OPINION
NETTESHEIM, Judge.This inverse condemnation case is before the court after trial to determine liability. The trial was bifurcated at defendant’s request. Plaintiff asserts that the subsurface drainage to his Arizona farmland, located in the Mohave Valley near the Colorado River and Topock Marsh, was taken by the United States without just compensation when high groundwater levels invaded the root zone of plaintiff’s crops, thereby rendering the land unusable for agricultural purposes. As the cause of this high groundwater, plaintiff ascribes causation to the system of dams and reservoirs on the Colorado River and to Topock Marsh, either independently of or in conjunction with each other. The United States operates both these public projects. Plaintiff seeks to recover for a temporary taking of his entire farm from 1983 to 1987 and a permanent taking of the western half of his property.
*87FACTS
1. Hydrologic features
This case requires a familiarity with the major hydrologic features of the Mohave Valley (the “valley” or “Mohave”). These features include the Colorado River (the “river”), the dams, reservoirs, and levees along it, and Topock Marsh (the “marsh”).
The Mohave is an “alluvial” valley1 located along the river on the border of California, Arizona, and Nevada. The valley is also a continuous “aquifer,” signifying that it is capable of subsurface storage and transmission of water. The Mohave is roughly 33 miles long and from two to five miles wide, with its northern end approximately ten miles south of Davis Dam and its southern end at Topock, Arizona.
The Colorado River begins its 1,400-mile path in the Rocky Mountains of Colorado and Wyoming, travels through portions of Utah and Arizona enroute to Mexico, and finally empties into the Gulf of California. As the second longest river system in the United States, it drains approximately 242,-000 square miles of territory, or one-twelfth the area of the contiguous United States.
In its natural state, the river often meandered, or changed its course, and would divide and split into various channels. Regarded as authoritative by the parties, Geological Survey Professional Paper 486-J by D.G. Metzger and O.G. Lerbtz entitled “Geohydrology of the Needles Area, Arizona, California, and Nevada,” published by the United States Department of the Interior in 1973 [hereinafter “486-J”], discloses that usually during the months of April through July, snowmelt from the mountains entered the river system. Such inflow caused the river annually to overflow its banks, thereby flooding large parts of the flood plain, some of which were distant from the river. 486-J further reports that normally the river migrated across the flood plain via abandoned channels. Although some water returned to the river, much of it became trapped and formed sloughs and oxbow lakes. Similarly, Reports on the Parker, Fort Mohave, and Cibola Irrigation Projects, Arizona, 1922: Hearings on HR. 11449 Before the Committee on Irrigation of Arid Lands, Appendix A, 67th Cong., 2d Sess. 144 (1922), states that “[practically the entire Mohave Valley is subject to overflow to a considerable depth.” The natural river channel near Needles, California, had the capacity to transport between 25,000 to 30,000 cubic feet per second (“c.f.s.”) of water.
As early as 1891, farmers attempted to divert river water, either by gravity or pumping, for irrigation purposes. These attempts failed due largely to the uncontrolled flow of the river. Subsequent attempts to irrigate the land from pumping wells also proved unsuccessful. As a result, irrigation agriculture was practically nonexistent throughout the 1940’s, with only a few hundred acres in the valley irrigated for farming purposes.
Prior to the construction of the first dam and reservoir on the river, high flows occurred seasonally and damaging flows were common. Annual spring floods rendered the valley largely useless for agricultural purposes. To contain these floods and put the river to effective use, the Government constructed a flood control system along the entire Colorado River.
The inceptive stage of this flood control system began in 1928 when Congress enacted the Boulder Canyon Project Act, ch. 42, 45 Stat. 1057 (1928) (codified as amended at 43 U.S.C. § 617 (1988)), which authorized construction of Hoover Dam. Section 6 of the Act provides, in pertinent part:
*88The dam and reservoir ... shall be used: First, for river regulation, improvement of navigation, and flood control; second, for irrigation and domestic use and satisfaction of present perfected rights ...; and third, for power____
In addition to Hoover Dam, the United States, through its Bureau of Reclamation (the “Bureau”), erected several other dams along the river. Starting in the north and tracing the river’s general downstream path southward, the northernmost dam of interest for this case is Glen Canyon Dam, located near Page, Arizona. Lake Powell filled behind Glen Canyon Dam between 1963 and 1980, has a storage capacity of 25,000,000 acre feet of water, and is located roughly 360 miles upstream from Hoover Dam. Seventy percent of the annual flow of the Lower Colorado River2 derives from snowmelt which gathers in Lake Powell.
Proceeding south, the next dam of importance is Hoover Dam. Completed in 1935, it is located near Boulder City, Nevada, and is approximately 108 “river miles”3 north of Topock, Arizona. Lake Mead filled behind Hoover Dam during the years 1935 to 1941 and has a storage capacity of 27,000,-000 acre feet of water. Releases from Hoover Dam, based largely on energy requirements of utilities companies, vary greatly from hour to hour. For example, as power demands increase during daylight and business hours, releases must increase proportionally to provide higher energy output.
The next significant dam to the south is Davis Dam, which is 67 river miles south of Hoover Dam, 41 river miles north of To-pock, Arizona, and about 10 miles north of the Mohave Valley. After completion in 1953, Lake Mohave formed behind Davis Dam. This dam and lake facilitate compliance with a treaty between the United States and Mexico that requires the United States to divert a predetermined amount of water to Mexico. While it is possible to utilize Davis Dam for flood control purposes, the capacity of Lake Mohave for water storage is comparatively small. Therefore, it serves primarily as a source of water for Mexico.
Further south Parker Dam is the next major point of interest for this case. With the dam’s completion in 1938, Lake Havasu began filling. This lake is 45 miles long, has a storage capacity of 648,000 acre feet of water, and has a surface elevation of 450 feet above mean sea level (“m.s.l.”). Parker Dam is 83 river miles south of Davis Dam, 42 river miles south of Topock, 150 miles south of Hoover Dam, and 12 miles Northeast of Parker, Arizona.
Just north of Lake Havasu, the Colorado River passes through an area known as Topock Gorge. Topock Marsh, a fish and wildlife preserve operated by the United States Department of the Interior Fish and Wildlife Service, has its southern end slightly north of Topock Gorge. Situated on the Arizona side of the Colorado River and roughly midway between Davis and Parker Dams, the marsh area consists of some 20,000 acres. The southern end of the marsh connects with the Colorado River, joining it at the entrance to Topock Gorge. An inlet channel also connects the river to the northern end of the marsh. The marsh’s present northern border is opposite Needles, California, which is located approximately ten river miles north of To-pock.
Acting both as a wintering site for migrating waterfowl and as a valued fishing spot, the marsh is a significant hydrologic feature in the Mohave Valley, but is not an integral link in the operation of the Colorado River as a flood control project. Its functions are largely recreational and environmental. Currently, the marsh is a part of the Havasu Lake National Wildlife Refuge.
*89The marsh did not exist in nature as it does today. It came into existence as an unanticipated consequence of the river control project. As explained in a letter of June 13, 1956, from one of the Bureau’s Regional Directors to his counterpart in the Fish and Wildlife Service, “Creation of the ... [marsh] was not planned nor foreseen____ [H]ad the conceivers and designers of the two dams visualized the extent to which the Topock ... [Marsh] would develop, they would have incorporated in the plans an effective control to avoid creation of the ... [marsh]____”
At the valley’s southern end, its width ranges from two to five miles. Throughout history the river meandered considerably within this area. Upon completion of Parker Dam in 1938, Lake Havasu started to fill above the dam, and, accordingly, the river’s flow speed decreased as river water approached the lake. Sediment deposition on the river floor resulted.4 Due to this deposition, the elevation of the river floor generally increased at Topock Gorge and the upper end of Lake Havasu. Since the river bottom elevation increased, so, too, did the river surface elevation.
By 1944 an emergency situation existed. With the closure of Parker Dam and the filling of Lake Havasu in 1938, continued river aggradation in the lower portion of the valley caused “stages,” or river surface elevations, and consequently groundwater levels, to increase in that area.5 This aggradation caused the obliteration of any definable river channel, thereby leading to the river’s spreading out over practically the full width of the lower valley. With the corresponding reduction of river flow speeds, silt dropped out of the river, causing numerous sandbars to form in the river channel. This process turned the river channel into virtually a densely overgrown lake or swamp. The resulting buildup of backwater threatened both the city of Needles and the Atchison, Topeka, and Sante Fe Railroad with gradual surface water and groundwater inundation.
The Bureau’s “Report on The Needles Situation” by C.P. Vetter, issued in December 1944 [hereinafter the “Vetter Report”], described the aggradation, in the following manner:
The Needles situation has been in the making for many years. Only sporadic water surface observations were made prior to the year 1931, but such records as are available indicate that, at least since 1902, there has been an apparently unbroken increase in the water level at Needles for a given discharge of the river____
Between 1902 and 1935, that is, up to the time when Boulder Dam was closed, the water surface for a discharge of 10,000 cubic feet per second increased approximately 11 feet, or an average of 0.35 foot per year. During the following few years, until the beginning of 1941, while Lake Mead was being filled and discharges were being held to a minimum, the water surface at Needles held relatively constant, fluctuating around elevation 469.00____ Beginning in 1941, the water surface at Needles again began to rise at a steady rate, a rise which has continued to date. While there have been minor fluctuations in the rate, the average has been close to 1.5 feet a year *90or 4.3 times the average for the years 1902 to 1935.
Both parties relied heavily on this report.
In an effort to alleviate this situation, the Bureau undertook a program of dredging operations, levee construction, and channel alignment from Big Bend6 to Topock.7 The program lasted from 1949 to 1959; routine upkeep continues perpetually.8 This effort lowered the river surface elevation by roughly five feet and stabilized the river channel in the Mohave Valley. However, some 8,700 acres of open water remained in the area now known as Topock Marsh.
Marsh water elevations originally varied seasonally according to river surface levels, with summer highs reaching 458 feet above m.s.l. By 1951 Topock Marsh had become such an important wildlife refuge that the Government decided to preserve the marsh in order to sustain the indigenous animal life. To accomplish this goal, water levels must be kept roughly at 455 feet above m.s.l.
Due to the period of aggradation leading up to the marsh’s creation, protective measures to maintain the marsh’s artificially high level were unnecessary initially. However, by 1964 a prolonged period of degradation in the Topock area, due in part to the earlier dredging and channel alignment, made it obvious that, unless protective action was taken, water outflow from the marsh might render it completely dry, thus endangering the wildlife living in the refuge area. Protective measures were undertaken, including the construction of a retention dike in 1965, designed to prevent water from flowing southward out of the marsh into the river; an inlet channel in 1966, which transported water directly from the river into the northern end of the marsh; and inlet and outlet structures, which allowed controlled volumes of water to exit or enter the marsh through the dike. These combined features controlled to some extent water level fluctuations in the marsh. To a large degree, however, due to the hydraulic connections of the inlet channel and the inlet and outlet structures, water levels in Topock Marsh are still directly proportional to the surface elevation of the river. Therefore, water levels in the marsh are likely to be higher in the summer and lower in the winter.
[[Image here]]
*912. Flood and water control
Before completion of Hoover Dam in 1935 and the subsequent filling of Lake Mead, no reservoirs or control structures existed to regulate the chameleon-like water supply. River flows varied drastically from year to year and from season to season.9 For example, average monthly flows below Hoover Dam fluctuated from a low of roughly 4,000 c.f.s. to a high of approximately 130,000 c.f.s. 486-J discloses that on June 22, 1921, the maximum discharge reached 175,000 c.f.s., and that the differences between yearly maximum and minimum discharges averaged about 100,000 c.f.s. Normally, the lowest flows occurred in January and the highest took place in July. When Lake Mead started to fill in 1935, these extremely high flows stabilized moderately, as Lake Mead simply stored most excess water. Average monthly releases during this time ranged from a low of approximately 5,000 c.f.s. to a high of about 22,000 c.f.s. Lake Mead filled in 1941, and, as there was no extra reservoir capacity in the system at that time, discharges became increasingly variable, ranging from a low of roughly 8,000 c.f.s. to a high of around 33,000 c.f.s. This regime continued until the completion of Glen Canyon Dam and the creation of its new reservoir in 1963.
When Lake Powell began filling behind Glen Canyon Dam, excess water supplies were stored in the lake, thus significantly reducing the likelihood of downstream flooding. During this interval discharges from Hoover Dam attained a new level of consistency. Average monthly flows below Hoover Dam varied from a low of about 4,000 c.f.s. to a high of only 18,000 c.f.s. The large flows of the pre-Hoover Dam area no longer resulted. For this period releases were a function of irrigation requirements and energy needs. When Lake Powell filled in 1980, the system of reservoirs attained its water storage capacity limit, thereby creating a potential for high flows once again.
DX 504 charts the natural flow into the system from 1906-1990:10
[[Image here]]
*92DX 505 depicts the flows below Hoover Dam from 1906 through March 1990:
[[Image here]]
The Bureau of Reclamation is the government agency that operates the water control structures on the Colorado River. After consulting water requirement schedules provided by all major water and power users, including Mexico, the Bureau determines the schedule of releases from Hoover Dam. See Gasser v. United States, 14 Cl.Ct. 476, 492 (1988).
Since virtually all of the water flow in the Lower Colorado River passes through Glen Canyon Dam, predictions of water supply to Lake Powell are of critical importance in determining the timing and extent of releases from both Glen Canyon and Hoover Dams. The forecast period, which runs from January to July of each year, is the time during which the Bureau utilizes the predictions to plan the amount of vacant space needed in the reservoir system.11 If a forecast predicts a large water supply from the annual snowmelt, early, preventive flood control releases called “space building” can be made which create room for the expected run-off. An average water supply for a year is 15,000,000 acre feet of water. The highest yearly water supply approximates 23,000,000 acre feet, and the lowest is close to 5,000,000 acre feet.
The 1983-1987 period represents the highest five consecutive years of inflow on record for the river. See DX 504 supra p. 91. From April 1983 to roughly August 1987, the Bureau released unusually large volumes of water from Hoover and Davis Dams into the Lower Colorado River. The parties jointly stipulated that these sizable discharges resulted in high river flows through the Mohave Valley, which, in turn, caused high groundwater conditions on plaintiffs property.
In 1983 the early water supply predictions provided to the Bureau called for a basically normal year. The January forecast called for an above average inflow, while the February and March forecasts predicted a below normal inflow. These predictions were in error. Due to an un*93usually cold spring, heavy late snows, and a sudden warming trend, an accurate forecast of the water supply was not possible until much later than usual. Additionally, heavy rains further confused and delayed the prediction by contributing to the influx of water into the river system. The actual water supply for 1983 was 23,000,000 to 24,000,000 acre feet, or over 50 percent above the normal water supply.
As a result of this late prediction and large water inflow, adequate space building did not take place; hence, insufficient storage capacity existed in the reservoirs. Accordingly, the Bureau released large amounts of water through Hoover Dam. It attempted to keep releases at a 40,000 c.f.s. level, but flows often reached 44,000 c.f.s. Peak releases sometimes reached as high as 65,000 c.f.s. Additionally, at one point water poured over the spillway gates, adding another 28,000 c.f.s. of water into the river flow. Having reached Davis Dam, these large releases exited quickly, as effective storage was not possible in the smaller Lake Mohave. Without the dams a daily peak flow would have reached roughly 128,000 c.f.s., the value corresponding to a volume of inflow into Lake Powell on one day in June 1983.
Above average water supplies continued into 1984, when the supply for the year was 24,000,000 acre feet. The supply for 1984 was slightly higher, then, than for 1983. However, with the benefit of an earlier and more accurate forecast, the Bureau was able to plan accordingly and release sizable amounts of water from Hoover Dam before the snowmelt entered the river system, thus averting the need to repeat the large discharges of 1983. For the year the maximum average monthly release from Hoover Dam was 35,000 c.f.s.
In 1985 with a water supply of 20,000,000 acre feet, the average monthly release through Hoover Dam was 26,000 c.f.s. In 1986 a supply of 21,000,000 acre feet caused the average monthly release to be 30,000 c.f.s. The 1987 average monthly discharge of 20,000 c.f.s. resulted from a water supply of 16,000,000 acre feet. After 1987 the water supply lowered significantly. In 1988 and 1989, the supply levels were roughly 11,000,000 and 9,000,000 acre feet, respectively. The projected 1990 water supply is about 8,000,000 acre feet.
The high flow years of 1983 to 1987 also impacted the size and elevation of Topock Marsh. Between January 1981 and April 1983, the average monthly water surface elevation of the marsh varied between highs of 454 feet and a low of 448.5 feet. The marsh reached the upper level on four separate occasions. The high river flows beginning in 1983 dramatically increased the marsh elevation.12 The average monthly elevation peaked in July 1983 at 461.6 feet. In 1984 through 1989, the elevation ranged from low to high, as follows: in 1984, from 456.8 to 459.6 feet; in 1985, from 454.3 to 457.6 feet; in 1986, from 454.3 to 458.8 feet; in 1987, from 451.6 to 456.0 feet; in 1988, from 450.1 to 454.5 feet; and in 1989, from 449.6 to 450.1 feet. Through May of 1990, the average monthly water surface elevation for that year ranged from 449.4 feet to 453.8 feet. The optimal elevation for the marsh is 455 feet above m.s.l.
An increase in the surface elevation of the marsh water causes a corresponding increase in the surface area covered by Topock Marsh. Thus, in 1983 when the marsh was at its highest level, significant amounts of farmland to the north of the marsh were inundated with marsh overflow. At the peak of the marsh flooding, its northern border was roughly 2.5 miles from the eastern border of plaintiff’s property. When maintained at its usual elevation of 455-456 feet, the marsh’s northern border is approximately five miles removed from plaintiff's farmland.
3. Groundwater levels
The primary sources of groundwater in the Mohave Valley are the Colorado River, unused .irrigation water, run-off from pre*94cipitation, and Topock Marsh. The river is the principal source of groundwater “recharge.” 13 Generally, the groundwater moves from the river into the flood plain and then discharges by transpiration or evaporation. Due to a hydraulic connection between the river and the groundwater system, river stages directly influence the rate of recharge to and discharge from the groundwater system, as well as impact the amount of groundwater stored in the system. The river surface elevation and groundwater levels are identical at the river’s banks, but as the distance from the river increases, so, too, does the depth to water.
The current groundwater regime differs considerably from its natural state due largely to the presence of dams, reservoirs, and levees on the Colorado River. 486-J discloses that, with the completion of Hoover Dam in 1935 and the ending of much of the annual flooding of the river, the groundwater recharge resulting therefrom was either reduced or eliminated to a large degree. Parker Dam also impacted the groundwater system by causing aggradation of the river channel in the lower Mohave Valley. As of 1973 486-J reports that this sediment aggradation caused the average river stages at Needles and Topock to increase eight and 27 feet, respectively, over the stages found in 1902-1903.
Generally, groundwater levels in the valley vary annually within a range of two feet, with exceptions for areas near working wells, the river, and irrigated land. Due to the river’s somewhat predictable flow fluctuations, the valley’s water table rises normally in the summer and drops in the winter.
Evidence of historic groundwater levels in and around plaintiff’s property, or in the general Mohave Valley area, is sketchy. Apparently, no systematic study of groundwater levels in the valley was undertaken prior to this litigation. Some evidence can be gleaned from the existence or absence of a type of vegetation known as phreatophytes, which include, among other plants and trees, salt cedar, arrowweed, mesquite, willow, and tules. These types of vegetation thrive on large amounts of water and, hence, flourish where groundwater is prevalent. Therefore, it can be inferred that if phreatophyte growth is prevalent at a given time, the groundwater level is likely to be high for the same time span. Nonetheless, phreatophytes are hardy and have been known to survive when water levels recede by as much as 30 to 50 feet.
In 1922 the groundwater level in the valley appeared close to the surface, in that phreatophytes including mesquite and willow were then thriving in the valley. On one date in 1935, the depth to groundwater on land that later became plaintiff’s farm was four feet. By 1944, as a result of the condition illustrated in the Vetter Report, the subsequent rise in groundwater levels caused portions of the area to become waterlogged. However, the dredging and channelization project reduced the river surface elevation, which corresponded to a decrease in the depth to water. From 1961 to 1962, the groundwater elevations simultaneously at some locations near Section 3 equalled six feet, at other points 12 feet, and at further portions an elevation in between. 486-J discloses that in 1973 the general groundwater level for the valley was roughly nine to 12 feet, with a minimum depth to water of six feet and a maximum of 20 feet.
Close to plaintiff’s property the depth to water was 12 to 13 feet in 1979. The groundwater level in wells near the property in 1980 was ten feet below the ground’s surface. At some point in late 1983 the groundwater was only two feet below the surface just off plaintiff’s land. In April 1984 the groundwater elevation was nine feet. The most current data reflect that the depth to groundwater at the center of plaintiff’s farm is six feet.
4. Plaintiffs property
Plaintiff owns property in the Mohave Valley, described as the southern half of *95Section 3, Township 17 North, Range 22 West of the Gila and Salt River Base and Meridian, Mohave County. Consisting of 320 acres and located approximately one mile east of the Colorado River, the property is 14 river miles north of Topock, Arizona, and five miles north-northwest of the present northern border of Topock Marsh. Situated near river mile 249, plaintiff's property is one mile east of the river, as the parties stipulated.
Plaintiff purchased the property in 1978 in a foreclosure sale with the intent of growing crops as feed for cattle located on a ranch owned by the Laughlin Land & Cattle Co. Previously unused, the property required clearing of its very heavy growth of phreatophytes, such as mesquite and salt cedar, as well as brush. At the time of purchase, there were no manmade improvements on the land. Consistent with his intention to use the land for agriculture, plaintiff cleared the land and surveyed, leveled, and slip-plowed it to a depth of about ten feet. Plaintiff then divided the property into eight separate fields, with two rows of four fields each, running from west to east. The field numbers increase numericauy from west to east, with field one in the southwest corner, field four in the southeast corner, field five in the northwest corner, and field eight in the northeast corner. An irrigation ditch constructed by plaintiff through the middle of the property abuts the east and west borders of the farm, and a gravel roadway traverses the middle of the property in a north to south direction. Plaintiff devoted the entire 320 acres to farming purposes.
According to Bemie M. Osbom, who managed the subject property for plaintiff, prior to the initial plantings, plaintiff’s farmer, Lee Banning, treated the ground with fertilizer, but added no minerals to the soil.14 Additionally, plaintiff drilled two wells on his property. Well number one, completed on October 13, 1980, reached a depth of 234 feet. Well number two, finished on November 13, 1980, was 222 feet deep. Field number five is the location for both wells, with well one in the southwest comer of the field, and well two in its southeast corner. These are the only wells on plaintiff’s land.
Planting of the first crop on plaintiff’s land occurred in the spring of 1981 and consisted of milo and barley. The crop was a success, with 90 to 95 percent of the property yielding produce. In 1982 after this initial harvest, Mr. Banning sowed alfalfa, which was also a success. The “root zone” for alfalfa is six feet, signifying that the roots reach a depth of six feet. Generally, this is also the root zone for the other crops on plaintiff’s land. Since the land had not been farmed previously, it was Mr. Osborn’s opinion that Mr. Banning regarded the first two crops as experimental. These early crops hopefully would provide an indication of what the land could produce consistently. As a perennial crop, alfalfa does not require replanting.
Late in 1983 the alfalfa began to perish, especially on the western side of plaintiff’s property. This failure occurred soon after the flood and high discharges of 1983, which began in April of that year. The flood caused high groundwater levels below plaintiff’s property. During this period of high flows, a process called “capillary action” transported water and salts through the plants’ root zone.15 Salt travelled to the surface and blanketed 95 percent of the land’s surface. This high salt deposition was not unique to plaintiff’s property, although its impact did vary from farm to farm. Eventually, all the crops on plaintiff’s property died.
Mr. Osborn reported that soon after salt first appeared, Mr. Banning, having consulted various valley farmers, attempted to correct the high salt levels in the soil by spreading' sulfur, gypsum, and “gin *96trash”16 on the ground. All efforts proved fruitless, and up until the beginning of 1984, any seeds planted grew only sporadically, if at all.
During this period of high groundwater, the land’s surface softened, especially on its western half, so that vehicles would sink if they attempted to move across the property. Also, the soil took on a white color due to the deposition of salts on the surface. Additionally, a pit, dug just off field three to the south of the property in late 1983 had a groundwater level that was only two feet below the surface of the property.
Mr. Osborn testified that while this same high groundwater period was ongoing, Mr. Banning made numerous attempts to cultivate crops on the farm, particularly on its eastern half where the groundwater levels appeared to be lower. Eventually, according to Mr. Osborn, since Mr. Banning had been losing money, Mr. Banning calculated that if he were to remain any longer and attempt to farm the land, he would become insolvent. Mr. Banning stopped farming plaintiff’s property in 1984 and Victor Wakimoto replaced him. Mr. Osborn explained that plaintiff paid Mr. Wakimoto for costs incurred in his attempts to rejuvenate the land, thus enabling the farmer to avert any risk associated with low yielding crops.
In 1984 the eastern side of plaintiff’s property did not yield enough crops to pay for the farming. While some plants did grow and plaintiff reaped a harvest, the eastern side produced only one half of its 1981 and 1982 output. Slowly, the groundwater receded and the conditions on the eastern side improved so that 1990 was the first break-even year for the eastern portion since the high flows of 1983 through 1987.
The western side of the property did not experience the same success, however. Each successive spring, attempts were made to induce crop growth and every attempt met with failure. The westernmost portion of the land, the closest to the river, was in the worst condition. Mr. Osborn stated that until recently a 15-acre section of field one went “back to nature” because, for a number of years, equipment could not be driven through it without sinking. He testified that the water table is currently at such a high level that the salt cannot be “leached”17 out of the soil. As of trial in August 1990, the western portion of the land remains largely unproductive.
Defense witness Charles Sherrill farms the northern half of Section 3, the southern half of which plaintiff owns and farms. In addition to several other farms in the area, Mr. Sherrill also farms Section 11, located catty-corner to the southeast corner of Section 3, and Section 27, located further north in the valley. In total, the witness farms nearly 5,000 acres in the Mohave Valley.
Mr. Sherrill first farmed the northern half of Section 3 in 1980. In 1983 his crops on this parcel consisted of cotton, alfalfa, and small grains; his Section 11 crops included alfalfa, cotton, and wheat. The presence of water in the crops’ root zone and salts in the root zone and on the surface of the land forced him to engage, in his words, in “aggressive” and “specialized” farming. Generally, these procedures required that the farmer, first, give daily attention to the non-waterlogged portion of the soil to ensure that it was not over- or under-watered and, second, push the salt out of the seed line of newly planted crops and water only every other row until the crops strengthened adequately to survive in spite of the salt.
Based in large part on his aggressive and specialized farming practices, Mr. Sherrill experienced a successful harvest in 1983. When defense counsel queried if he experienced groundwater-related problems in farming Section 11 that year, the witness replied: “Well, we may have had to be a little more careful in the beginning when we were planting our crops, but ... I don’t *97remember any yield reduction____” In response to a subsequent question concerning whether he encountered any farming obstacles in general, the witness responded: “No. Like I said, ... [the land] just had to be aggressively farmed. It had to be watched a little more closely____ You had to watch your ground a little bit more and go out there with a shovel and dig down and see what your ... roots were doing.”
Additionally, even though the depth to water on Section 27 reached one and one half to two feet, Mr. Sherrill experienced only a 15 to 20 percent crop yield reduction for that section. The witness possessed no knowledge of the exact farming methods used by plaintiffs farmers, but did deem Mr. Wakimoto knowledgeable in Mohave Valley farming practices and an effective, productive farmer.
5. Expert testimony
At trial plaintiff called William Gerald Matlock, a doctorate who eschews the title, as an expert witness qualified in the areas of hydrogeology, irrigation, drainage, groundwater resources, and agricultural productivity. Mr. Matlock’s testimony spanned numerous topics, including, but not limited to, river water levels, river discharge as a source of groundwater in the valley, the cause of groundwater conditions on plaintiff’s property, the cause of interference with plaintiff’s subsurface drainage, and the effect of high groundwater on plaintiff’s farmland.
His testimony consisted of, in part, the following facts and opinions. First, Mr. Matlock stated that Topock Marsh, which did not exist in nature, permanently excludes groundwater outflow from the southern end of the valley. He attributes this effect of the marsh to its artificially high water surface elevation and the sediment associated with the marsh. These two factors reduce the “gradient”18 of the aquifer, making it closer to level, thereby inhibiting adequate subsurface drainage under plaintiff’s farm. Generally, then, water that eventually impacts plaintiff’s land leaves the river, passes through sediments, enters the aquifer, and flows down the gradient until it passes underneath plaintiff’s property and into the cone of depression. See infra note 20 and accompanying text. Mr. Matlock added that when the marsh’s size expanded as a result of the 1983 flood, the subsequent rise in surface elevation created more opportunity for groundwater recharge northward into the valley. He stated that this recharge from the marsh also occurs when its elevation is only 456 feet above m.s.l. Second, Mr. Matlock stated that, inasmuch as plaintiff’s property is found only one mile from the river and since the aquifer has a very high “transmissivity” value,19 any change in river stages would quickly impact the groundwater level on plaintiff’s property.
Mr. Matlock testified that the Colorado River and Topock Marsh are the two principal sources of groundwater recharge in the valley. He maintained that regularization of the river channel created a new discharge regime, which translates into different discharge levels. After noting that the river elevation is higher now than in 1902 and 1903, Mr. Matlock concluded that the rate of recharge from the river into the aquifer has risen generally over the past 80 years. Also, since the marsh’s surface level is higher today than in nature, the gradient in the lower valley has been altered to such a degree that groundwater actually flows northward from the marsh into the valley.
*98He further testified that, without the marsh’s flooding of 1983 and 1984, the groundwater invasion under plaintiff’s farm would have been reduced significantly. Based on readings from two wells— one on plaintiff’s land and one off it — Mr. Matlock hypothesized that if the control structures of the marsh did not then exist, the rise in groundwater under plaintiff’s property would have been five feet less than what actually resulted, thereby causing only a small western portion of the land to experience water damage. The witness went on to assert that if Topock Marsh did not exist at all, the groundwater invasion under plaintiff’s property likely never would have taken place. In part, the different gradient resulting from the 27-foot increase in depth to groundwater if the marsh were absent would have made northward inflow into the valley unlikely.
Additionally, Mr. Matlock stated that the high groundwater levels remained under the center of the property for almost four years. He obtained this information from readings of a well that is just off plaintiff’s property, but near the mid-point of his land. Also, the water remained at high levels in the western portion for about four years and three months and is currently on the border of the root zone, which is six feet. This high groundwater, through capillary action, brought salts up through the root zone and onto the land’s surface. Since common agricultural crops cannot grow in a highly saline environment, the presence of salt at this level killed or severely damaged the existing plants. Attempts to leach out the salts by flooding the land would be ineffective because there was no place for the water to drain.
Next, Mr. Matlock testified that another flow of 15,000 c.f.s. would place the groundwater levels under plaintiff’s property right at the root zone of his crops. A flow of 20,000 c.f.s. or above, in fact, would invade the root zone. In the witness’ view, such flows are very likely to recur, since there have been only eleven years out of the last 84 wherein the average annual flow did not reach 15,000 c.f.s. After considering that the Bureau attempts to maintain discharges at 15,000 c.f.s. in order to meet the demands of the Lower Colorado River, and after noting that the reservoirs are now full, Mr. Matlock asserted that a very real possibility exists that flows in excess of 15,000 c.f.s. will recur frequently, if not on a regular basis. He opined that the river system now more closely approximates its natural state than at any time while the reservoirs were filling.
Defendant’s expert in the field of hydrology was Dr. Paul Sebenik. The thrust of his testimony concerned the level and cause of groundwater conditions under plaintiff’s land.
Dr. Sebenik stated that although both the river and the marsh are sources of groundwater recharge to the valley, the river has more impact on plaintiff’s property because the river is a linear recharge system, while the marsh is circular. Linear systems recharge the groundwater system more quickly than circular. Also, the river is only one mile from the property, which is much closer than the marsh.
This witness went on to testify that the marsh has no practical impact on the groundwater level at plaintiff’s property. Theoretically, the marsh has a minuscule 0.08 foot impact on the groundwater at plaintiff’s land. However, any impact the marsh might have is overwhelmed by the river, which might change elevation by as much as five to seven feet in one day. This negligible impact is due largely to a “cone of depression” or, unartfully, a “sink hole,”20 which can be found between plaintiff’s property and the marsh. The bottom of the cone represents the lowest groundwater elevation in this area. Since gravity causes water to seek the lowest level, the cone collects all groundwater flowing from all directions, thereby preventing any of the marsh’s groundwater outflow from reaching or impacting plaintiff’s farmland.
*99Dr. Sebenik additionally stated that the marsh is not impeding groundwater outflow from the valley. Outflow can be blocked, in the witness’ words, only by plugging up the valley with a dam or a concrete wall placed deep into the ground. If not, the water will flow out somewhere, even if it must travel around the marsh. For reasons discussed in the section of this opinion dealing with causation, Dr. Sebenik’s testimony concerning the cone of depression has been discounted.
Plaintiff filed suit in this court on September 8, 1988. He claims the right to recover for a temporary taking of all of his property from 1983 through 1987 and asserts that loss of use is the correct measure of compensation. Plaintiff additionally claims the right to recover for a permanent taking of the western portion of his property. Diminution in value, due to the permanent presence of salt, is the alleged correct measure of damages for this taking.
Defendant contends that, first, the statute of limitations precludes jurisdiction; second, the high groundwater levels are not the natural and probable consequence of the public projects; third, the soil’s high salinity level renders the property unproductive; fourth, even if a causal relationship exists, the high river flows that led to the high groundwater levels are neither permanent nor inevitably recurring; and, finally, the doctrine of relative benefits precludes recovery.
The court has found and concluded that plaintiff is not entitled to recover because he failed to establish that either the Bureau’s actions on the Colorado River or Topock Marsh, or both, caused the damage to his property; because the damage was unintended and consequential; and, even assuming that the Bureau caused the damage, because plaintiff’s property sustained more benefit from the projects than damage.
DISCUSSION
The particular facts of each suit determine the outcome of takings cases. Herriman v. United States, 8 Cl.Ct. 411, 417 (1985) (citing Berenholz v. United States, 1 Cl.Ct. 620, 626 (1982), aff'd mem., 723 F.2d 68 (Fed.Cir.1983)). When the government operates water or flood control projects, a taking may result even if there is no direct flooding. Interference with the land’s natural drainage is a sufficient predicate to find a taking, and a taking may result from “percolation or rising groundwater.” Barnes v. United States, 210 Ct.Cl. 467, 475, 538 F.2d 865, 870 (1976) (citing United States v. Kansas City Life Ins. Co., 339 U.S. 799, 70 S.Ct. 885, 94 L.Ed. 1277 (1950)).
1. Statute of limitations
Defendant contends that the six-year statute of limitations, 28 U.S.C. § 2501 (1988),21 bars recovery for any claim based on blocked subsurface drainage under plaintiff’s property. Specifically, defendant argues that with the construction of the marsh level control works in 1965, the groundwater levels in the valley stabilized not more than ten years thereafter, or by 1975. Since plaintiff instituted this suit more than six years after 1975, defendant asserts that the statute of limitations deprives the court of jurisdiction.
The statute of limitations is jurisdictional. Parker v. United States, 2 Cl.Ct. 399, 402 (1983) (citing cases). A court may not waive the statute. Id. at 402. As recently stated by the Federal Circuit: “A claim first accrues and the six year statute of limitations begins to run (28 U.S.C. § 2501), ‘when all the events have occurred which fix the liability of the Government and entitle the claimant to institute an action.’ ” Hart v. United States, 910 F.2d 815, 817 (Fed.Cir.1990) (quoting Kinsey v. United States, 852 F.2d 556, 557 (Fed.Cir.1988) (citation omitted)); Hopland Band of Pomo Indians v. United States, 855 F.2d 1573, 1577 (Fed.Cir.1988).
*100In the context of determining whether permanent surface flooding of property constitutes a taking, the Supreme Court has held “that when a taking is caused by a continuous process, it is not complete, for purposes of determining when the claim arose, ‘until the situation becomes stabilized.’ ” Cooper v. United States, 827 F.2d 762, 764 (Fed.Cir.1987) (quoting United States v. Dickinson, 331 U.S. 745, 749, 67 S.Ct. 1382, 1385, 91 L.Ed. 1789 (1947)). “[W]hen the Government chooses not to condemn land but to bring about a taking by a continuous process of physical events, the owner is not required to resort to either piecemeal or to premature litigation to ascertain the just compensation for what is really ‘taken.’ ” Dickinson, 331 U.S. at 749, 67 S.Ct. at 1385. In other words, “[a]n owner of land flooded by the Government ... [may properly] postpone bringing a suit against the Government for the flooding until the consequences of inundation have so manifested themselves that a final account may be struck.” Id.22 The moment when a taking becomes sufficiently definite to sanction a claim for compensation varies with each case. Cooper, 827 F.2d at 764. This is a practical finding and not a rigid legal rule. Dickinson, 331 U.S. at 749, 67 S.Ct. at 1385.
In Dickinson, where the landowner’s property sustained permanent flooding by a dam’s impoundment of river water, the Court did not state a precise time at which the six-year statute started to run. However, in holding the claim not time barred, the Supreme Court did reject the Government’s position that the statute began to run either when the dam first impounded water or when the dam was fully capable of operation. 331 U.S. at 747-49, 67 S.Ct. at 1384-85. Similarly, in Cooper the issue was when the standing flood water caused sufficiently stable destruction of the owner’s timber so that the owner could determine the amount of timber taken. Although water invaded the property’s surface and the trees began to perish in 1979, the Federal Circuit ruled that the claim did not accrue until the extent of the damage was first discemable in 1984, the year in which the owner filed suit. 827 F.2d at 764.
Plaintiff filed suit on September 8, 1988, and, therefore, his claim must have accrued no earlier than September 9, 1982, for jurisdiction to lie in the Claims Court. With regard to plaintiff’s claims for temporary and permanent flowage easements, the inquiry is, then, when the last event occurred which fixed liability on the Government so as to entitle plaintiff to institute an action. Kinsey, 852 F.2d at 557-58.
Between 1963 and 1980, Lake Powell was filling behind Glen Canyon Dam. During this period the average river flows were at an extremely low level. Since river stage levels directly affect groundwater elevation, groundwater levels in the valley receded at least somewhat during this 17-year period. In 1983 through 1987, the Bureau released uncommonly large volumes of water from Hoover and Davis Dams, which resulted in high river flows, thereby causing groundwater levels to increase at plaintiff’s property. But, more importantly, plaintiff did not become aware of the impact of the groundwater invasion until 1983 when his crops first began to perish.
In this case the low groundwater elevation consequent to the filling of Lake Powell, completed in 1980, is inconsistent with defendant’s argument that the groundwater regime under plaintiff’s property stabilized in the 1970’s. In a statute of limitations context, even assuming that defendant is correct that the marsh’s effect on groundwater levels stabilized by 1975, any influence by the marsh cannot be imputed retroactively. While the record reflects no precise moment in time at which the stat*101ute of limitations began to run, it is held that, for both claims, conditions did not stabilize on or under plaintiffs property until after September 9, 1982. See Dickinson, 331 U.S. at 749, 67 S.Ct. at 1385; Hart, 910 F.2d at 817. Since plaintiff filed his claims within the six-year statute of limitations, the Claims Court has jurisdiction to consider them.
2. Natural and probable consequence
For a compensable taking to exist, the flooding under the property must be the natural and probable consequence of government action. Bartz v. United States, 224 Ct.Cl. 583, 593, 633 F.2d 571, 577 (1980) (“Here the damage was not the natural consequence of government action.”), cert. denied, 450 U.S. 967, 101 S.Ct. 1484, 67 L.Ed.2d 616 (1981). The Court of Claims stated further: “The United States is not liable for flood damages unless directly attributable to governmental action. Indirect or consequential damages are not compensable.” Id.; Hartwig v. United States, 202 Ct.Cl. 801, 809, 485 F.2d 615, 619 (1973). Other cases utilize the same requirement of natural and probable causation, but label it the “direct and proximate” result of government action. Loesch v. United States, 227 Ct.Cl. 34, 43, 645 F.2d 905, 913 (citing cases), cert. denied, 454 U.S. 1099, 102 S.Ct. 672, 70 L.Ed.2d 640 (1981); Baskett v. United States, 8 Cl.Ct. 201, 209 (1985) (citing cases), aff'd, 790 F.2d 93 (Fed.Cir.) (Table), cert. denied, 478 U.S. 1006, 106 S.Ct. 3300, 92 L.Ed.2d 714 (1986).
Causation is a factual, not legal, determination. Loesch, 227 Ct.Cl. at 43, 645 F.2d at 913. In this case a preliminary matter is whether the river and the marsh merit treatment as separate or combined hydro-logic features for purposes of determining causation. Resolution of this issue affects the analysis of causation only insofar as it creates either a separate or unified examination of these features. Acknowledging that the river and marsh are interactive, the court has determined to treat the river and the marsh separately in terms of causation. The parties presented their respective cases in this manner, and the analysis would not profit from the added complexity of melding the two.
Causation becomes a twofold inquiry. The first issue is whether the Bureau's operation of the Colorado River as a flood control project caused the high groundwater table at plaintiff's property. The second issue is whether the existence and/or operation of Topock Marsh produced the high water elevation at plaintiff's farm. Regarding both issues, the court finds causation lacking.
a. The Colorado River
Indisputably the Bureau of Reclamation controls the timing and extent of discharges from Glen Canyon, Hoover, Davis, and Parker Dams. The Bureau bases these discharges on numerous factors, including, but not limited to, downstream irrigation, flood control, water conservation, improvement of navigation, electrical energy requirements, and water needs of Mexico. Thus, the Colorado River is not solely a flood control project.
Plaintiff contends that the releases in excess of 20,000 c.f.s. will invade and damage the root zone of his crops. The fundamental point to note, though, is that although the Bureau does control discharges, it does not control the amount of precipitation, usually in the form of snowmelt, that flows into Lake Powell behind Glen Canyon Dam. Nature controls this inflow. Since no excess storage capacity currently exists in the system’s reservoirs, any precipitation in excess of existing storage space must flow downriver.23 Through the water it supplies, nature largely dictates the releases the Bureau must make. It is the Bureau’s responsibility, as safely as *102possible, to regulate discharges of the water naturally added to the system.
A similar issue arose in Bartz. That case concerned riparian farmland whose owners attempted to recover for inverse condemnation due to construction and operation of a dam. The main complaint was that the presence of the dam and reservoir caused backed-up river water to linger on plaintiffs’ properties for longer periods than in nature, thereby detrimentally interfering with farming procedures and production. In denying relief, the Court of Claims stated: “Excessive precipitation was the root cause of the flooding experienced by plaintiffs’ in the wet years of which they complain. The Government’s manipulation of releases from the dam played only a secondary role.” 224 Ct.Cl. at 593, 633 F.2d at 577; see Hendricks v. United States, 14 Cl.Ct. 143, 156 (1987) (“[T]he most significant factor causing the flooding was the high natural precipitation over the past six years with some undeterminable effect by the reservoir____”); cf. Warrior & Gulf Navigation Co. v. United States, 864 F.2d 1550, 1553-54 (11th Cir.) (“unprecedented rain,” not the Government’s operation of a lock and dam system, proximately caused a series of river accidents), cert. denied, — U.S. -, 110 S.Ct. 236, 107 L.Ed.2d 187 (1989). The court in Bartz stated further: “Faced with the need to reduce the high levels of the Reservoir because of actual or impending heavy rainfalls in the area the Corps had no alternative to increasing the rate of discharge from the dam, despite temporary consequences to downstream farmers.” 224 Ct.Cl. at 592, 633 F.2d at 577.
Plaintiff relies heavily on Gasser, a surface flooding case which also addressed the issue of causation by the Colorado River. In that case due to the Government’s purposeful impoundment of water in Lake Powell, only low volume flows reached Mexico, resulting in aggradation of the river floor in that area. Whenever the large flood control releases of 1983 through 1985 did reach Mexico, the water overflowed the aggraded channel and flooded the two plaintiffs’ land. The court concluded “that the construction of the dams, the filling of the lakes, and the flood control releases ... were under the exclusive control of the United States____” 14 Cl.Ct. at 502. This statement is correct in that only the Government was in control of the completion of these construction projects and in charge of the releases once water flowed into the system. However, Gasser does not address the issue of control in the context of inflow into the Colorado River. Accordingly, Gasser does not stand for the proposition that the Government controls, or could control, the amount of water that flows into the system each year.
The case at bar does not involve a situation wherein the Bureau made random extended releases of 20,000 to 65,000 c.f.s. for reasons unconnected to insufficient storage space in the reservoirs. See Bartz, 224 Ct.Cl. at 592, 633 F.2d at 577. If this were the case, a basis might exist for finding that the Bureau was in charge of the discharges. However, due to incorrect early forecasts and an extremely late accurate one, the Bureau’s space building was unable to deal adequately with the large water supply of 1983. Similarly, large inflows continued through 1987. These inflows caused the Bureau to make the corresponding large discharges. The Bureau’s responsibility is to regulate flows to minimize the impact of heavy discharges. To attach liability to the Bureau as plaintiff wishes every time the Bureau made releases in response to insufficient storage and groundwater invaded the root zone of his crops would make a government agency responsible for whatever climatic conditions nature chooses to deliver. By undertaking to regulate flows on the Colorado River, the Bureau did not assure landowners that surface flooding and corresponding raised groundwater levels would not return. The Bureau did what it could with the situation presented to it by nature.
That a government agency could have forecast more accurately the heavy snow-melt does not place liability for a taking on the Government. Indeed, the theory ap*103pears tortious.24 Plaintiff did not establish that the Bureau could have accommodated the inflow without the heavy discharges had the predictions been accurate. Nor did plaintiff establish that earlier, more extensive flood control releases by the Bureau would have resulted in less or no damage to plaintiffs farmland. On this point, the following discussion between plaintiff’s counsel and defense witness Alden L. Briggs, Chief of the Bureau’s River Operations Policy Branch, Lower Colorado Region, took place:
Q. So, if you anticipated ... [the high flows from 1983 to 1987] by reducing the reservoirs, you could have avoided flood levels?
A. In theory, you’re right, if they were empty or reduced. Practically speaking, the primary purpose of the system as a whole — not just Lake Mead and Hoover Dam — is conservation of storage for future use____ [L]et’s say we had some ways of anticipating such large inflows, ... it would probably be impossible to drain reservoirs with some kind of anticipation of a flood threat.
Q. But, before November of ’82, you had no idea of the probability of flooding?
A. [I]f you define flooding as a probability of flood control releases, I did anticipate that.
Q. Well, what about flood control releases of a damaging magnitude?
A. I could not have anticipated that.
The evidence supports a finding that the large discharges on the Colorado River system were due to the exceptional inflow of 1983-1987. The court finds that natural precipitation, and not the Bureau, caused the high releases for this five-year period.
Gasser differed in another respect from the case at bar. Due to the aggradation of the riverbed near the plaintiff’s property resulting from low flows after the closure of Glen Canyon Dam, the owners’ land in Gasser experienced surface flooding. In this case, though, plaintiff failed to prove that the flood control structures caused aggradation of the river channel near plaintiff’s land to such an extent that the resulting higher river stages caused damaging groundwater levels under his property. In comparing historical water surface elevation levels to corresponding more modern levels at river mile 249, Mr. Matlock made the following statement:
The river surface elevation in ’82-’83 is above where it was historically [meaning 1902 through 1903], but below where it was in ’62-’63; the reason for that being that the high water flows of the ’83-’84-’85 period reduced the water elevation in this part of the river below what it was in ’62-’63 but still well above where it was in nature.
Concerning the same subject, the witness made the following specific statements during direct examination:
Q. Did you[r] study lead you to conclude that the flows in the river were higher in the 1960’s than they had been in nature, or lower on the average?
A. (No response.)
Q. Higher in elevation?
A. Yes, yes. I think it’s clear from the figures in 486-J that the elevations of the river surface in the 60s were higher than they were in the 1902/03 period.
Q. And the 1902/03 period, was a period before the river system improvements?
A. Yes.
*104Q. And the measurement of higher versus lower was at a given volume of water?
A. Yes.
The witness next testified that the present river channel near plaintiff’s property is a transition point between scour and deposition where the riverbed elevation changes very little, if at all.
Robert C. Brose, Chief of the Bureau’s River Development Branch, Lower Colorado Region, apparently both agrees and disagrees with Mr. Matlock’s testimony. Mr. Brose stated that the river bottom degraded near plaintiff’s farm by two to three feet during 1983 to 1985. Concerning historical versus modern riverbed elevations, the following dialogue between defense counsel and Mr. Brose took place:
A. [T]he elevation in the vicinity of the Plaintiff’s property is very close to the elevation of 1902/1903. Perhaps a little bit lower.
Q. Okay---- What’s ‘very close’ or ‘a little bit’?
A. Within a foot or two.
While Mr. Matlock’s testimony may be accurate in the technical sense, it mischaracterizes the flood control system’s impact on the change in water surface elevation from 1902-1903 to present. Figure 19 in 486-J relied on by Mr. Matlock compares only years 1902-1903 with years 1962-1963. It fails to break down the changes in elevation by year and thereby reflects only the net impact, not the intermediate changes. Therefore, it does not delineate the impact of either Hoover or Davis Dams on natural river surface elevations.
Figure 1 of the Vetter Report includes this critical information omitted from 486-J. It reflects that during the natural period of 1902 to 1935, the water surface elevation at Needles increased eleven feet. Given this eleven-foot natural increase and after considering that: 1) 486-J reports that the average river stage at Needles in 1973 was merely eight feet higher than in 1902-1903; 2) no evidence suggests that the average stages increased since that time; and 3) plaintiff’s property is located at a present transition point, the court finds that plaintiff failed to demonstrate that the river control structures aggraded the average riverbed elevation (and consequently increased the river stages) so as to cause damaging groundwater levels at his farm. Moreover, since the causation in Gasser resulted from aggradation after the closure of Glen Canyon Dam and this case concerns a period of degradation for the identical period, Gasser is distinguishable factually from this present case.
A surface flooding case, Cotton Land v. United States, 109 Ct.Cl. 816, 75 F.Supp. 232 (1948), involved both the Colorado River and the land plaintiff now owns. As a manifestation of the aggradation that occurred after the closure of Parker Dam in 1938, a large portion of the plaintiff’s 12,-000 acres became flooded or inaccessible. The Court of Claims held that the sediment deposition and corresponding flooding resulted directly from the dam’s impoundment of water. Hence, the court found a permanent taking.
Due to factual differences, Cotton Land’s holding does not predetermine the outcome of this case. Cotton Land involved surface flooding that inundated large portions of property. Plaintiff demonstrated easily that this flooding resulted directly from the closure of Parker Dam and the aggradation of the river channel resulting therefrom. An entirely different situation exists in the case at bar which involves subsurface flowage easements, not surface flooding. This is an inflow case, not a backwater or aggradation case. Further, the cause of the Cotton Land case has been removed to the extent that this same land is no longer endangered by backwater buildup from Lake Havasu.
More importantly, though, this case is much more nuanced than Cotton Land. Only one possible component of causation existed in the other case: the construction of Parker Dam. Here, however, causation is much more subtle. Damage to plaintiff’s land occurred 35 years after Parker Dam and the marsh were in place. Additionally, plaintiff attributes causation to the river, the marsh, or a combination of *105the two. Thus, the current condition is extremely dynamic and remarkably interactive. Unlike Cotton Land, no obvious single factor played a dispositive role in causation.
The court finds that neither action by the Bureau nor the existence of any river control structure caused damage to plaintiff’s subsurface drainage.
b. Topock Marsh
The analysis begins by noting that, in the area that is now Topock Marsh, a body of water exists where none was present in nature. The existence of this body of water raised the groundwater levels in the aquifer directly under the marsh. This artificially high water table altered to some degree the gradient of the aquifer in the part of the valley near the marsh. The question becomes, then, whether the presence of the marsh altered the gradient of the aquifer to such an extent as to cause the high groundwater conditions underneath plaintiff’s property.
Mr. Matlock testified extensively on the subject of. Topock Marsh’s impact on the groundwater regime in the Mohave Valley and, in particular, in the area near plaintiff’s land. When questioned as to how the presence of the marsh influences the groundwater in the valley, Mr. Matlock responded:
In the south end of the Mohave Valley, the Topock Marsh now forms a barrier to ground water outflow and there is flow from the marsh back into the valley____ [W]hat that does is to act to raise the elevation of this area, which I described as the ground water sink for the Valley, to new elevations.
This raising of the elevation of the cone of depression recognized by both Mr. Matlock and Dr. Sebenik altered the gradient of the aquifer between the cone and plaintiff’s farm. Dr. Sebenik disagrees with Mr. Mat-lock’s testimony that the marsh blocks all groundwater outflow. Dr. Sebenik testified that since water is fluid, the only way to prohibit outflow is to “build a dam or put concrete down 100 feet [into the ground].”
Through reliance on his own exhibit PX 34A, Mr. Matlock continued to testify concerning the marsh’s role in causation. The witness stated that in 1984 the groundwater invaded the root zone of plaintiff’s crops by three to five feet. Mr. Matlock next testified that without the marsh flooding, the groundwater invasion of the root zone would have been reduced. According to Mr. Matlock, if the inlet channel was not present during the high river flow periods, the invasion into the root zone would have been merely one to 1.5 feet in only the western portion of the property. In discussing how the witness arrived at the projections for the "without-the-flooding” and the “without-the-inlet-channel” marsh levels, plaintiff’s counsel asked the following question:
Q. Now, how did you make those projections related to what would have happened had there not been the marsh improvements?
A. For the [without-the-inlet] projections ..., we hypothesized that the marsh would have been operating on the average at an elevation approximately 5 feet lower or at an elevation of ... [451] feet and for the other projection without the marsh flooding, of course, we were using the existing 456.
On cross examination regarding the process by which the witness arrived at the without-the-inlet projection, Mr. Matlock responded: “It was an estimate based on our best judgment of what the effect would have been if the marsh elevation were somewhat lower than it is with its inlet channel in operation.”
Plaintiff’s counsel and the witness discussed next the impact on the groundwater levels at the property if the marsh did not exist at all:
Q. I understand it would have been less than 27 foot, but in terms of the impact on the Laughlin property root zone, would it have been significantly lower?
A. If this is a projection without the inlet works to the marsh and we assume in the next projection that there might have been a change of even 5 feet, it’s *106likely that there would have been no invasion of the root zone at all on the Laughlin property.
Q. Is that because the steeper the gradient, the greater the outflow?
A. Well, partly. But it’s also because of the fact that the water surface elevation throughout ... would be at a lower elevation.
(Emphasis added.) In response to a subsequent question by defense counsel on the same subject, the witness testified:
I told you that it was an estimate____ Now, we know certain things about the elevation of the water level in that area, but we have no history of what the water level in that area would have been without the marsh. So the only thing we could do would be to estimate what the effect might be.
Mr. Matlock’s theory of marsh causation, as derived from his testimony, reduced the process to these essentials: The marsh acts mainly to impede any groundwater outflow to the south of the valley and continuously to recharge the aquifer, even when the marsh is maintained at its normal level of 455-456 feet. The amount of recharge increases whenever the elevation of the marsh increases, as might occur due to factors such as water inflow from a heavy rainfall, a breach in a dike, or the creation of an inlet channel. Once out of the marsh, the recharge follows the gradient and flows northward into the cone of depression or sink hole, thereby raising the elevation of the bottom of the cone. This raised elevation changes the gradient between plaintiff’s property and the cone, making its slope more gradual. Since gradient is the difference in elevation between two points divided by their distance apart, the more gradual the slope, the higher the groundwater levels at a specific point. Thus, the groundwater table at plaintiff’s property is higher than it would be if the marsh or the marsh improvements were not in place. Based on this reasoning, Mr. Matlock attributes the high water table at plaintiff’s farm to the marsh.
While Mr. Matlock’s testimony on marsh causation is plausible, it is substantially theoretical in that it is based largely on projections, estimates, and assumptions. Furthermore, his analysis misconstrues the fundamental component of causation in this case. Relative to the period before 1983, no evidence suggests that the marsh’s existence or operation caused damaging groundwater levels at plaintiff’s property. Rather, the evidence shows that plaintiff’s land, and the land in the neighboring areas, was useable as productive farmland. For example, plaintiff was able to grow successful crops in 1981 and 1982. The inferences drawn by Mr. Matlock are the only evidence pointing to the marsh as a cause of blocked groundwater drainage during this time.
Damaging groundwater levels occurred only after the high water flows of 1983. After these high flows ended in 1987, the depth to water at plaintiff’s farm began to increase and gradually approach pre-1983 conditions. The court recognizes that plaintiff did not have a break-even year on the eastern half of his property until 1990 and that the western half remains largely unproductive. However, the high discharges of 1983 through 1987, and not the marsh, caused the damaging groundwater levels at plaintiff’s property. Without this five-year high flow period, no damaging water invasion would have occurred. Since the court found previously that the Government is not responsible for damage caused by large river flows in this situation, no role in causation can be ascribed to the marsh.
Moreover, the evidence plaintiff did introduce concerning actual causation by the marsh is insufficient to satisfy his burden of proof. Mr. Matlock’s analysis posited that the groundwater invasion would have been reduced if there was no marsh flooding or marsh improvements. Plaintiff did not prove that, but for the marsh, there would have been no groundwater invasion at all or that absent Topock Marsh the drainage from plaintiff’s property toward the cone of depression would have been non-damaging. Accordingly, Topock Marsh is not the direct cause of the damage to plaintiff's property.
*107The court has not based its findings on Dr. Sebenik’s riposte to Mr. Matlock. Due to no apparent fault of the witness, Dr. Sebenik’s causation analysis imploded with the introduction of a faulty computer-generated exhibit. Indeed, Dr. Sebenik’s theory that the cone of depression between the marsh and plaintiff’s property limited any groundwater invasion is plausible; unfortunately, the multipage computer-generated groundwater contour maps manifested irreconcilable interna] discrepancies. The court is unable to ascribe any weight to Dr. Sebenik’s testimony on point, although it should be noted that the witness was earnest and candid to a fault. The following statement not only reflects Dr. Sebenik’s frustration with the meager facts with which he had to deal, but accurately characterizes the paucity of facts upon which both experts constructed their opinions:
THE COURT: Have you testified before?
THE WITNESS: [Yes.] But this is the most difficult ... complicated case because we have so little real information and we have net effects of how the system operates____ I’m an expert now seven times. And this is by far the double or triple because of the complexity of the problem, the unknowns involved as far as making some opinions because we do not know — again, the system is so interactive. We don’t know how — we have recharges. We have discharges____ [A]ll we have is the net effect of those water levels.
The credibility of the experts played no role in the court’s findings relative to causation. For Mr. Matlock’s part, the witness’ testimony was almost consistently elicited by leading questions — without defendant’s objection. His testimony is considered at face value with no additional weight accorded for credibility. Dr. Sebenik’s testimony on the buffer effect of a cone of depression is entitled to no weight. Nonetheless, Mr. Matlock’s testimony is not binding on the trier of fact. Even standing alone his opinions are not decisive. “[The judgments] of an expert can be no better than the soundness of the reasons that stand in support of them.” Fehrs v. United States, 223 Ct.Cl. 488, 508, 620 F.2d 255, 265 (1980). “ ‘[I]t is axiomatic that the trier of fact is not bound to accept expert opinion, even if it is uncontradicted.’ ” Del Mar Avionics, Inc. v. Quinton Instr. Co., 836 F.2d 1320, 1325 (Fed.Cir.1987) (quoting Minnesota Mining & Mfg. Co. v. Berwick Indus. Inc., 532 F.2d 330, 333 (3d Cir. 1976)); see Sternberger v. United States, 185 Ct.Cl. 528, 535-36, 401 F.2d 1012, 1016-17 (1968) (“Even uncontradicted opinion testimony is not conclusive if it is intrinsically nonpersuasive.”)
Turning to the case law, Kansas City Life, 339 U.S. 799, 70 S.Ct. 885, 94 L.Ed. 1277, relied on by plaintiff, is inapposite. Before construction of a dam on the Mississippi River, respondent’s farm land drained adequately. After construction, the dam raised quickly the average level of the river, thereby destroying the land’s agricultural value through blockage of its surface and subsurface drainage. High river flows did not influence causation. The Supreme Court held that a taking had occurred. Like Cotton Land, Kansas City presented a circumstance where only one possible factor, the construction of a dam, could be responsible for causation. Because no such elementary solution is present in this case, Kansas City is not controlling.
3. Indirect and consequential damage
Assuming, arguendo, that the high groundwater levels at plaintiff’s farm did result from Topock Marsh, plaintiff still cannot prevail. It must be remembered that the marsh’s formation was accidental; plaintiff offered no evidence that formation was foreseeable. Further, no evidence suggests that the marsh can be removed. Before construction of the river control structures, i.e., the dams, reservoirs, and levees, plaintiff’s land was subject to the risk of continuous periodic overflows by floodwater, accompanied by a corresponding threat of increased groundwater levels. Any increase in threat to the land caused by the marsh is indirect and consequential.
In Sanguinetti v. United States, 264 U.S. 146, 44 S.Ct. 264, 68 L.Ed. 608 (1924), *108due to its location between a river and a slough, appellant’s property was overflowed continually. In an effort to improve river navigation, the Government constructed a canal connecting the slough and the river. Despite the engineers’ best calculations, the capacity of the canal proved insufficient to carry the water flow and, after periods of heavy rain, the canal overflowed, thus flooding appellant’s property. Appellant argued for a compensable taking. The Supreme Court disagreed, stating:
Prior to the construction of the canal the land had been subject to the same periodical overflow. If the amount or severity thereof was increased by any reason of the canal, the extent of the increase is purely conjectural____ It was not shown that the overflow was the direct or necessary result of the structure; nor that it was within the contemplation of or reasonably to be anticipated by the Government____
The most that can be said is that there was probably some increased flooding due to the canal and that a greater injury may have resulted than otherwise would have been the case ... [T]he injury was in its nature indirect and consequential, for which no implied obligation on the part of the Government can arise____
Id. at 149-50, 44 S.Ct. at 265 (citing cases); see United States v. Sponenbarger 308 U.S. 256, 265, 60 S.Ct. 225, 228, 84 L.Ed. 230 (1939) (no taking exists when flood control system aggravates volume of floods that would take place absent the system); ARK-MO Farms, Inc. v. United States, 209 Ct.Cl. 116, 119, 530 F.2d 1384, 1386 (1976) (no taking exists when system increases duration of raised water levels at lower elevations but decreased peaks, duration, and frequency of floods).
In Horstmann Co. v. United States, 257 U.S. 138, 42 S.Ct. 58, 66 L.Ed. 171 (1921), a situation arose where lakes’ surface elevations rose 19 feet after completion of Government-sponsored improvements, thereby destroying the value of appellants’ land. The Horstmann Court held:
[I]t would border on the extreme to say that the Government intended a taking by that which no human knowledge could even predict. Any other conclusion would deter from useful enterprises on account of a dread of incurring unforeseen and immeasurable liability. This comment is of especial pertinence____ [Tjhere is obscurity in the movement of percolating waters, and ... there was no evidence to remove ... [the improvements] in the present case, and necessarily there could not have been foresight of their destination nor purpose to appropriate the properties.
Id. at 146, 42 S.Ct. at 60; see Danforth v. United States, 308 U.S. 271, 286-87, 60 S.Ct. 231, 237, 84 L.Ed. 240 (1939) (where appellant found his land with the same level of flood protection after the construction of a new levee as before construction, no taking occurred). Where lake surface levels after completion of a public project were not as high as the natural levels before the project, “it is difficult to imagine that the ... [plaintiffs] can show a ‘direct appropriation’ resulting from ...” government conduct. Miller v. United States, 583 F.2d at 864, on remand, 480 F.Supp. 612 (E.D.Mich.1979) (finding no taking).
Thus, even assuming that government action damaged plaintiff’s property, the impact of the marsh is at most conjectural, indirect, and consequential. The Claims Court lacks jurisdiction to award damage for claims sounding in tort. 28 U.S.C. § 1491(a)(1) (1988); e.g., Somali Dev. Bank v. United States, 205 Ct.Cl. 741, 751-52, 508 F.2d 817, 822 (1974).
4. Salinity levels
Defendant, through witness Stanley Curtis Conway, the Bureau’s Chief of Irrigation Management Services Branch, attempted to demonstrate that the crop failure on the farm evolved from a combination of plaintiff’s failure to leach effectively the western portion of the property and the land’s history of high salinity and low productivity, and not from the river or the marsh. Defendant failed to substantiate this contention.
*109Mr. Conway testified concerning a Bureau land classification survey of Section 3 completed in 1966. The survey classified land according to its productivity and based its conclusions on the amount of expenditures required to create profitable land out of unproductive property, i.e., the more money necessary to generate profitable acreage, the lower its classification. Based on the relatively low classification of the southern half of Section 3, the witness inferred that the land was unproductive in 1966. Mr. Conway further testified that soil samples taken on April 19 and 20,1990, reflected high salinity on plaintiff’s property generally, with extremely high levels in some portions. The witness next stated that, in theory, leaching the property would accomplish salt removal.
While agreeing with the substance of the witness’ testimony, the court finds that the impact of these conditions on the property does not demonstrate fault or causation on plaintiff’s behalf. In 1966 plaintiff’s land received a low classification because numerous improvements were necessary to transform the land into a productive farm. Plaintiff undertook these betterments, thereby rendering moot the earlier classification. Further, plaintiff does not deny that the salinity level on the property’s western portion remains high, but, rather, argues that due to the high groundwater level leaching is ineffective to remove these salt depositions from the surface and the root zone. Plaintiff acted reasonably in attempting to eradicate the salt from his farm. Neither plaintiff’s farming practices nor the history of high salinity levels and low productivity are responsible for the salt which currently remains on the western portion of his property.
5. Permanent or frequent and inevitable recurrence
Assuming that the court found that government action was the cause of damage to plaintiff’s property, the court must characterize the temporal impact of the causation event. In a taking case, a claimant must demonstrate permanent or inevitable recurrence of the damaging government-caused condition or event in order to recover on his claim when he pleads either a permanent condition or intermittent events that allegedly will occur indefinitely. For example, in Barnes, wherein plaintiffs alleged a taking based on past intermittent flooding, plaintiffs succeeded because they offered evidence that future floods would “continue to be sufficiently frequent to constitute a taking.” 210 Ct.Cl. at 478, 538 F.2d at 872. This rule applies to flowage easements. “There is no difference of kind, but only, of degree, between a permanent condition of continual overflow by back-water and a permanent liability to intermittent but inevitably recurring overflows; and, on principle, the right to compensation must arise in the one case as in the other____” United States v. Cress, 243 U.S. 316, 328, 37 S.Ct. 380, 385, 61 L.Ed. 746 (1917); accord Accardi v. United States, 220 Ct.Cl. 347, 357, 599 F.2d 423, 429 (1979); Hartwig, 202 Ct.Cl. at 809, 485 F.2d at 619.
Mr. Matlock asserted at trial that when river flows reach a 15,000 c.f.s. level, groundwater invasion of the root zone of plaintiff’s crops would then begin. At 20,-000 and 25,000 c.f.s. levels, the invasion of and damage to the root zone would be more substantial. This witness drew those conclusions from PX 34A, discussed supra at p. 105. Due to the exhibit’s theoretical quality, it is not determinative as to when the onset of damage to plaintiff’s subsurface drainage actually began. For this case it is clear that plaintiff’s farm suffered damage only after the high flows of 1983-1987. The specific threshold of damaging flows, which is probably less than the 40,000 to 65,000 c.f.s. flows of 1983, is uncertain.
However, according to a joint 1982 Army Corps of Engineers and Bureau of Reclamation report entitled “Colorado River Basin, Hoover Dam, Review of Flood Control Regulation”:
As the Colorado River reservoir system fills ..., the flood potential along the river below Hoover Dam approaches conditions of the post-1935 period when flood releases from Hoover Dam in ex*110cess of 30,000 ... [c.f.s.] were more frequent____ Flood frequency studies in which a repetition of historical flows was assumed show that, when the reservoir system is filled, there would be about a 10-percent chance in any given year for sustained releases averaging 28,000 ... [c.f.s.] or more for 1 month from Hoover Dam____
The first conclusion of this report declares that “[a] high probability exists that flood control releases of a damaging magnitude will be required from Hoover Dam within the next 10 years.”
Mr. Matlock converted the measurements of DX 504, discussed supra at p. 91, from millions of acre-feet into c.f.s. in an attempt to explain the river flow in terms of c.f.s. This effort led Mr. Matlock to conclude that, in 84 years of river flow record-keeping, the average annual water supply to the river exceeded 15,000 c.f.s. 73 times, 20,000 c.f.s. 45 times, and 25,000 c.f.s. 22 times. This testimony is inclusive concerning recurrence. DX 504 reflects only the natural inflow into the river system. As noted by plaintiffs counsel, DX 504 does not purport to depict actual river flow or discharges; DX 505, discussed supra at p. 92, is the exhibit which details river flow. Any inference that because the average yearly inflow equals 20,000 c.f.s. the average river discharge would equal the same amount would be unfounded. The most that can be said is that DX 504 represents an average inflow. How this inflow translates into actual river flow is an entirely different question.
Since the filling of Lake Powell in 1980, the reservoir system has been full. Without excess storage capacity, the system now more closely resembles natural flow conditions than it did before 1980. The Bureau’s ability to control river flow is reduced, and correspondingly the probability of damaging flows is increased. With respect to plaintiff’s claim for a permanent taking, based on the above probabilities and the fact that the reservoir system has now reached capacity, the court finds that damaging high groundwater levels under the western half of plaintiff’s property will inevitably recur on a frequent basis.
Arguing for a temporary taking of all his property for 1983-1987, plaintiff reads First English Evangelical Lutheran Church v. County of Los Angeles, 482 U.S. 304, 107 S.Ct. 2378, 96 L.Ed.2d 250 (1987), as completely eliminating from takings jurisprudence the requirement that a taking be permanent or inevitably recurring. While agreeing in part with plaintiff, this court deems this assertion over-broad.
In certain circumstances the Government may be liable for a taking when the invasion of a plaintiff’s property is neither permanent nor inevitably recurring. Such situations arise when the Government conduct completely, but only temporarily, denies a landowner all use of his or her property. E.g., First English, 482 U.S. 304, 107 S.Ct. 2378, 96 L.Ed.2d 250 (where a county land-use ordinance deprived an owner all use of his property, subsequent revocation of the ordinance without compensation for the fair value of the property’s use was an insufficient constitutional remedy); Yuba Natural Resources, Inc. v. United States, 821 F.2d 638 (Fed.Cir.1987) (when the Government precluded an owner from exercising his mineral rights, but retracted this prohibition six years later, owner received compensation for six-year period). Thus, a taking may be found, absent permanent or inevitable recurrence, if government interference with the owner’s interest is pervasive but not perpetual. First English applies only to temporary takings, not to permanent ones. Contrary to plaintiff’s assertion, First English did not articulate a new rule applicable to all takings cases.
Apparently, only one case discusses a temporary taking in context of a flowage easement, the basis of plaintiff's claim.25 Cooper v. United States, 11 Cl. Ct. 471, rev’d on other grounds, 827 F.2d 762 (Fed.Cir.1987), involved a situation where by 1979 the Government’s construction of a waterway between the Tennessee River *111and the Gulf of Mexico resulted in blockage of the Tombigbee River near the Cooper farm. This blockage caused large volumes of floodwater to invade the low-lying areas of plaintiff’s land. This water remained standing during the spring, summer, and winter of each year and would inevitably recur each year thereafter. In 1984 the blockage was removed and the Cooper farm was no longer subject to flooding.
The Claims Court stated “the evidence ... show[s] clearly that the United States, ... did take by inverse condemnation a temporary flowage easement over the bottom land on the Cooper farm during the 1979-84 period....” Id. at 474-75. The court went on to hold that, since plaintiff was not the owner of the land when the taking occurred in 1979, he did not have standing to sue for a taking. Id. at 475-76. The Federal Circuit reversed on other grounds, stating, among other things, that plaintiff requested compensation for loss of timber, not a flowage easement. Cooper, 827 F.2d at 763. However, the court noted in dictum:
The trial court’s judgment ... is consistent with precedent dealing with takings of flowage easements. We think that, although the government may have taken a flowage easement, the plaintiff does not seek compensation for it. Therefore, this case is not controlled by the cases cited by the trial court dealing with flow-age easements.
Id. Thus, it appears that the Federal Circuit considers temporary takings of flow-age easements a valid part of takings jurisprudence.
In the instant case, from 1983 to 1987, high groundwater permanently saturated the root zone of plaintiff’s entire farm to such an extent that the farm was virtually non-productive. After 1987 the groundwater levels receded a sufficient amount so that some productivity returned to the eastern half of the property. Plaintiff effectively was deprived of any use of his farm for a five-year period. With regard to his claim for a temporary taking, plaintiff has met his burden by proving that any presumed taking was absolute, even if for only a temporary five-year time span. See Cooper, 11 Cl.Ct. at 475.
6. Relative benefits
Assuming that it were found that the Government’s action caused the high groundwater table, the inquiry does not end. Even if a causal relationship exists between the Government’s action and plaintiff’s damage and even if the invasion is either permanent or inevitably recurring, no liability attaches if the Government’s conduct bestowed more benefit than detriment on plaintiff’s property. This doctrine of relative benefits espouses that “if governmental activities inflict slight damage upon land in one respect and actually confer great benefits when measured in the whole, to compensate the landowner further would be to grant him a special bounty. Such activities in substance take nothing from the landowner____”26 Sponenbarger, 308 U.S. at 266-67, 60 S.Ct. at 229 (emphasis added); ARK-MO Farms, Inc., 209 Ct.Cl. at 119, 530 F.2d at 1386 (in denying relief, the court stated that the case involved, at most, minimal injury in comparison with much greater benefits conferred) (citing Sponenbarger); see also Danforth, 308 U.S. at 286, 60 S.Ct. at 237; Sanguinetti, 264 U.S. at 149-50, 44 S.Ct. at 265; Horstmann, 257 U.S. at 146, 42 S.Ct. at 59; Miller, 583 F.2d at 864. Where the Government installs a public works project, “the Fifth Amendment does not make the Government an insurer that the evil of floods be stamped out universally before the evil can be attacked at all.” Hartwig, 202 Ct.Cl. at 811, 485 F.2d at 621 (quoting Sponenbarger, 308 U.S. at 266, 60 S.Ct. at 229); see also Bartz, 224 Ct.Cl. at 594, 633 F.2d at 578; Accardi, 220 Ct.Cl. at 347, 599 F.2d at 429. While true that “only benefits *112inuring specifically to the condemnee, rather than to the community at large, are relevant to an analysis of the government’s liability under the Fifth Amendment”, City of Van Buren, Ark. v. United States, 697 F.2d 1058, 1062 (Fed.Cir.1983), “[i]t is settled that special benefits do not become general merely because other lands in the area are similarly benefited.” United States v. Crance, 341 F.2d 161, 167 (8th Cir.), cert. denied, 382 U.S. 815, 86 S.Ct. 36, 15 L.Ed.2d 63 (1965) (citing United States v. River Rouge Improvement Co., 269 U.S. 411, 415-16, 46 S.Ct. 144, 145-46, 70 L.Ed. 339 (1926)); see also United States v. Easements & Rights Over Certain Land, 259 F.Supp. 377, 380-81 (E.D.Tenn.1966) (where owner received direct benefit from the elimination of flooding and flood-related damage benefits were specific and not general). Although direct evidence of special benefits is helpful, see Herriman, 8 Cl.Ct. at 419 (data existed on the amount and duration of surface flooding absent the dam), a court may infer their existence. See, e.g., Bartz, 224 Ct.Cl. at 593, 633 F.2d at 577; Gasser, 14 Cl. Ct. at 505-06. In analyzing relative benefits, a court must balance the overall benefits and detriments of the Government’s conduct in relationship to their impact on a claimant’s property. Thus, in this case, the court will view the river and the marsh jointly in determining relative benefits.27
The relevant inquiry is threefold. First, what would be the use of plaintiff’s land without the dams, reservoirs, levees, and marsh. Second, what damage does plaintiff’s land realize because of these hydrologic features. And, third, which impacts plaintiff’s property more significantly, the benefits or the detriments.
In its natural state, the river often formed a braided stream, with its main artery 1000 feet wide in places. It meandered and migrated across the Mohave Valley and had a natural capacity of 25,000 to 30,000 c.f.s. near Needles. In 1905 the river meandered to within roughly a half-mile of the property plaintiff now owns.
Prior to construction of Hoover Dam, irrigation farming in the valley was virtually nonexistent. High flows along the lower river, caused by snowmelt and large rainstorms in the upper basin, occurred seasonally. Both Cotton Land, 109 Ct.Cl. at 821, 75 F.Supp. 232, and the Vetter Report state that damaging flows were common, with periods of surface flooding lasting one to two months. Furthermore, virtually the entire valley was at risk of overflow to a considerable depth. The closure of Hoover Dam in 1935 ended the annual spring floods in the Mohave Valley and consequently eliminated or greatly reduced a large amount of groundwater recharge resulting therefrom. Although the specific harm to the Mohave Valley historically was surface flooding, the experts concurred that higher groundwater levels result from higher river flows. Thus, the evidence illustrates that harm from surface flooding implies a corresponding harm through higher groundwater levels.
Mr. Matlock testified along similar lines when, in response to a question on direct examination concerning the characteristics of the river before Hoover Dam, he stated:
The river was typical of rivers in arid regions, that was extreme variability in the rates and duration of discharges; there was a high annual flood in the spring; the river was uncontrolled; it was meandering through the valleys where the width of the valleys were great enough for it to meander; it was scouring and depositing sediment intermittently as it moved through these valleys.
In the case at bar, since the levee system can safely contain flows in excess of at least 50,000 c.f.s., and possibly up to 75,000 c.f.s., see supra note 7, only the most extreme flood event would endanger the surface of plaintiff’s property under the *113present flood control system. As an example of the protection afforded by the levees, the discharges in 1983 were relatively high, varying at times from 44,000 to 65,000 c.f.s. If the river’s natural channel near Needles flowed at its 25,000 to 30,000 c.f.s. capacity without any levees in place, the flow would have escaped the channel and flooded parts of the valley. The same situation would have repeated itself in 1984 and 1986, when the average monthly discharges were 35,000 c.f.s. and 30,000 c.f.s., respectively. Instead, the levee improvements contained the flow in the channel and prevented any surface flood from occurring.
Mr. Brose testified persuasively that, absent the system of levees, plaintiff’s land, and a point paralleling plaintiff’s land approximately 1.5 miles to the east, would flood with discharges of 70,000 c.f.s. While identical discharges have not occurred in the recent past — Mr. Brose could not recall any — larger flows took place frequently before construction of Hoover Dam in 1935. See DX 505 supra at p. 92. Although the flows peaked at 65,000 c.f.s. during 1983, no direct evidence was introduced that plaintiff’s land would have been inundated without the protection of the levees. However, given the channel’s natural capacity close to Needles (two to three miles from plaintiff’s land) of 25,000-30,000 c.f.s.; the proximity of plaintiff’s land to the channel; and Mr. Brose’s DX 534, depicting the incursion of water into the valley at 70,000 c.f.s., the record supports the inference that absent the system of dams and levees, plaintiff’s land would have been flooded during the 1983-1987 interval.
After finding liability, the Claims Court in Gasser permitted recovery for one plaintiff, Mr. Mosqueda, on the basis that since the Government constructed the dams after the levees in Mexico were built, the dams did not provide a special benefit to this claimant’s property. 14 Cl.Ct. at 505. Here, though, the case is distinguishable in that the dams were built prior to the levees. More importantly, as discussed above, the dams did specifically benefit the plaintiff’s farm land.
Concerning the other plaintiff in Gasser, Mr. Bailey, the court denied recovery on the basis that the dams did benefit this property located adjacent to the river. In support of this finding, the court identified numerous historical high flows in the vicinity of Mr. Bailey’s property, and the flooding they caused. In conclusion, the court stated:
This review of historical data, contrasted with the location of Bailey’s property, persuades the court that without Hoover and Glen Canyon Dams, frequent and large scale overflow would have prevented development of Bailey’s lot. Because those uses and improvements for which he seeks compensation were made possible by the dams, the court concludes that the flooding of Bailey’s property did not result in a compensable taking.
Id. at 506. Apparently, the court relied on no direct evidence of pre-Hoover Dam flooding of Mr. Bailey’s land. Rather, the court inferred that since land in the vicinity of the disputed property flooded under natural conditions, Mr. Bailey’s property would have suffered a similar fate without the protection afforded by the dams.
The ending of the surface flooding and the general regularization of river flow also impacted the farmability of land in the Mohave Valley. Before 1950 there were at most a few hundred acres of irrigated valley farmland. A precise present acreage of such farmland is not evident in the record, but it is obvious that most, if not all, of the historical flood plain is now used as irrigated farmland. Without the flood control system, plaintiff’s land in its natural state largely would have been unusable for farming purposes. Thus, Hoover Dam and its successor improvements directly benefited plaintiff’s land by providing a stable source of irrigation water and by greatly reducing the likelihood of surface flooding and, hence, damaging groundwater levels. As stated by the Trial Commissioner in Cotton Land. “Between June 1914 and 1940 the land, taken as a whole, was not used for agricultural purposes, except that part of the time part of it was used for grazing cattle.” Cotton Land v. United States, No. 46422, ¶ 23 (Cl.Ct. Comm’r *1141951) (unpubl.). In spite of this finding, the Court of Claims on review did not consider relative benefits in its ruling.
Without the dams, reservoirs, and levees, plaintiffs land would be at risk of surface water flooding every spring. Likewise, groundwater levels at plaintiffs farm would increase dramatically whenever surface water flooded or encroached upon the property. With the construction of the flood control system, such possibilities no longer are a hazard to the land plaintiff farms. The dams, reservoirs, and levees regulate the river flow so effectively as to practically eliminate the threat of surface flooding to plaintiffs land.
Since plaintiff did not prove causation, no detriment to his property resulted from any act of the Government. Consequently, the benefit to the farmland outweighs the detriment and plaintiffs claim fails under the doctrine of relative benefits. However, assuming that plaintiff did succeed on every element of causation that he advanced, he would have established the following detriments to his land: First, the river improvement projects caused the river to aggrade at a point parallel to plaintiffs land, thereby raising groundwater levels in the area including plaintiffs farm. Second, the existence of the marsh raises the elevation of the cone of depression to such an extent that groundwater now invades the root zone of his crops. Third, without the marsh or marsh improvements, the groundwater invasion of the root zone would have been either nonexistent or greatly reduced in 1983 through the present.
After balancing these detriments against the benefits of the river improvement projects, plaintiffs claim cannot overcome the relative benefits balancing test. While acknowledging that plaintiffs property suffered some harm from the river and marsh (assuming he had proved that the Government’s operation of either caused the damage), his property realized far greater benefits than detriments. Without the projects plaintiff’s land would have flooded and the groundwater level would have reached the surface, instead of two feet below. Plaintiff cannot succeed in his claim that the Government’s actions caused the subsurface drainage of his property to be less effective than it would be without the river improvement projects or the marsh.
CONCLUSION
Based on the foregoing, judgment shall enter for defendant, and the Clerk of the Court is directed to dismiss plaintiff’s complaint.
IT IS SO ORDERED.
No costs.
. An "alluviar valley is one composed of different sediments, such as silt, sand, gravel, and clay. The Colorado River gradually deposited these sediments over the entire floor of its flood plain. This flood plain encompasses the portion of the valley covered by river floods prior to construction of Hoover Dam, but is wider than the meandering course of the river. The maximum width of the flood plain in the valley is five miles, which includes plaintiffs property.
The court notes that flood plain does not necessarily imply that every time a flood occurred river water submerged the entire surface of the valley. Rather, it more likely implies that when the river flooded, it expanded to cover various parts of the valley at different times.
. The Lower Colorado River starts just below the Glen Canyon Dam. Utah is the rough border between the river’s upper and lower regions.
. A "river mile” refers to the distance between the point where the Colorado River crosses the international border with Mexico and any other given point along the river to the north. Thus, river mile 250 is 50 miles closer to Mexico than river mile 300. The river mile nearest plaintiffs farmland is 249.
. The capacity of a body of water to transport sediment is largely a function of the velocity and sediment content of the flow. Other factors governing sediment transportation capacity are depth and temperature of flow. When water leaves a dam, it is "clean," Le., it carries little sediment. It also travels at a relatively high rate of speed. These two factors combine to create "scour,” a process by which river water picks up sediment consisting of, in part, dirt, sand, gravel, and rock, and carries it downstream. This has the effect of lowering the river floor, a process called “degradation.”
Further downstream the river water reaches a state of equilibrium wherein it neither picks up sediment from the river bottom nor deposits it on the river floor. This occurs at the "transition point" in the river in between the dam and downstream lake. Below this transition point, as the water approaches a lake, the river water’s velocity slows and sediment falls out. This deposition process is known as "aggradation” and has the effect of raising the elevation of the river floor.
. See infra note 13 and accompanying text.
. Big Bend, located approximately five miles south of the Davis Dam and 13 miles north of Needles, is a section of the river where it forms an elongated horseshoe-shaped bend.
. The design of the levee system allowed it to safely contain river flows in excess of 50,000 c.f.s. Defense witness Robert C. Brose, Chief of the Bureau’s River Development Branch, Lower Colorado Region, estimated that the levees presently can accommodate flows in excess of 75,-000 c.f.s.
. In 1955 the Bureau undertook a related project in which it dredged and widened the river in order to create a "settling basin.” This basin is located south of Needles and stretches for about five miles above the Topock Bridges. By reducing water flow speeds, the basin forces sediment to drop out of the water and collect on the river bottom. The Bureau dredges the basin whenever the sediment levels accumulate sufficiently, thereby theoretically preventing a recreation of the 1944 emergency situation.
. Recordation of river flow data began in 1906.
. The court deleted projections for the year 1990.
. Regulations require that Lake Mead have a minimum of 1.5 million acre feet of storage every day of each year which increases gradually to 5.35 million acre feet by January 1 of each year. In November 1982 the Bureau had 6.6 million acre feet of storage in Lake Mead.
. In 1983 the control dike at the marsh’s southern end ruptured, causing large amounts of water to flow northward from the river into the marsh. The water level in the marsh then was controlled directly by the surface elevation of the river.
. “Recharge" means the groundwater flow out of a body of water, such as a river or marsh, and into the aquifer.
. Although numerous farms elsewhere were under Mr. Banning’s control, plaintiffs was the only property in the valley that he farmed.
. Capillary flow results when the capillary forces exceed the gravitational pull, so that the water, and the salts mixed therein, actually moves upward from the water table. Due to capillary action, the water below plaintiffs farm rose between three and four feet above the groundwater level. This distance that the water rises is called the "capillary fringe.”
. "Gin trash” is a by-product of the cotton refining process.
. "Leaching” is a process where the land’s surface is flooded with water for an extended period, thus allowing water to carry salts downward out of the root zone. For leaching to be effective, the groundwater level must be low so the floodwater will have drainage potential.
. Loosely speaking, "gradient” means slope between two points. Specifically, it is the difference in vertical elevation between two points divided by the amount of their horizontal separation. In the context of this case, gradient signifies the slope of the aquifer that the groundwater follows downward. If the gradient is steep, groundwater will flow quickly. If the gradient is subtle, or near level, groundwater flow virtually will be nil.
. “Transmissivity” means the ability of the aquifer to carry water. A high value denotes the possibility of transferring large quantities of groundwater very quickly. A low transmissivity value implies that only low-speed, low-volume groundwater transfers are possible. The transmissivity value depends largely on the composition and texture of the aquifer. The aquifer in the Mohave Valley has the highest value of any aquifer along the Colorado River.
. A "cone of depression” is a place where the groundwater levels are lower than the groundwater levels which surround it. Hence, the surrounding water flows, or "sinks”, into this cone or hole. Pumping of water from a well, especially from a series of proximate wells, commonly causes the creation of these cones.
. The statute in its entirety provides: “Every claim of which the United States Claims Court has jurisdiction shall be barred unless the petition thereon is filed within six years after such claim first accrues."
. The Supreme Court went on to imply that it is unjust to place "on the owner the onus of determining the decisive moment in the process of acquisition by the United States when the fact of taking could no longer be in controversy." Dickinson, 331 U.S. at 748, 67 S.Ct. at 1384; Alder v. United States, 785 F.2d 1004, 1008 (Fed.Cir.1986) (quoting the identical language from Dickinson).
. For example, assume that on January 1, 1983, there exist 5,000,000 acre feet of vacant storage in Lake Mead. Further assume a water inflow into Hoover Dam for 1983 of 23,000,000 acre feet. Simple physics instruct that the Bureau has no choice but to release a minimum of 18,000,000 acre feet of water in order to offset the inflow. Additionally, if the Bureau intends to recreate the 5,000,000 acre feet of vacant storage, it must then discharge the entire 23,-000,000 acre feet of inflow.
. To the extent that plaintiff's claim is tortious, 33 U.S.C. § 702c (1988), bars recovery. The statute states, in pertinent part: "No liability of any kind shall attach to or rest upon the United States for any damage from or by floods or flood waters at any place____” "Because of section 702c, the government’s acts to store, divert, and release waters to further flood control are subject to no risk of liability." James v. United States, 760 F.2d 590, 603 (5th Cir.1985); see also Aetna Ins. Co. v. United States, 628 F.2d 1201 (9th Cir.1980); Hayes v. United States, 585 F.2d 701 (4th Cir.1978). "[W]hen a landowner is damaged by release of water through a dam or a change in flooding patterns is caused by construction of levees, courts have frequently denied compensation, saying that the flooding involves at most a tort, not a taking or an implied contract claim____” Miller v. United States, 583 F.2d 857, 863 (6th Cir.1978).
. While United States v. 79.39 Acres of Land, 440 F.2d 1190 (6th Cir.1971) briefly mentions temporary flowage easements, the case does not address the concept's validity.
. The Sponenbarger Court went on to state: "While this Court has found a taking when the Government directly subjected land to permanent intermittent floods to an owner’s damage, it has never held that the Government takes an owner’s land by a flood program that does little injury in comparison with far greater benefits conferred____” 308 U.S. at 267, 60 S.Ct. at 229 (footnotes omitted).
. Although the court treated the river and the marsh separately for causation purposes, it is treating these features differently for relative benefits purposes. The reason for this disparate treatment is that plaintiffs two theories of causation must be considered separately. Regarding relative benefits, however, since the benefits and detriments must be measured "in the whole,” there is no reason to balance these features in isolation of each other.