This lawsuit stems from the unfortunate drowning of three fishermen near Lock and Dam 13 (LD13) on the Mississippi River. Plaintiffs filed under the Suits in Admiralty Act, 46 U.S.C. § 741-52, claiming that the accident was the result of inadequate warnings posted in and around LD13 and misconduct by dam employees, who alleged*288ly opened the dam gates in an attempt to wash decedents downstream. The trial court, sitting without a jury, found that the warnings posted on LD13 were adequate under the circumstances and held that there was insufficient evidence to support the allegation of employee misconduct. The court determined that decedents’ own negligence in fishing near the dam was the sole cause of the accident. Plaintiffs now claim that the court’s findings of fact are clearly erroneous.
I. Facts
LD13 spans 1066 feet across the Mississippi River, from Illinois on the east to Iowa on the west. On the eastern portion on LD13 there is an operating lock chamber and an inoperable auxiliary lock. The dam is made up of thirteen movable gates that control the flow of water. The middle gates, numbers 5, 6, and 7, are large, “roller” gates. The gates on either side are smaller, “tainter” gates. The gates are separated by concrete “piers” that extend 40 feet into the river from the gates. Ladder rungs are built into the downstream face of the pier, known as the “piernose.”
Downstream from LD13 there are a series of “baffle blocks” that are built into the concrete foundations in the river. The baffle blocks are designed to prevent erosion by dissipating water energy. The blocks produce varying amounts of surface and subsurface turbulence, depending on factors such as the gate settings and water level. The surface turbulence may be visible as boiling water or white caps, while the subsurface turbulence is not visible to boaters. The dissipation of water energy also produces a phenomenon known as a “break-line.” A breakline is created when the water near the baffle blocks is raised above the level of the water directly in front of the dam, causing the current to flow back to the dam. The water on the downstream side of the breakline continues to flow away from the dam. As there are several sets of baffle blocks there will be several break-lines. The location of the breaklines will vary according to gate settings and water levels, but anyone fishing near the dam will notice that the back-current is carrying fishing lines and debris toward the dam.
Other than the physical design of the dam, the facts were contested. After reviewing the evidence presented by the parties the court made extensive findings of fact. The court found that LD13 is an attractive area for fishing, but is also a hazardous one. Until 1968 the public was advised to stay 300 feet away from the downstream side of the dam. In 1968 the “restricted” area was changed to 100 feet. There was conflicting testimony at trial as to how often fishermen violated the restriction and came within 100 feet of the dam. Plaintiffs presented testimony indicating that boaters frequently entered the restricted area, while dam employees testified that such instances were rare. The court did not resolve this contradiction, but did note that there was no credible proof that anyone had tied up to the pier as decedents did, or had even come as close to the pier as decedents did. The court also noted that the lockmen were supposed to warn boaters observed entering the restricted area, but were under no duty actively to police the area.
On the evening of June 22, 1977, decedents went fishing near LD13 in a 16-foot fiberglass boat. All three men were avid fishers, and decedent Schlatter had fished within the 100 foot restricted area in the past. Around 7:00 p.m. decedents approached Charles Munson, who was fishing 200 feet downstream from LD13, and discussed the fishing conditions with him. After trolling behind Munson for awhile, decedents proceeded to the dam. At about 7:30 p.m. Munson observed one of the decedents tie a line from their boat to the ladder on the piernose between gates 6 and 7. Another fisherman, Terry Cram, arrived in the fishing area about 7:30 p.m. and observed decedents fishing 500-600 feet downstream from the dam, indicating that Munson must have observed decedents tie their boat to the piernose sometime after 7:30 p.m.
Carolyn and Richard Trude observed decedents fishing after they tied to the pier-nose. Carolyn testified that one of the decedents would pull the boat up to the pier *289and then let it drift downstream for 10-15 feet. Decedents repeated this maneuver several times while Carolyn watched. The Trudes testified that they heard cries for help around 8:00 p.m., as did another fisherman, John Burman. No witness actually saw decedents’ boat capsize. The first person to see a post-accident occurrence was fisherman Matzen, who immediately began unsuccessful rescue efforts. After reviewing Matzen’s testimony the court concluded that it was consistent with an 8:00 p.m. accident.
After being informed of the accident, the lockmen closed the gates. Decedents’ empty boat was caught in the current in front of gate 6 and was being slammed into the gate. The mooring line, trolling motor and part of the gunwale were still attached to the piernose. Decedents’ bodies were recovered three days later. None of the men were wearing life-vests, although the boat was equipped with an ample supply. Trude, Martzen, and Burman all testified that they saw decedents’ heads and arms above water as they were washed downstream.
II. Warnings
Plaintiffs’ principal contention is that the warnings posted on and around LD13 were inadequate in that they did not warn boaters of the type of danger presented by the dam. The complained of warnings consisted of the following:
(1) A sign, eight feet by five feet, posted on the intermediate wall of the auxiliary lock, which read “Restricted. Keep Below This Point.”
(2) Three red lights that appear to be aligned when viewed from the 100 foot restriction line, but which lose their alignment when viewed from within the 100 foot restriction.
(3) Three orange lights that work the same way as the lights in (2), but warn boaters when they are within 300 feet of the dam.
(4) A sign, 20 inches by 28 inches, posted on the downstream face of the river wall next to gate 1 that reads: “Danger. Keep 100 Feet From Dam.”
(5) A sign similar to that in (4), posted on the pier between gates 12 and 13, which advises boaters to keep 300 feet from the dam.
(6) At two different places across the downstream side of the dam the word Restricted appears, which can be read from the river more than 300 feet downstream.
(7) A yellow sign with black letters covers the width of pier 6. Prior to 1968 the sign said: “Restricted. Keep 300 Feet From Dam.” In 1968 the restricted area was reduced to 100 feet and the digit “3” was painted over with white paint. A digit “1”, however, was not painted in, leaving the sign to read “Restricted. Keep 00 Feet From Dam.” Anyone near the sign could see that the “3” had been painted over. Decedents tied their boat to the pier directly below this sign.
At trial various witnesses testified that “Restricted” meant “stay out” and was an absolute prohibition rather than simply a warning of danger. One witness, Raymond Wainscott, saw the sign on pier 6 when fishing with decedent Schlatter several weeks before the accident. Wainscott, who also saw the sign on the intermediate lock wall, testified that he understood the signs to mean not to go within 100 feet of the dam. The one witness who did not read the signs this way was plaintiffs’ expert witness, Dr. Casky. Dr. Casky testified that “restricted” meant “stay off” and that the signs could be read as merely prohibiting boaters from climbing on the dam. The trial court did not credit Dr. Casky’s reading as it required one to ignore “Keep 00 Feet From Dam,” which clearly indicated more than that one should not climb on the dam. The court found that it was sufficient that the decedents were warned of an absolute prohibition against entering the 100 foot restricted area and that it was not necessary that the nature of the danger be specified.
After the trial in this case had ended this court decided Callas v. United States, 682 F.2d 613 (7th Cir.1982). In Callas decedents *290allowed their boat to drift into the auxiliary lock chamber on LD8 on the Mississippi River, unaware that they would be swept into a roller gate by the back-current created by the dam. Decedents’ boat was swept into the gate and capsized. LD8 was equipped with only one sign, which read “Danger — Keep 00 Feet Away” because the overlay for the digit “1” had not arrived. LD8 also employed warning lights similar to those at LD13. We agreed with the district court that the warnings were inadequate because they did not reveal the nature of the danger they purported to warn of. “Ultimately, we believe that this was a case in which the government induced reliance on a belief that it was providing something which, in fact, it was not providing — an adequate warning of the dangers posed by the dam.” Id. at 623. The Government, having “lulled [decedents] into a false sense of security” with the sign, was responsible for the fatal consequences.
Plaintiffs argue that Callas requires, as a matter of law, that the Government specify the nature of the danger. We do not agree. Callas did nothing more than apply the long recognized rule that when the Government warns the public of a danger, it must do so with due care. 682 F.2d at 622; see also Indian Towing Co. v. United States, 350 U.S. 61, 76 S.Ct. 122, 100 L.Ed. 48 (1955). In Callas we agreed with the district court that the Government was negligent in posting the specific sign in issue because it purported to describe the dangers presented without actually doing so. That is not the case here. Unlike a “danger” sign, which merely advises boaters that there are risks attendant upon boating near the dam, a “restricted” sign tells boaters that they may not enter the area under any circumstances. That the other warnings, such as the red and orange lights, may be ineffective is irrelevant.
Plaintiffs’ expert witness admitted that it is sufficient either to apprise boaters of the nature of the danger or to describe what action must be taken to avoid danger. In this case the signs indicated that boaters should keep out of the restricted zone. In fact, Dr. Casky was also a witness in Callas, and in that case stated that the problem with the lone sign at LD8 was that it failed to state that boaters should keep away from the dam, a problem not present with the signs at LD13. The Government, having clearly warned boaters to stay away from the dam, was under no further duty to explain the nature of the danger.1
The remaining issue is whether the sign, sans the digit “1,” was sufficient to inform decedents to stay away from the dam. Dr. Casky testified that the sign, reading “Keep 00 Feet From Dam,” simply meant to keep off the dam. The district court did not accept this reading, and neither do we. The sign clearly meant that boaters were to stay away from the dam, not just off of it, and the only ambiguity was the distance boaters were to keep between themselves and LD13. A reasonable interpretation was that boaters should keep *291at least 100 feet from the dam.2 Even assuming that the average boater would not read the sign this way, any ambiguity was cleared up by the sign on the intermediate lock wall, a sign that was visible to anyone in decedents’ position.
The district court specifically found that dam employees were not aware that boaters were coming as close to the dam as decedents did and rejected the claim that the lockmen acquiesced in the violations of the restriction. We cannot say that the court’s findings on this are clearly erroneous and must reject plaintiffs’ claim that defendants knew that the warnings were ineffective and were under a duty to take further steps to enforce the 100 foot restriction. Similarly, there was no evidence that any of the lockmen saw decedents tie up the night of the accident. While the lock-men may have been under a duty to warn boaters observed violating the restriction, they were not obliged actively to police the area to protect boaters from their own folly-
III. Opening of Gate 6
Plaintiffs’ second claim before the trial court was that the lockmen, observing decedents fishing in front of the dam, opened the gate in an attempt to wash decedents downstream and out of the restricted zone. The resulting surge of water caused the boat to capsize and led to the drownings. In advancing this theory plaintiffs rely upon an unexplained rise in the water level, fisherman Matzen’s testimony regarding decedents floating downstream on a surge of water, and the type of damage done to the boat.
In rejecting this claim the trial court did not just adopt the defendants’ proposed findings of fact, but rather reviewed the evidence and made extensive findings on its own. On review we are not empowered to retry the case but are instead limited to determining whether the district court’s determination that plaintiffs failed to carry the burden of proof on their claim was clearly erroneous. Fed.R.Civ.P. 52(a). We may not reweigh the credibility of the witnesses, as plaintiffs would have us do, and we may only reverse the court’s decision if we are “left with the definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum Co., 333 U.S. 364,395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948); see also Nickerson v. Commissioner, 700 F.2d 402, 405-06 (7th Cir.1983).
A recording mechanism located downstream from the dam measured a rise in the water level at 7:42 p.m. The water level did not recede until after the accident was reported and the gates closed around 8:30 p.m. Plaintiffs claim that the rise in the water level was caused by the lockmen opening gate 6, which in turn led to the accident. The linchpin of this claim is, of course, proving that the accident happened contemporaneously with the rise in the water level. The district court considered the testimony of the fishermen near LD13 at the time of the accident, much of which has been set forth in the beginning of this opinion, and determined that the accident occurred at 8:00 p.m., negating any claim that decedents’ boat capsized because of a gate opening.
The testimony of various fishermen was not exact. Nobody actually saw the accident and, of necessity, the district court had to rely on estimates provided by the boaters in determining the time of the accident. Most of this evidence consisted of guesses concerning the amount of time spent on *292activities prior to and after the accident, so it is not surprising that plaintiffs can now point to evidence that contradicts the court’s conclusion. None of the evidence plaintiffs rely on is the type of irrefutable, physical evidence that would compel us to reverse the court’s findings. We see no reason to discuss the conflicting testimony, but suffice it to say that we do not find the court’s conclusion regarding the time of the accident to be clearly erroneous.
Because the accident did not occur when the rise in the water level took place, there is no reason to spend much time discussing plaintiffs’ remaining evidence in support of the gate opening theory. We certainly would be going beyond our appellate province to try the case de novo, which to some extent the appellants seem to want us to do. The court was not required to credit Matzen’s testimony that he saw a surge of water, and his claim that he heard of lock-men opening gates on boaters 10-15 years earlier is of minimal value. Similarly, the court was not required to disbelieve the express denials of the lockmen, all of whom were engaged in locking through a vessel during the rise in water level — which may account for the increase. We are equally unconvinced that plaintiffs’ expert established that the damage sustained by decedents’ boat could only have been caused by a surge of water rather than being slammed into the roller gate.
Although it is not necessary to review the court’s finding that decedents’ negligence was the sole cause of the accident in light of plaintiffs’ failure to establish any basis for holding defendants liable, we do note that the evidence is clear that decedents, grown men with extensive boating experience, chose to tie their boat in front of the gate of an operating dam despite warning signs and the obvious danger of placing themselves in this position. Decedents’ carelessness was compounded when they failed to take the simple precaution of wearing the life vests that were in the boat, a step that might well have saved their lives. Given decedents’ willingness to take risks with their lives, the resulting tragedy could reasonably have been expected.
For the reasons stated herein, the decision of the district court is Affirmed.
. In addition to the clear message conveyed by the sign, the court found that the cause of the danger was obvious from the surface turbulence. Plaintiffs argue that the court was not permitted to consider testimony concerning the water conditions on the night of the accident because the parties stipulated that several photographs, taken long after the accident, represented conditions similar to those existing the night of the drownings. Plaintiffs argue that this stipulation required the court to accept that the water was calm the night of the accident. In making this argument plaintiffs ignore that the water depicted in the photographs is not calm, and is in fact quite turbulent near the gates. Furthermore, assuming that the water was calm, this does nothing to negate that decedents, experienced boaters, knew that they had tied their boat in front of the gate of an operating dam and must have known that this involved substantial risk. Witnesses at trial testified that decedents had obviously placed themselves in a dangerous situation and that nobody in the past had been so foolhardy as to tie up to a piemose as decedents did. In addition, from decedents’ position the backcurrent and breakline should have been visible and decedents were undoubtedly aware of the risk they were taking. In short, the facts fully support the court’s determination that the danger was open and obvious and that decedents’ own negligence was the sole cause of the accident.
. In Callas the court rejected the Government’s claim that “Keep 00 Feet Away” should be read as indicating a minimum of 100 feet, principally because this interpretation “assumes that boaters would have recognized the operational defect in the sign.” 682 F.2d at 623. In Callas the sole sign was new and was equipped with replaceable digits. The digit “1” had not arrived and there was no indication from the sign itself that a digit was missing. Here, however, the digit “3” had been painted over and it was clear that the sign was incomplete. In this situation it would be unreasonable to read the sign as actually “00 Feet.” Furthermore, the sign was not the only one posted and a reasonable boater would have read it in conjunction with those signs indicating 100 feet.