Tarrant Regional Water District v. Richard Johnson and Sharkara Johnson, Individually and as Personal Representatives of the Estate of Brandy Johnson

IN THE SUPREME COURT OF TEXAS ══════════ No. 17-0095 ══════════ TARRANT REGIONAL WATER DISTRICT, PETITIONER, v. RICHARD JOHNSON AND SHARKARA JOHNSON, INDIVIDUALLY AND AS PERSONAL REPRESENTATIVES OF THE ESTATE OF BRANDY JOHNSON, RESPONDENTS ══════════════════════════════════════════ ON PETITION FOR REVIEW FROM THE COURT OF APPEALS FOR THE SECOND DISTRICT OF TEXAS ══════════════════════════════════════════ Argued October 30, 2018 JUSTICE BLACKLOCK delivered the opinion of the Court. JUSTICE BUSBY did not participate in the decision. On January 16, 2013, Brandy Johnson attempted to cross the Clear Fork of the Trinity River in Fort Worth by walking across Trinity Park Dam No. 2. She was five months pregnant and wearing heavy clothing. In the middle of the dam is a ten-foot-wide kayak chute, through which the river swiftly flows. Brandy lost her footing on the dam, slipped into the river, and drowned. Brandy’s parents sued the Tarrant Regional Water District, which constructed and maintains the dam from which Brandy fell. The District, however, is a governmental entity and therefore generally immune from suit unless the Legislature has waived its immunity. Our task in this appeal is to examine the Tort Claims Act to determine whether the Legislature has authorized this suit against the District. We conclude that it has not. For the reasons explained below, section 101.056 of the Tort Claims Act, known as the discretionary function exception, applies here. As a result, governmental immunity bars the Johnsons’ claims. We therefore reverse the judgment of the court of appeals in part and render judgment dismissing all claims against the District. I. BACKGROUND The Tarrant Regional Water District is a water control and improvement district created under article XVI, section 59 of the Texas Constitution. Tarrant Regional Water Dist. v. Gragg, 151 S.W.3d 546, 549 (Tex. 2004). Its responsibilities include flood protection. TEX. WATER CODE § 51.121(b). The District is responsible, along with the U.S. Army Corps of Engineers, for maintaining flood control along the Fort Worth Floodway, which includes the Clear Fork of the Trinity River. In the 1960s, the District built a series of dams as part of a joint effort with the Corps of Engineers to channelize part of the Trinity River to help with flood control. The original dams were vertical-face dams, which have a steep drop off. The Corps’ original design plans from 1966 indicated that, below the dam, the river bottom would be graded to a depth of approximately 3.5 feet in order to facilitate water flow. In 2002, the District redesigned several dams along the Clear Fork of the Trinity River, including Dam No. 2, to allow for kayaks and other small vessels to pass through the dams as they navigate the river. In 2003, the District demolished the old dams and constructed new ones. As the District was constructing the new version of Trinity Park Dam No. 2, it discovered that the old dam had created a “scouring effect” downstream, the result of water flowing vertically over the dam and eroding the bottom of the river. According to the affidavit of Louis Verreault, the District’s Dam Levee Safety Engineer, by 2003 the river bottom had eroded to a depth of roughly eight or nine feet below the bottom of the dam. Darrell Beason, the District’s then-Director of Operations, testified that rather than filling in the eroded area and raising the depth to the 1966 design level of 3.5 feet, the dam design engineer decided the deeper river bottom should remain in 2 place to prevent kayakers and tubers from injuring themselves when passing through the chute. Beason testified that, under the new dam design, the riverbed immediately downstream from the kayak chute needed to remain at “a depth of at least eight feet.” The District graded the river bottom accordingly. After Brandy’s death, her parents, the Johnsons, sued the District. The Johnsons sought to overcome the District’s immunity by alleging the accident was caused by the use or condition of the District’s tangible or real property. See TEX. CIV. PRAC. & REM. CODE § 101.021(2). They alleged both premise-defect and special-defect theories of property-owner liability. See TEX. CIV. PRAC. & REM. CODE § 101.022. In their third amended petition, the Johnsons alleged that the kayak chute was slippery and that the current running through it was deceptively dangerous and difficult to see. They also alleged that there were no signs warning of the strong current and slippery surface. Additionally, the Johnsons claimed that the scour hole at the base of the dam and the “hydraulic boil effect” it created were not visible or obvious to a person crossing the dam and that no signage warned specifically of these dangers. They further alleged that the District had knowledge of these dangerous conditions because of previous incidents at this location. The District filed a plea to the jurisdiction. Among other arguments, it contended that all the Johnsons’ claims are barred because of section 101.056 of the Tort Claims Act, which creates an exception to the waivers of immunity otherwise provided by the Act. Section 101.056 states: This chapter does not apply to a claim based on: (1) the failure of a governmental unit to perform an act that the unit is not required by law to perform; or (2) a governmental unit’s decision not to perform an act or on its failure to make a decision on the performance or nonperformance of an act if the law leaves the performance or nonperformance of the act to the discretion of the governmental unit. 3 “The provision generally preserves immunity not only for the state’s public policy decisions, but also for the state’s failure to act, when no particular action is required by law.” Stephen F. Austin State Univ. v. Flynn, 228 S.W.3d 653, 657 (Tex. 2007). The District argued that section 101.056 shields it from any liability that might otherwise flow from its discretionary decisions related to the design of the dam and its safety features. Additionally, the District argued that the Johnsons could not establish a premise-defect claim sufficient to waive its immunity. The trial court denied the District’s plea to the jurisdiction. On interlocutory appeal, the court of appeals affirmed in part and reversed in part. 514 S.W.3d 346, 350 (Tex. App.—Fort Worth 2016). The court rejected the Johnsons’ effort to establish a waiver of immunity based on either a misuse-of-personal-property theory or a special- defect theory. That left only the premise-defect theory remaining. The District argued that the Johnsons had not properly stated a premise-defect claim sufficient to waive immunity since the danger to Brandy was open and obvious. The court of appeals rejected this argument. It determined that because the District only argued in its plea to the jurisdiction that the dangers related to the kayak chute and swift current were open and obvious, and not that the dangers related to the scour hole and alleged boil effect were open and obvious, the trial court properly denied the District’s plea to the jurisdiction on that basis. The court of appeals held that the Johnsons’ claims regarding the dam itself (i.e., the slippery nature of the kayak chute and the swiftly flowing current passing through the chute) were barred in light of section 101.056 because those features of the dam were the result of the District’s intentional design decisions. See Flynn, 228 S.W.3d at 657 (distinguishing “the design of public works, for which there is immunity, from their maintenance, for which there is not immunity”). The court of appeals also held that the Johnsons’ claims regarding the adequacy of the District’s warning signs were all complaints related to the design of 4 the dam and were thus barred by immunity due to section 101.056. However, the court of appeals held that the scoured or eroded riverbed and the resulting possibility of a “hydraulic boil” were not related to the dam’s original design but resulted from the District’s failure to maintain that design. On this basis, the court of appeals upheld the trial court’s denial of the District’s plea to the jurisdiction as to the Johnsons’ premise-defect claim based on the scour hole and alleged boil effect. The District petitioned for review. The Johnsons did not. In this Court, the District continues to argue that, by operation of section 101.056, immunity bars all the Johnsons’ claims, including those based on the scoured area of the riverbed and the alleged boil effect. The District also re-urges its argument that the Johnsons’ premise-defect claim does not support a waiver of immunity because Brandy chose to encounter the open and obvious risk that she might drown. 1 II. STANDARD OF REVIEW A. Texas Tort Claims Act “Sovereign immunity protects the state and its various divisions, such as agencies and boards, from suit and liability, whereas governmental immunity provides similar protection to the political subdivisions of the state, such as counties, cities, and school districts.” Travis Cent. Appraisal Dist. v. Norman, 342 S.W.3d 54, 57–58 (Tex. 2011). Tarrant Regional Water District is a political subdivision of the State of Texas. See TEX. WATER CODE § 51.149(e). It is therefore 1 The Johnsons attempt to raise a cross-appeal issue in their brief on the merits. They ask the Court to examine the court of appeals’ conclusion that their claims related to the rapid current flowing through the kayak chute, the slippery bottom of the chute, the original scour hole, and the adequacy of the District’s warning signs were all barred by operation of section 101.056 because those features were discretionary design decisions. Texas Rule of Appellate Procedure 53.1 states that “[a] party who seeks to alter the court of appeals’ judgment must file a petition for review.” The Johnsons did not file a petition for review challenging the court of appeals’ dismissal of these claims. As a result, we do not consider these arguments. 5 generally immune from suit and liability absent a waiver of its immunity. See Gragg, 151 S.W.3d at 550. The Legislature may waive immunity by statute or legislative resolution. Fed. Sign v. Tex. S. Univ., 951 S.W.2d 401, 405 (Tex. 1997). The Texas Tort Claims Act creates a limited waiver of immunity. TEX. CIV. PRAC. & REM. CODE §§ 101.001–.109. The Act waives sovereign and governmental immunity, subject to restrictions, in the following three areas: “use of publicly owned automobiles, premises defects, and injuries arising out of conditions or use of property.” Tex. Dep’t of Parks & Wildlife v. Miranda, 133 S.W.3d 217, 224 (Tex. 2004) (quoting County of Cameron v. Brown, 80 S.W.3d 549, 554 (Tex. 2002)); TEX. CIV. PRAC. & REM. CODE §§ 101.021– .022. B. Evidentiary Standard An assertion of governmental immunity implicates the trial court’s jurisdiction. Rusk State Hosp. v. Black, 392 S.W.3d 88, 91 (Tex. 2012). Thus, immunity is properly asserted in a plea to the jurisdiction. Houston Belt & Terminal Ry. v. City of Houston, 487 S.W.3d 154, 160 (Tex. 2016). Parties may submit evidence at the plea-to-the-jurisdiction stage, and the trial court’s review generally mirrors the summary judgment standard. Sampson v. Univ. of Tex., 500 S.W.3d 380, 384 (Tex. 2018). “If the evidence creates a fact question regarding the jurisdictional issue, then the trial court cannot grant the plea to the jurisdiction, and the fact issue will be resolved by the fact finder. However, if the relevant evidence is undisputed or fails to raise a fact question on the jurisdictional issue, the trial court rules on the plea to the jurisdiction as a matter of law.” Miranda, 133 S.W.3d at 227–28; see also Klumb v. Houston Mun. Emps. Pension Sys., 458 S.W.3d 1, 8 (Tex. 2015). The trial court’s ruling on the plea is reviewed de novo on appeal. Miranda, 133 S.W.3d at 228. 6 III. ANALYSIS A. The Johnsons’ Premise-Defect Claim The Tort Claims Act waives immunity for “personal injury and death so caused by a condition or use of tangible personal or real property if the governmental unit would, were it a private person, be liable to the claimant according to Texas law.” TEX. CIV. PRAC. & REM. CODE § 101.021(2). A premise-defect claim is a common instance of a claim for “injury and death caused by a condition . . . of real property.” Id.; see State v. Shumake, 199 S.W.3d 279, 281 (Tex. 2006). If the claim “arises from a premise defect, the governmental unit owes to the claimant only the duty that a private person owes to a licensee on private property . . . .” TEX. CIV. PRAC. & REM. CODE § 101.022(a). The duty owed to a licensee on private property requires that “a landowner not injure a licensee by willful, wanton or grossly negligent conduct, and that the owner use ordinary care either to warn a licensee of, or to make reasonably safe, a dangerous condition of which the owner is aware and the licensee is not.” Sampson, 500 S.W.3d at 385 (quoting State Dep’t of Highways & Pub. Transp. v. Payne, 838 S.W.2d 235, 237 (Tex. 1992)). The Johnsons’ sole remaining claim is a premise-defect claim based on the alleged dangerous condition of the scour hole and the hydraulic boil. The Johnsons allege that, because of previous incidents at the site, the District was aware of the dangerous condition—the scour hole and alleged hydraulic boil—that they claim caused Brandy’s death. The Johnsons further allege that Brandy did not understand the dangers associated with the scour hole and boil effect because neither were visible, and a nineteen-year-old would have no reason to know of such dangers without clear and noticeable warning signs. The Johnsons allege that the District’s small signs were inadequate to convey the dangers of the scour hole and boil effect. 7 B. The Discretionary Function Exception The outcome of this appeal turns on section 101.056 of the Tort Claims Act, often referred to as the discretionary function exception. Section 101.056 provides: [The Tort Claims Act] does not apply to a claim based on: (1) the failure of a governmental unit to perform an act that the unit is not required by law to perform; or (2) a governmental unit’s decision not to perform an act or on its failure to make a decision on the performance or nonperformance of an act if the law leaves the performance or nonperformance of the act to the discretion of the governmental unit. Section 101.056 preserves immunity “for the state’s failure to act, when no particular action is required by law.” Flynn, 228 S.W.3d at 657. The exception “avoid[s] judicial review or interference with those policy decisions committed to the other branches of government.” Id. In Flynn, we characterized the existing case law on section 101.056 as articulating “more than one test for determining when questioned conduct involves a protected ‘discretionary’ determination.” Id. (quoting 19 WILLIAM V. DORSANEO III, TEXAS LITIGATION GUIDE § 293.12[7] (2007)). The first test “distinguishes between policy-level decisions and operational-level decisions.” Id. The second “distinguishes the design of public works, for which there is immunity, from their maintenance, for which there is not immunity.” Id. Flynn illustrated the latter test by suggesting that while a “government is not liable for designing a bridge without lighting, it may be liable for failing to maintain the lighting on a bridge designed to be illuminated.” Id. at 657– 58. Although described at times as two distinct inquiries, both of these “tests” have the effect of distinguishing “between policy-level decisions and operational-level decisions.” Id. at 657. The design versus maintenance “test” is simply the policy-level versus operational-level test applied to public works. The interconnectedness of the two “tests” is illustrated in our mingled application 8 of them in Flynn: “[T]he decisions here concerning when and where the water was to spray were operational- or maintenance-level decisions, rather than policy formulation.” Id. at 658. The dam at issue in this case is a public work, and the court of appeals focused its analysis on the “design versus maintenance” dichotomy suggested in Flynn. See 514 S.W.3d at 356–63. The court of appeals essentially drew a sharp line between “design” and “maintenance,” under which anything on the “maintenance” side of the line is not covered by the discretionary function exception. The Johnsons’ complaint regarding the scour hole and boil effect is that the District failed to maintain the riverbed below the dam at its designed depth. The court of appeals concluded that because this claim is based on an alleged “failure to maintain,” it is not covered by the discretionary function exception. Id. at 362. Of course, neither “design” nor “maintenance” appears in the text of section 101.056. For that matter, neither do the terms “policy-level” or “operational-level.” These labels provide useful conceptual frameworks that aid courts and litigants in the application of the rather obscure text of section 101.056. Without doubt, section 101.056 is difficult “to meaningfully construe and consistently apply.” Flynn, 228 S.W.3d at 662 (Hecht, J., concurring). But no court, including this one, can alter or augment statutory text by announcing tests to aid the application of the statute. City of Rockwall v. Hughes, 246 S.W.3d 621, 631 (Tex. 2008) (“[C]hanging the meaning of the statute by adding words to it, we believe, is a legislative function, not a judicial function.”). The words of section 101.056 protect a governmental entity’s failure to act or its decision not to act (or its failure to make any decision at all) on matters within “the discretion of the governmental unit.” They do not protect a governmental entity’s failure to act when a particular action is “required by law.” TEX. CIV. PRAC. & REM. CODE § 101.056. The statute’s textual distinction—“discretion” versus “required by law”—is not always easily applied on its own terms, which is why 9 interpretational rubrics like “design versus maintenance” and “policy-level versus operational- level” have arisen and serve an important purpose. But the interpretational rubrics are only useful to the extent they yield results faithful to the statute’s textual distinction between discretionary and non-discretionary government decisions. City of Rockwall, 246 S.W.3d at 628 (Tex. 2008) (“[B]y not reading language into the statute when the legislature did not put it there, we do not risk crossing the line between judicial and legislative powers of government as prescribed by article II of the Texas Constitution.”). Examining this Court’s prior decisions reveals that our application of section 101.056 to public works has consistently been grounded in the statute’s focus on preservation of the government’s discretionary decision-making authority, rather than on the often-useful but extra- textual distinction between design and maintenance. We first explicitly held that the design of public works is a discretionary function protected by section 101.056 in State v. Rodriguez, 985 S.W.2d 83, 85 (Tex. 1999) (per curiam), overruled on other grounds by Denton Cty. v. Beynon, 283 S.W.3d 329 (Tex. 2009). In Rodriguez, a truck driver died when he hit a bridge abutment on the side of a highway detour. Id. The detour, which routed traffic onto a frontage road to circumvent a construction project on U.S. Highway 77, involved a 90-degree turn. Id. The driver failed to make the turn, instead side-swiping the bridge abutment, which resulted in his truck rolling over. Id. The driver’s widow alleged the detour was unreasonably dangerous and had inadequate warning signs. Id. We concluded that “[u]nder section 101.056, the State retained its immunity for the detour design because the design was a discretionary act.” Id. at 86. Further, we held that “[d]esign of any public work, such as a roadway, is a discretionary function involving 10 many policy decisions, and the governmental entity responsible may not be sued for such decisions.” Id. at 85. 2 State v. San Miguel involved a car accident on an elevated freeway ramp in Houston. 2 S.W.3d 249, 250 (Tex. 1999). Part of the guard railing was missing on the exit ramp, and the Texas Department of Transportation placed barrels in front of the missing railing to warn drivers of the danger. Id. A driver lost control of his vehicle, drove through the barrels, and fell to the ground. Id. We held that the State retained its immunity because “[d]ecisions about highway design and about what type of safety features to install are discretionary policy decisions.” Id. at 251. Additionally, we noted that “[a] court should not second-guess a governmental unit’s decision about the type of marker or safety device that is the most appropriate.” Id. Likewise, in Tex. Dep’t of Transp. v. Ramirez, we held that the Department of Transportation was immune from a suit alleging a dangerous highway median because “the median’s slope and the lack of safety features, such as barriers or guardrails, reflect discretionary decisions for which TxDOT retains immunity under the Act’s discretionary-function exception.” 74 S.W.3d 864, 867 (Tex. 2002). In County of Cameron v. Brown, 80 S.W.3d 549 (Tex. 2002), we first recognized the possibility of a distinction between the design of public works and their maintenance. In Brown, a driver died on the Queen Isabella Causeway, a bridge connecting South Padre Island and the mainland, after losing control of his vehicle. Id. at 552–53. When the accident occurred, over thirty streetlights on the bridge at the scene of the accident were not working. Id. at 553. The State owned the bridge, but the county was responsible for maintaining the streetlights. Id. The 2 Similarly, in State v. Burris, we were asked whether a vehicle making an illegal turn onto a highway or briefly stopping on the highway could constitute a premise defect or a special defect for which the State’s immunity is waived. 877 S.W.2d 298 (Tex. 1994). We held that it could not. While that decision did not directly implicate the discretionary function exception, we briefly noted at the end of the opinion that the State would also be immune for “discretionary roadway design under section 101.056 of the Tort Claims Act.” Id. at 299. 11 decedent’s family sued the county. Id. The trial court granted the county’s plea to the jurisdiction. Id. at 554. The court of appeals reversed, holding that the maintenance of the bridge’s streetlights was not a discretionary function covered by section 101.056. Id. We affirmed, although we did not address the court of appeals’ application of the discretionary function exception because the county did not challenge it in this Court. Id. at 559. In responding to concerns that our decision would require governmental entities to choose between designing public roads with no illumination or ensuring that no light ever malfunctioned, we stated, “[a] governmental unit’s sovereign immunity is not waived for failure to install lighting, which is a discretionary decision . . . . Our decision rests upon the causeway’s unique characteristics and the nature of the particular dangerous condition alleged.” Id. at 557. Importantly, the county in Brown had no decision to make about whether to keep the streetlights working. The county had a contract with TxDOT requiring the county to maintain the streetlights. The distinction in Brown between the State’s discretion to design the lighting system and the county’s lack of discretion to let it fall into disrepair provided a vivid example of the difference between discretionary design decisions protected by section 101.056 and non- discretionary maintenance functions, which are not protected. We employed this very example again in Flynn, where we suggested that while a “government is not liable for designing a bridge without lighting, it may be liable for failing to maintain the lighting on a bridge designed to be illuminated.” 228 S.W.3d at 657–58 (emphasis added). Neither Brown nor Flynn provide that, in every case involving premise-liability claims about a public work, a sharp distinction can or must be drawn between the design of the public work and its maintenance. Although Flynn characterizes “design versus maintenance” and “policy-level versus operational-level” as two separate tests, our application of the statute merged 12 the two tests into a single consideration of whether the governmental unit’s challenged action was covered by section 101.056. Id. at 658 (“[T]he decisions here concerning when and where the water was to spray were operational- or maintenance-level decisions, rather than policy formulation.”). To summarize, while the “design versus maintenance” dichotomy is a useful way to think about how section 101.056 operates with respect to public works, it is not an element of the statute. Narrow consideration of whether the claim involves a design function or a maintenance function must not completely displace consideration of the statute’s textual touchstone: “discretion.” Otherwise, “judicial review or interference with those policy decisions committed to the other branches of government” would be allowed whenever the policy decisions concern maintenance of a public work. Id. at 657. Section 101.056 does not permit that result. C. Application of the Discretionary Function Exception to the Johnsons’ Claims With these observations in mind, we must consider whether, applying section 101.056, immunity bars the Johnsons’ remaining claims related to the scour hole and hydraulic boil. The Johnsons do not dispute that the District’s decisions related to the depth of the river at the base of the dam are discretionary design decisions protected by section 101.056. Rather, they claim that the District never actually decided to grade the riverbed to a depth of eight feet following the 2003 reconstruction. They claim that the existing scour hole was not adopted as an intentional design decision and that in 2002 the District decided to bring the depth back up to the original 1966 depth of roughly three feet. They then claim that the reason the scour hole returned to its greater depth was that the District negligently filled the hole with loose materials which were not sealed in place, and as a result, the materials washed away. While these allegations, taken as true, might place the Johnsons’ claims outside section 101.056, they do not raise a material issue of fact 13 given the evidence offered by the District regarding the 2003 redesign of the dam. See Miranda, 133 S.W.3d at 227–28 (Tex. 2004) (“If the evidence creates a fact question regarding the jurisdictional issue, then the trial court cannot grant the plea to the jurisdiction, and the fact issue will be resolved by the fact finder. However, if the relevant evidence is undisputed or fails to raise a fact question on the jurisdictional issue, the trial court rules on the plea to the jurisdiction as a matter of law.”). The only record evidence regarding what depth the District chose for the scour hole after the dam’s 2003 reconstruction establishes that the District decided to grade the riverbed below the dam to at least eight feet in order to facilitate the safety of kayakers. None of the evidence the Johnsons point to contradicts the testimony of the District’s engineer and the District’s Director of Operations that the design for the new dam built in 2003 required the existing plunge pool (at the time roughly eight or nine feet below the dam) to remain at a depth of at least eight feet after the construction of the dam. In the face of this evidence, the court of appeals did not credit the Johnsons’ assertion that the only consciously chosen design of the dam had a 3.5- feet-deep plunge pool. Nor do we. The Johnsons also argue that, even if the District intentionally chose the eight-foot depth in 2003, the further deepening of the riverbed and possible hydraulic boil are the result of the District’s “failure to maintain” the eight-foot depth of the riverbed. The court of appeals agreed, concluding that injury caused by the District’s failure to keep the riverbed at a depth of eight feet, just as it was after the 2003 reconstruction, can be attributed to a “failure to maintain” the depth and therefore falls outside the scope of the discretionary function exception. 514 S.W.3d at 362. We disagree. To begin with, the record indicates the District intentionally graded the channel downstream from the dam to a depth of at least eight feet to ensure kayakers would not hit the river bottom while rolling or flipping. A design of at least eight feet suggests that any deepening 14 of the riverbed beyond that depth was consistent with the District’s design. The Johnsons, however, maintain that an eight-foot depth was the maximum allowed by the design. The court of appeals seems to have taken the Johnsons’ view of that question. Id. Even if it were correct that the 2003 redesign called for a static eight-foot riverbed depth, the court of appeals improperly elevated the “design versus maintenance” test over the underlying statutory inquiry into the District’s discretion. Section 101.056 “preserves immunity not only for the state’s public policy decisions, but also for the state’s failure to act, when no particular action is required by law.” Flynn, 228 S.W.3d at 657. We have never held that decisions about how to conduct maintenance on a public work can never rise to the level of a “policy decision committed to the other branches of government” protected by section 101.056. 3 Id. at 657. Any such holding would be tantamount to saying the government is “required by law” to maintain all its public works exactly as they were originally constructed. No such requirement exists in the law. 4 To the extent the law has anything to say about the District’s obligations with respect to river depth, it reinforces the District’s discretion. The Texas Water Code states that a water control and improvement district may provide for the control of its water and floodwater “by any practical means,” suggesting the Legislature vested significant discretion in the District regarding such matters. TEX. WATER CODE § 51.121(b), (c). Likewise, the federal regulations governing the maintenance of channels and floodways do not impose a requirement that channels be maintained at a particular depth. See 33 C.F.R. § 208.10(g)(1). 3 The District asserts that refilling the portion of the channel immediately below the dam would have cost roughly $310,000. This expense may have recurred as the riverbed continuously eroded. 4 Conversely, we have never held that the design of a public work can never exceed a governmental entity’s discretion or otherwise conflict with requirements of law. It plainly could, and in a hypothetical case, section 101.056’s protections might not apply to all decisions about the design of a public work. The imperfection of the “design versus maintenance” dichotomy thus works both ways. Here, we consider only the “maintenance” side of the equation because the Johnsons’ only remaining claim is that the District failed to properly maintain the dam. 15 It must be remembered as well that the “public work” the District is alleged to have improperly maintained is the natural bed of a flowing river. Analyzing the riverbed as if it were a structural public work already stretches credulity. The notion that the District had a legal obligation to keep this natural “public work” at a constant depth beneath an opaque and running body of water is unsupportable. We have previously held that section 101.056 makes a governmental body “immune from liability for discretionary decisions concerning the expenditure of limited resources for the safety of its citizens.” City of Corsicana v. Stewart, 249 S.W.3d 412, 416 (Tex. 2008) (per curiam); see also Ramirez, 74 S.W.3d at 867. That principle applies to this case and dictates the outcome. It does not yield every time the District’s discretionary decisions can be characterized as the “failure to maintain” a public work. 5 Nevertheless, the “design versus maintenance” test identified in Flynn remains useful. Notably, the faulty streetlight example from Brown and Flynn—the most common articulation of the “design versus maintenance” distinction—differs significantly from this case. A road is designed for drivers, who are invited to use it. While the road design may be discretionary, the failure to keep the road safe for drivers in accordance with its design may not be. Unlike a city that undertakes to build safe roads for drivers, the District did not set out to design a dam that would be safe for people to walk across or safe for those who slip off its surface into the river. It set out to design a dam that would channelize the river and allow kayakers to pass over it. In 5 The Johnsons also claim the District loses under the “policy-level versus operational-level” test because the negligent implementation of policy-level decisions is not protected by the discretionary function exception. Once again, the “design versus maintenance” test and the “policy-level versus operational-level” test are not two distinct inquiries. The “design versus maintenance” test is simply the “policy-level versus operational-level” test applied to public works. The Johnsons claim that the District’s failure to maintain the 1966 design grade of the dam constitutes policy implementation for which the District does not retain immunity. This argument fails. As discussed above, the record establishes that the District made a policy-level decision to grade the river bottom immediately below the dam to a depth of no less than eight feet in order to protect kayakers. The Johnsons take issue with this policy decision— not the District’s implementation of it. 16 allocating taxpayer resources, the District had discretion to direct its maintenance efforts toward the dam’s intended purposes of river control and kayaking rather than toward protection of those who inadvisably use the dam to cross the river. A government that invites drivers onto a bridge may in some cases have no discretion other than to “maintain the lighting on a bridge designed to be illuminated.” Flynn, 228 S.W.3d at 658. By contrast, any suggestion that the District had no discretion other than to maintain the riverbed at a constant level finds no support in this record or in any other source. 6 IV. CONCLUSION The District could have maintained the riverbed at a constant eight-foot depth if it chose to do so. It could also have raised it to three feet. But the District’s failure to make the judgments the Johnsons claim it should have made about the proper allocation of taxpayer resources is the kind of “policy decision[] committed to the other branches of government” that section 101.056 shields from judicial second-guessing. 7 Id. at 657. 6 We need not reach the District’s argument that the scour hole and alleged boil effect were open and obvious because we find that, irrespective of the answer to that question, the District maintains immunity under section 101.056. The District also argues that the Johnsons’ premise-defect claim fails because there was not evidence sufficient to raise a genuine issue of material fact as to the existence of a hydraulic boil which contributed to Brandy’s death. The District contends that the court of appeals erred in allowing the testimony of a layperson to controvert the District’s expert testimony regarding the existence of a hydraulic boil. Once again, we do not reach these arguments since we find that the District maintains immunity under section 101.056. 7 The District asks us to abandon our previous decisions interpreting section 101.056 and adopt the federal case law interpreting 28 U.S.C. § 2680(a), the federal discretionary function exception. This, the District believes, would remedy what it calls our “hazy construction” of section 101.056. The Johnsons and the State of Texas as amicus curiae oppose this suggestion. “The exceptions to the waiver of sovereign immunity contained in the Texas Tort Claims Act are patterned after those contained in the Federal Tort Claims Act.” Driskill v. State, 787 S.W.2d 369, 370 (Tex. 1990). While the language in the Texas statute is similar to the language in the federal statute, it is by no means identical. The fact that the Texas Legislature intentionally patterned the Texas statute after the federal statute while still using noticeably different language counsels against wholesale incorporation of the federal case law. This does not mean the federal cases should be ignored. At bottom, they struggle with the same question facing Texas courts—whether challenged government action “is of the kind that the discretionary function exception was designed to shield.” United States v. Gaubert, 499 U.S. 315, 322–23 (1991). Federal courts answer this question by asking whether the challenged decisions were “grounded in social, economic, and political policy.” Id. at 323. Here, the District’s decisions involved social and economic policy related to the purposes of the dam and the cost of maintaining the depth of the riverbed. 17 The judgment of the court of appeals is affirmed in part and reversed in part, and judgment is rendered dismissing all claims against the District. __________________________________ James D. Blacklock Justice OPINION DELIVERED: April 12, 2019 18