City of Austin v. Leggett

OPINION

At approximately 6 p.m. on November 15, 2001, seventeen year-old Nathan Leg-gett tragically drowned after attempting to drive through a flooded street in southwest Austin. That afternoon, it was undisputed that the Austin area had been hit by thunderstorms with intense rainfall, hail, torna-dos and widespread flooding. Nathan’s mother, Trudy Leggett, individually and as Nathan’s heir, sued the City of Austin for damages under the survival statute and wrongful death act. She alleged that the City’s negligent maintenance or design of a stormwater detention pond, located north of the intersection where Nathan drowned, had caused debris to clog a grate covering the pond’s designed drainage outlet, resulting in storm waters backing up and ultimately overflowing the pond, flooding the adjacent residential area and causing Nathan’s death.1

Leggett’s suit implicates the City’s governmental immunity, the long-established common-law doctrine that categorically bars suits for money damages against municipalities unless the legislature has consented to suit. See, e.g., City of Galveston v. State, 217 S.W.3d 466, 469 (Tex.2007); Tooke v. City of Mexia, 197 S.W.3d 325, 332 (Tex.2006); Reata Constr. Corp. v. City of Dallas, 197 S.W.3d 371, 374 (Tex.2006).2 Leggett purports to assert claims *461within the legislative waivers of immunity under the tort claims act for damages claims based on theories of premises defects and “special defects.” See Tex. Civ. Prac. & Rem.Code Ann. § 101.022(a)-(b) (West Supp.2007). Asserting that its immunity against Leggett’s claims had not been waived, the City filed a plea to the jurisdiction. The trial court denied the plea, specifically finding “as a matter of law that the condition was a special defect.” The City appeals this order. See id. § 51.014(a)(8) (West Supp.2007).3 Concluding that Leggett’s suit does not fall within the tort claims act’s waivers of immunity, we must reverse and render judgment dismissing the suit for want of subject-matter jurisdiction.

STANDARD OF REVIEW

A challenge to a trial court’s subject-matter jurisdiction may be asserted in a plea to the jurisdiction. Texas Dep’t of Parks & Wildlife v. Miranda, 138 S.W.3d 217, 225-26 (Tex.2004); Bland Indep. Sch. Dist. v. Blue, 34 S.W.3d 547, 554 (Tex.2000). Whether a court has subject-mat ter jurisdiction is a question of law. Miranda, 133 S.W.3d at 226. The determination of whether a trial court has subject-matter jurisdiction begins with the pleadings. Id. The pleader has the initial burden of alleging facts that affirmatively demonstrate the trial court’s jurisdiction to hear the cause. Id. (citing Texas Ass’n of Bus. v. Texas Air Control Bd., 852 S.W.2d 440, 446 (Tex.1993)). Whether the pleader has met this burden is a question of law that we review de novo. Id. We construe the pleadings liberally and look to the pleader’s intent. Id. If the pleadings do not contain sufficient facts to affirmatively demonstrate the trial court’s jurisdiction but do not affirmatively demonstrate incurable defects in jurisdiction, the issue is one of pleading sufficiency and the plaintiffs should be afforded the opportunity to amend. Id. at 226-27. If the pleadings affirmatively negate jurisdiction, then a plea to the jurisdiction may be granted without allowing an opportunity to amend. Id. at 227.

However, “a court deciding a plea to the jurisdiction is not required to look solely to the pleadings but may consider evidence and must do so when necessary to resolve the jurisdictional issues raised.” Bland, 34 S.W.3d at 555. “When a plea to the jurisdiction challenges the existence of facts alleged by the pleader to establish the trial court’s subject-matter jurisdiction, the trial court must consider relevant evidence submitted by the parties.” Miranda, 133 S.W.3d at 227 (citing Bland, 34 S.W.3d at 555). To varying degrees, a jurisdictional challenge may also “implicate the merits of the pleader’s cause of action.” Id. at 227-28 (describing overlapping jurisdictional and merits inquiry regarding challenge to whether parks and wildlife department acted with gross negligence so as to waive sovereign immunity under the recreational use statute). When the consideration of a trial court’s subject-matter jurisdiction requires the examination of evidence, the trial court exercises its discretion in determining whether the jurisdictional determination should be *462made at a preliminary hearing or await fuller development of the case, mindful that the jurisdictional determination must be made as soon as practicable. Id. at 227-28.

“[I]n a case in which the jurisdictional challenge implicates the merits of the plaintiffs cause of action and the plea to the jurisdiction includes evidence, the trial court reviews the relevant evidence to determine if a fact issue exists.” Id. at 227. This standard, which “generally mirrors that of a summary judgment under Texas Rule of Civil Procedure 166a(c),” seeks to reconcile “the fundamental precept that a court must not proceed on the merits of a case until legitimate challenges to its jurisdiction have been decided” while “protecting] the interests of the state and the ... claimants in cases ... in which the determination of the subject matter of the court implicates the merits of the parties’ cause of action.” Id. at 227-28. Accordingly, when reviewing a plea to the jurisdiction in which the pleading requirement has been met and evidence has been submitted to support the plea that implicates the merits of the case, we take as true all evidence favorable to the non-movant and indulge every inference and resolve any doubt in the non-movant’s favor. Id. at 228. Whether the evidence presents a fact question regarding a jurisdictional fact is a question of law that we review de novo. Id. Unless a pled jurisdictional fact is challenged and conclusively negated in this manner, it must be taken as true for purposes of determining subject-matter jurisdiction. Id. at 226.

THE RECORD

Pleadings

Leggett alleges the following pertinent facts:

The location of the drowning of Nathan is not known to have flooded before. A nearby retention pond, designated by the City ... as “Pond # 342,” was supposed to drain flood waters out of the opposite side of the pond, and away from the location of the incident. The City maintains and inspects Pond #342.

The City responded to a complaint received on April 12, 2001, during a rainstorm near Pond #342, complaining that the drainage grate was clogged with trash and debris, and the water level was rising. City workers removed trash sediment and debris from three inflow areas and cleaned out the outflow.

After Nathan’s death, Pond # 342 was inspected again. The inspector concluded that the openings in the outflow grate are too small to allow the type of debris generally to pass through the grate, resulting in the overflow problem. Upon inspection, the outflow grate was completely clogged, resulting in the flooding of the intersection where Nathan Leg-gett drowned.

Leggett further alleges that “the condition of the pond and flooded street posed an unreasonable risk of harm to the deceased,” that the City “knew or should have known of the danger to the deceased,” that the City “failed to exercise ordinary care to protect the deceased by failing to adequately warn him of the condition and failing to make the street reasonably safe,” and that the “condition of the property” proximately caused her injuries and damages. Leggett also pleads that “[t]he flooded intersection was an unexpected and unusual danger to ordinary users of the roadways.”

In the tort claims act, the legislature has waived sovereign or governmental immunity to the extent of liability for “personal injury or death caused by a *463condition or use of ... real property if the governmental unit would, if it were a private person, be liable to the claimant under Texas law.” Tex. Civ. Prac. & Rem Code Ann. §§ 101.021(1), .025 (West 2005). Leggett pleads that her damages were proximately caused by the City’s negligence in connection with two “conditions” of real property: (1) the flooded intersection and (2) the “condition of the pond.” Leggett recognizes that these are premises defect claims. See State v. Tennison, 509 S.W.2d 560, 562-63 (Tex.1974). For negligence claims arising from premises defects, the tort claims act generally limits a governmental unit’s duty solely to that owed by a private person to a licensee on private property. Tex. Civ. Prac. & Rem Code Ann. § 101.022(a) (West Supp.2007). That is, the governmental unit owes the claimant only the duties not to injure him by wilful, wanton or gross negligence, and that it use ordinary care either to warn the licensee of or to make reasonably safe a dangerous condition of which it is aware and the licensee is not. State v. Payne, 838 S.W.2d 235, 237 (Tex.1992); Tennison, 509 S.W.2d at 562. An exception to this duty limitation applies, however, to “the duty to warn of special defects such as excavations or obstructions on highways, roads, or streets,” Tex. Civ. Prac. & Rem Code Ann. § 101.022(b), in which case the governmental unit owes the duty a private landowner owes an invitee. Payne, 838 S.W.2d at 237. That duty requires the governmental unit to use ordinary care to reduce or eliminate an “unreasonable risk of harm” created by a premises condition of which it is or reasonably should be aware. Id. Thus, the two theories differ primarily in that an “ordinary” premises-defect plaintiff must prove that, among other elements, the governmental unit had actual knowledge of the unreasonably dangerous condition and that the entrant did not, while a special-defect plaintiff must prove the governmental unit’s actual or constructive knowledge of the condition and need not disprove the entrant’s actual knowledge. Id. Leggett’s pleadings, liberally construed, purport to allege both premise-defect and special-defeet theories.

Plea to the jurisdiction

The City filed a plea to the jurisdiction, asserting that Leggett had failed to state a claim within the tort claims act’s waiver of immunity for several reasons:

• Leggett cannot assert a premises defect claim or special defect claim predicated on the alleged “dangerous condition” of the trash grate or pond because the sole “dangerous condition,” for purposes of premises liability, was the flooded intersection where Nathan drowned.

• The flooded intersection where Nathan drowned was not a special defect but an ordinary premises defect.

• The City did not know and should not have known of the flooding at that location before Nathan Leggett’s accident, negating actual or constructive knowledge of that condition.

• Nathan Leggett was aware of the flooding but “unfortunately assessed it incorrectly and chose to encounter it.”

• To the extent Leggett is complaining about any design element of the pond or grate, her claim implicates discretionary City functions for which immunity is not waived.

• The detention pond, including the drain outlet and trash grate that covered it, were not designed, constructed or built by the City of Austin, but were originally designed and built for the Tanglewood Forrest Subdivision M.U.D.

• Although the City had maintained the detention pond, there is no evidence that negligent maintenance of the pond — as opposed to heavy rainfall that simply *464overwhelmed its design — proximately caused Leggett’s injuries. In other words, even if there had been no grate covering the outflow drain, the City maintains, the huge amount of rainfall would have overwhelmed the pond and flooded the adjacent area anyway.

• Even if the City’s governmental immunity would otherwise be waived, its immunity from all of Leggett’s claims is retained under sections 101.055 and 101.062(b) of the tort claims act and Leggett has not pled a waiver under either provision.4

Jurisdictional evidence

The City attached evidence in support of its plea. Leggett filed a response, also attaching evidence. The following summarizes the evidence with regard to issues bearing on our analysis.

The detention pond

The location and basic features of the detention pond are undisputed. The Tan-glewood Forest regional detention pond (designated “Pond 342” by the City) lies in the Slaughter Creek watershed in southwest Austin. It is bordered to the north by Slaughter Lane and by residential areas on its other sides. Designed to temporarily detain storm water, Pond 342 consists mostly of a large grassy field, roughly the size of six football fields, bordered by a grassy embankment. The large field is also used for recreational purposes like soccer. On the northern side of Pond 342 are inlet pipes designed to channel drainage from areas north of Slaughter Lane. Flow channels cross the pond leading to an outlet structure located at its southwestern corner — a 66-inch outlet pipe, leading to a grass-lined spillway channel and other drainage structures designed to divert storm water ultimately into a tributary of Slaughter Creek. Covering the outlet pipe at all relevant times has been some sort of metal grate.

Pond 342 was originally constructed in 1984 for the Tanglewood Forest M.U.D. It was intended, based on the calculations of the engineer who designed it, to detain up to the amount of storm water that would result from a 100-year frequency storm, or amount of rainfall for which there is a one-percent chance that a single rainfall will meet or exceed during a given year. The detention pond remained under the M.U.D.’s control until that area was annexed by the City in 1997. Thereafter, the City had maintained the pond.

The storm of November 15, 2001

The City presented evidence that, during the afternoon of November 15, 2001, a powerful thunderstorm cell moved into the Austin area, spawning intense rainfall, hail, and tornadoes. This evidence included documents from the National Weather Service reflecting the agency’s issuance, throughout the afternoon and evening, of multiple tornado warnings and severe thunderstorm warnings advising of “very heavy rainfall and flash flooding” in Austin and Travis County. George E. Oswald, a professional engineer and hydrologist, also testified by affidavit and deposition. On November 15, 2001, Oswald was the manager of the Watershed Engineering Division of the City’s Drainage Utility Department, in which capacity he served as “the chief floodplain administrator for the City of Austin land development regulations related to floodplain management.” According to Oswald, “[o]n November 15, 2001 severe thunderstorms struck Austin, Tex*465as” and “produced flash flooding, hail, and tornadoes.” Oswald further indicated that, “owing to the severe weather,” the City’s Office of Emergency Management was activated, and that he was ordered to report to the City’s Emergency Operations Center (EOC) at City Hall sometime between 3 and 5 p.m. He explained that the EOC “served as the command center during area wide emergencies” where “[n]u-merous agencies and City of Austin department heads or decision makers gather ... to collect information and to direct responses to those emergencies.” From there, Oswald and his group “received and reviewed weather information and information from the City of Austin Early Flood Warning System as well as flooding information from other sources.”

Oswald farther explained that “[o]n November 15, 2001, the protocol concerning flooding in the City of Austin was that all such information was provided to me or my group,” to enable them to “analyze the information and appropriately respond or have others appropriately respond.” Such responses, Oswald explained, included directing the barricading of streets, sending emergency personnel, and making public service announcements “such as don’t drive into high water, road closure information, severe rain and storm activity, tornado information, etc.” He recounted that “[t]he flash flooding encompassed vast area[s] of Austin,” “Bull [C]reek, Shoal Creek and Onion [C]reek flooded,” “[a]p-proximately 600 businesses and homes were flooded,” “[rjoughly 200 families were displaced and in need of emergency shelters,” and “[m]illions of dollars worth of damage was done.” Oswald added that then-Mayor Gus Garcia declared the City a “local state of disaster” and that the City sought relief from the federal government.

The City also presented the affidavit testimony of Joe Ramos. Ramos testified that he is a licensed professional engineer who, on November 15, 2001, was the Division Manager of the Street & Bridge Division of the City of Austin. In this role, he oversaw the division’s operations, including placing barricades and other warning devices on City roadways to warn of road hazards such as trees, flooding, or other obstacles. Ramos was ordered to report to the EOC between 3 and 5 p.m. on November 15, 2001. In the EOC, Ramos explained that “information concerning road hazards such as flooded streets would have been directed to me or my group so we could respond appropriately” and, “[i]n fact, we received numerous reports of road hazards and responded to them” by placing barricades to warn of trees on the streets, flooded streets, and downed power lines and poles.

Oswald testified that “[t]he storm of November 15, 2001, was the biggest storm we have had in over twenty (20) years.” Oswald and another expert also opined that the storm had been approximately a once-in-500 year event regarding Pond 342’s watershed. While not disputing that heavy rains occurred in Austin that afternoon, Leggett points to City internal documents that had assessed the storm as only a 10- to 20-year frequency rainfall, as well as evidence indicating that the amount of rainfall in Pond 342’s watershed was not necessarily uniform or capable of precise measurement.

Flooding on Alcott & Gwendolyn

It is undisputed that during the storm, waters in Pond 342 overflowed its berm near its southeastern corner and flooded the adjacent residential area. Among the flooded area were portions of Alcott Lane and Gwendolyn Avenue. Alcott is a residential street that runs east-west roughly one block south of Pond 342. Alcott is intersected in two locations by Gwendolyn, a half-loop whose ends t-intersect Alcott *466from the north, and whose north-side residents abut Pond 342 to their rear. From its eastern intersection with Gwendolyn, Alcott continues eastward until it intersects Riddle Road, which runs south from Slaughter. Approximately six or seven house-like structures lie along Alcott’s south side between Gwendolyn and Riddle, with approximately four structures situated on the north side of the block.5

Austin police reports and City studies performed in the storm’s aftermath indicate that waters flowed from the southeastern corner of Pond 342 down an alley connecting with the northeast corner of Gwendolyn, down Gwendolyn, and into Alcott. The force of the current destroyed a privacy fence near the alley; pushed two ears, as well as a railroad tie, down Gwendolyn; scoured concrete and tore up asphalt. Approximately 25 homes on Gwendolyn and Alcott were flooded; the depth of the flood waters in some homes on Alcott south of the Alcott-Gwendolyn intersection exceeded three feet. An estimated fifteen vehicles on Alcott and Gwendolyn also suffered flood damage.

The City’s evidence included the depositions of two Austin firefighters — Lieutenant Tom Reiner and Don Bissell — who were dispatched at 5:40 p.m. on a call concerning possible flooding at 2617 Alcott. Reiner, Bissell, and a third firefighter, A1 Papi, drove their fire engine to the intersection of Alcott and Riddle, arriving at what Reiner estimated was approximately 5:45. Bissell recounted that Alcott had flooded from the west to within about 20 feet of the Riddle intersection; Reiner noted that the water began after the first driveway on Alcott. Reiner characterized Alcott as “obviously impassible.” The firefighters parked their engine and began wading down Alcott in an attempt to located 2617 Alcott. Reiner and Bissell each estimated that the waters were between eighteen inches and two feet deep.

Under the conditions at the time, Reiner recounted that the fire department was receiving an “extremely high” volume of calls and that the firefighters were operating under “Recon 3” response level or guidelines regarding how they handled calls. According to Reiner, Recon 3 denoted “we have an extreme situation where we don’t necessarily have all of the resources to match the needs of the citizens.”

The accident

Meanwhile, Nathan Leggett was driving east on Alcott in his red pickup, approaching the Gwendolyn intersection from the west. According to Trudy Leggett’s deposition testimony, Nathan had been driving home from his job at a local grocery store. Nathan and his mother, with other relatives, lived at 2600-A Alcott, located at or near the Alcott-Riddle intersection. Thus, the route by which Nathan was driving home led directly through the flooded portion of Alcott between the Gwendolyn intersection area and Riddle.

Police reports reflect the accounts of witnesses who saw a motorist, later identified as Nathan, drive into the flood waters and stall in the middle of the Alcott-Gwendolyn intersection. By some accounts in the reports, the water in the intersection had been as high as four feet. After stalling, Nathan had attempted to exit his truck on the “upstream” side and was quickly swept under the truck by the cur*467rent.6 He did not resurface. Witnesses called 911, and the police reports indicate that a call was received at 6:01 p.m. indicating that a male was stuck under his pickup in a flooded roadway in the 2600 block of Alcott.

Back at the eastern end of Alcott, according to Bissell, the firefighters had determined that 2617 Alcott was not in the block through which they were wading. In fact, 2617 Alcott was located one or two houses west of the Gwendolyn intersection. Lt. Reiner testified that, while the crew was still near the Riddle intersection, he noticed individuals standing on the northwest corner of Alcott and Gwendolyn attempting to flag down his crew. He also saw a red pickup in the Alcott-Gwendolyn intersection that he had not previously noticed. Reiner testified that he decided the crew should proceed to the location via their fire engine rather than on foot because wading the flooded street was dangerous and because the fire engine contained their equipment. Because Reiner regarded Alcott as impassible, the crew drove their engine to the Alcott-Gwendolyn intersection area via Slaughter, reaching it from the west. Reiner estimated that his crew had spent between 10-15 minutes near the Alcott-Riddle intersection,7 that it took them less than five minutes to drive to the Alcoti>-Gwendolyn intersection, but that it was reasonable to assume that they had arrived sometime after 6 p.m.8

Upon the firefighters’ arrival, witnesses informed them that a motorist had been swept under the pickup parked in the Alcott-Gwendolyn intersection and had not emerged. According to Reiner, two witnesses informed him that the motorist had disappeared approximately 15 minutes earlier. Both Reiner and Bissell indicated that the flood waters were flowing down Gwendolyn very swiftly; Bissell described them as “white rapids.” The three-person crew attempted to wade to the pickup in a “pyramid” formation. After Bissell, the front man in the formation, felt the current sweeping his feet away, the firefighters abandoned that strategy. They instead decided to reach the pickup by having Bissell try to drive the fire engine into the waters with Reiner and Papi positioned on its front bumper.

These efforts enabled Reiner and Papi to reach the pickup and locate Nathan underneath, where he had become lodged against the transmission. Reiner and Papi extracted Nathan and moved him to the front yard of a house in the northeast corner of Alcott and Gwendolyn. He was dead. The cause of death was later determined to be accidental drowning.

Reiner and Bissell indicated that, during their rescue attempt, the waters in the intersection had been between eighteen inches and two feet in depth, though Rein-er thought they might have receded slightly by the time they moved Nathan to the yard. Reiner testified that the waters had initially reached the bottom of Nathan’s pickup. By the time they extracted Nathan, Reiner observed, the water level appeared to be a couple of inches below the pickup’s bottom. Reiner also thought that the current might have subsided somewhat, pointing to the facts that the fire*468fighters had been able to pull Nathan upstream and that Bissell had been able to walk from the engine to the yard a few minutes later.

The City’s knowledge of the flooding on Alcott & Gwendolyn

The City presented evidence that it had no knowledge of the flooding in the Alcott-Gwendolyn intersection before the incident, nor any knowledge that Alcott, Gwendolyn, or Pond 342 had ever flooded. Oswald explained that from his vantage point in the EOC, “if I had received information that the intersection of Alcott and Gwendolyn Streets were flooded, I, or my group, would alert the Street & Bridge Division (Joe Ramos) of the situation so he could direct the placement of barricades and warnings.” Oswald recounted that “[o]n the evening of November 15, 2001, we were unaware that the intersection of Alcott and Gwendolyn Streets had flooded prior to plaintiffs accident.” He added that the City “maintains a list or database of flood prone areas and streets” and that “[t]he intersection of Alcott and Gwendolyn Streets, which lies in the Slaughter Creek Watershed, is not on the list or in the database. We were unaware of and had no history of these streets ever flooding.” Similarly, Oswald testified, “[pjrior to Plaintiffs accident on November 15, 2001, the City had no history of this pond, Pond 342, flooding or overflowing its banks and/or spillway.”

Joe Ramos echoed Oswald’s testimony that the Alcott-Gwendolyn intersection was not in the City’s list or database of flood-prone areas and streets and “[w]e were unaware and had no history of these streets ever flooding before.” He added that “[w]e were unaware that the intersection of Alcott and Gwendolyn Streets were flooded prior to plaintiffs, Nathan Leg-gett’s, accident.”

Cause of the flooding on Alcott & Gwendolyn

The City acknowledges that the flood waters on Alcott and Gwendolyn consisted of on-site rain and stormwater from the overflowing Pond 342. Furthermore, a February 2003 City report, introduced by both parties, attributed Pond 342’s overflow to two chief causes: (1) the magnitude of the November 15, 2001, storm exceeded the design capacity of the detention pond system; and (2) “grass clippings and leaf litter had accumulated on the pond outlet structure trash grate, reducing the design flow capacity of the pond discharge and emergency overflow system.” In fact, following the storm, Pond 342’s outflow grate was found to be 95% clogged with grass and debris. Leggett also presented evidence — including City internal documents — reflecting that Pond 342 had been inspected less frequently than City guidelines had specified and that, in April 2001, the City had received a report from a nearby resident that the grate covering Pond 342’s outlet pipe had become clogged with trash and debris, causing water to rise behind it during a rainstorm. In response, City crews had cleaned the grate of trash and debris. The City had received no further reports of blockages pri- or to Nathan’s drowning.

Leggett also presented evidence that, at some point prior to Nathan’s drowning— the evidence is disputed whether it occurred pre- or post-annexation — the original grate covering the pond’s outlet pipe had been replaced with one having individual openings of 3.5 square inches, much smaller than that specified in the original design. Leggett attached both expert opinion testimony and evidence of communications by City employees stating that the small grate openings tended to clog and provided inadequate drainage. She also introduced expert testimony that *469Pond 342⅛ design afforded inadequate “freeboard,” or difference between the pond’s water level during the 100-year frequency flood and the top of its berm.

The parties presented competing expert testimony regarding the extent to which Pond 342’s overflow and Nathan’s ultimate drowning were attributable to the state of Pond 342 or its outflow grate versus the magnitude of the storm. As noted, the parties disputed the precise amount of rainfall — the City’s experts opined that Pond 342’s watershed had received a 500-year frequency rainfall, while Leggett’s experts point to City estimates of a 10- to 20-year rainfall. Based on its calculations, the City’s experts opined that Pond 342 would have overflowed its berm even if the outflow pipe had no grate. Leggett’s experts responded with calculations purporting to demonstrate that if Pond 342’s grate had not been clogged, it either would not have overflowed or that the volume and current of any overflow would not have pinned Nathan under his truck and caused his death.

Ruling

A hearing on the City’s plea to the jurisdiction was held on March 21, 2007. The trial court’s order indicates that no additional evidence was presented at the hearing. On May 8, the trial court denied the City’s plea to the jurisdiction, finding “as a matter of law that the condition was a special defect.” The City appealed.

ANALYSIS

On appeal, the City brings forward the same grounds for dismissal that it had asserted in its plea to the jurisdiction, discussed above, and attempts to raise an additional ground — that it discharged any duty to warn of the flooded intersection by making public safety announcements during the day of the storm warning of possible street flooding and not to drive into high water. We need only address the following grounds, as they are decisive.

What is the “unreasonably dangerous condition?”

As a threshold matter, we must consider whether Leggett can allege a premises defect or special defect theory predicated upon the “unreasonably dangerous condition” of Pond 342 (i.e., what she contends was the “defective” drainage grate or the height of its berm), as opposed to the flooding itself. Leggett argues that “[t]he condition of the real property complained of ... is the defective Trash Grate” and that “the City failed to make the Trash Grate reasonably safe and to prevent flood water from rushing into the street.” The City urges that any complaint regarding the design of Pond 342 or its grate implicates its discretionary functions and is barred by governmental immunity. It adds that under the Texas Supreme Court’s precedents, the relevant “condition” for purposes of Leggett’s premises claim must be the flooding in the Alcott-Gwendolyn intersection, not any condition of the Pond that allegedly contributed to that flooding. We agree with the City.

To the extent Leggett seeks to impose liability based on the design features of Pond 342 itself, as opposed to an unreasonably dangerous condition to which those features allegedly contributed, her claims implicate City discretionary functions and are barred by governmental immunity. E.g., Texas Dep’t of Transp. v. Ramirez, 74 S.W.3d 864, 867 (Tex.2002) (design of roadways and installation of safety features are discretionary decisions that are immune from suit); Tex. Civ. Prac. & Rem.Code Ann. § 101.056(2) (West 2005). On the other hand, the City’s maintenance of the detention pond or its grate is not considered a discretion*470ary function. See County of Cameron v. Brown, 80 S.W.3d 549, 555 (Tex.2002) (contrasting maintenance of roadway lighting with discretionary decisions regarding whether and what kind of lighting to install). In either case, however, the relevant “dangerous condition” for purposes of premises liability is not the condition of the pond or grate, but the flooding in the location where Nathan drowned.

The supreme court has repeatedly held that the relevant unreasonably dangerous condition in a premise liability case is generally the condition at the time and place injury occurs, not some antecedent condition or situation that helps create a dangerous condition. E.g., Brookshire Grocery Co. v. Taylor, 222 S.W.3d 406, 408-10 (Tex.2006) (“Ordinarily, an unreasonably dangerous condition for which a premises owner may be liable is the condition at the time and place injury occurs, not some antecedent situation that produced the condition”; holding that ice on floor, and not drink dispenser that tended to produce that condition, was the unreasonably dangerous condition); CMH Homes, Inc. v. Daenen, 15 S.W.3d 97, 99-102 (Tex.2000) (platform unit was not an unreasonably dangerous condition merely because owner knew it would eventually become unstable with use); City of San Antonio v. Rodriguez, 931 S.W.2d 535, 536-37 (Tex.1996) (per curiam) (leaky roof was not itself a dangerous condition, but a cause of the dangerous condition of a wet floor below).

Rodriguez, on which the City extensively relies, illustrates this principle. At trial, Rodriguez recovered damages from the city for injuries he allegedly incurred when falling on a wet basketball court in a city-owned recreation center. The evidence was undisputed that the water on the floor came from a leak in the roof and that the city knew the roof leaked, but the city disputed whether it actually knew of the water on the floor at the time. The charge had not defined the relevant “dangerous condition,” and the court of appeals had held that the jury could have concluded that the leaky roof was the dangerous condition in the premises, so that the city’s actual knowledge was conclusively established. The supreme court held that the jury should be instructed on retrial that the allegedly dangerous condition was the water on the floor. It explained that “[t]he leaky roof was not itself a dangerous condition; it could only cause a dangerous condition.” Id. at 536-37. Consequently, “[t]he City was not required to warn of leaks in the roof or repair them; it was required only to prevent the water that leaked through the roof from causing a dangerous condition.” Id. at 536.9

*471Applying the rationale of these holdings, we agree with the City that the unreasonably dangerous condition that is the proper focus of Leggett’s premises claim is the flooding in the intersection where Nathan drowned, not the allegedly defective grate or other antecedent condition related to Pond 342 that Leggett claims contributed to the flooding. See Brookshire Grocery, 222 S.W.3d at 408-10; Rodriguez, 931 S.W.2d at 536-37. In other words, any duty of the City here could arise only from its knowledge of the flooding in the intersection, not merely its knowledge regarding the grate or other condition of the pond condition, and such duty would be to warn or make reasonably safe the flooded area rather than anything in the pond. Id. (“The City was not required to warn of leaks in the roof or repair them; it was required only to prevent the water that leaked through the roof from causing a dangerous condition.”). To the extent that Leggett purports to assert a claim predicated upon an “unreasonably dangerous condition” in the pond, as opposed to the flooding where Nathan drowned, we hold that it is barred by governmental immunity.

Although an antecedent condition or situation is not itself the unreasonably dangerous condition for purposes of premises liability, actual knowledge of an antecedent condition may, under some circumstances, help support the inference of actual or constructive knowledge of the dangerous condition. See Rodriguez, 931 S.W.2d at 537; cf. City of Corsicana v. Stewart, 249 S.W.3d 412, 414-416 (Tex.,2008) (per curiam). We consider those implications below in our discussion of the actual-knowledge element of Leggett’s premises-defect claim.

Special or “ordinary” premises defect?

Also pivotal to determining the nature of any duty potentially owed by the City is whether the flooded intersection is classified as a special defect or an ordinary premises defect. If the condition is classified as an ordinary premises defect, Leg-gett must prove, among other elements, that (1) the City had actual knowledge of the dangerous condition before Nathan Leggett’s accident, and (2) Nathan did not himself know of the condition before his accident. See Payne, 838 S.W.2d at 237. By contrast, if the condition is classified as a special defect, Leggett can satisfy the notice element by proving the City’s actual or constructive knowledge of the flooding before the accident and need not disprove Nathan’s knowledge of it. See id.

The classification of a condition as a premises defect or a special defect presents a question of duty involving statutory interpretation, analyzed on a case-by-case basis. See id. at 238. Statutory construction presents a question of law that we review de novo. State v. Shumake, 199 S.W.3d 279, 284 (Tex.2006). We seek to discern the legislature’s intent, as manifested first and foremost in the statutory text. When possible, we ascertain the legislature’s intent from the plain meaning of the words. Id. To that end, we consider statutory language in context, not in isolation. Jones v. Fowler, 969 S.W.2d 429, 432 (Tex.1998); see Tex. Gov’t Code Ann. § 311.011(a) (West 2005). These issues of duty and statutory construction are questions of law for the court to decide. Payne, 838 S.W.2d at 238. However, *472where the underlying facts surrounding the condition are disputed, those issues are for the trier of fact. Villegas v. Texas Dep’t of Transp., 120 S.W.3d 26, 32 (Tex. App.-San Antonio 2003, pet. denied).

The legislature did not define “special defect” in the tort claims act, but cited representative, non-exclusive examples “such as excavations or obstructions on highways, roads, or streets.” Tex. Civ. Prac. & Rem.Code Ann. § 101.022(b); see Harris County v. Eaton, 573 S.W.2d 177, 179 (Tex.1978). In Eaton, the supreme court observed that these examples “provide an understanding of the kinds of dangerous conditions against which the legislature intended to protect the public.” Id. “One characteristic of the class that should be considered,” it reasoned, “is the size of the dangerous condition.” Id. The condition in Eaton was a large oval-shaped hole in a two-lane asphalt road that varied between six and ten inches deep and four and nine feet wide, covering over ninety percent of the highway’s width. Such a defect, the court concluded, “had reached the proportions of a ditch across the highway” and constituted a special defect. Id.

The Eaton court rejected the assertion that a special defect could be created only by the governmental unit, holding that one could arise from weather or other natural forces:

It is our view that an excavation or obstruction need not have been created by the governmental unit itself. Nothing in the statute expresses that idea. For example, an avalanche which clogs a mountain road would be an obstruction although the governmental unit did not create it. The same may be said for an excavation. Whether created by the governmental unit, by natural forces, or by third persons, the dangerous condition on the roadway is the same.

Id. at 179-80. Whether the governmental unit created the condition, the court reasoned, would instead go to the unit’s notice of the special defect, as “the government will have actual knowledge of its existence if it created the condition.” Id. at 180.

More recently, the supreme court identified an additional characteristic of “excavations” and “obstructions” that distinguishes special defects: “all present an unexpected and unusual danger to ordinary users of roadways.” Payne, 838 S.W.2d at 238; see also Stokes v. City of San Antonio, 945 S.W.2d 324, 326 (Tex. App.-San Antonio 1997, no writ) (“A special defect is distinguished by some unusual quality outside the normal course of events”; contrasting “a longstanding, routine, or permanent condition”) (quoting Harris County v. Smoker, 934 S.W.2d 714, 718-19 (Tex.App.-Houston [1st Dist.] 1996, writ denied)). “Unexpectedness” is an objective standard. See City of Fort Worth v. Gay, 977 S.W.2d 814, 819 (Tex.App.-Fort Worth 1998, no pet.).

Subsequently, emphasizing the “unusual” or “unexpected” character of special defects it had identified in Payne, the supreme court held that ice that had formed on a bridge was not a special defect “under the circumstances of this ease.” State Dep’t of Highways & Pub. Transp. v. Kitchen, 867 S.W.2d 784, 786 (Tex.1993) (per curiam). It reasoned that “[wjhere there is precipitation accompanied by near-freezing temperatures, as in this case, an icy bridge is neither unexpected nor unusual, but rather entirely predictable. Unlike an excavation or obstruction, an icy bridge is something motorists can and should anticipate when the weather is conducive to such a condition.” Id.

The Texas Supreme Court has yet to address whether or when a flooded roadway could constitute a special defect, but some of our sister courts have. Miranda *473v. State, 591 S.W.2d 568 (Tex.Civ.App.-El Paso 1979, no writ), involved a suit arising from the death of a motorist who drowned after he drove his car into two feet of flood waters over a low-water crossing in predawn darkness and was swept away. See id. at 569. The court, relying on Eaton and the dictionary definition of “obstruction,” held that the plaintiff, by specifically pleading that the “torrential floodwaters” had “flooded and obstructed the roadway,” had alleged a special defect. Id. at 570-71. We note that Miranda pre-dated Payne and Kitchen, and that the Miranda court’s analysis was limited solely to determining whether the flood waters were an “obstruction,” never addressing whether they were “unexpected” or “unusual” under the circumstances.

In a more recent case, the Fort Worth Court of Appeals, applying the supreme court’s rationale in Kitchen, held that “a flooded low-water crossing during flash flood conditions is neither unexpected nor unusual” but “entirely predictable to the ordinary motorist traveling in such weather” and, therefore, not a special defect. Corbin v. City of Keller, 1 S.W.3d 743, 747 (Tex.App.-Fort Worth 1999, pet. denied). It reasoned that “[m]otorists can and should anticipate flooding in low lying areas when the weather is conducive to flooding.” Id. The court further observed that “[t]he open and obvious nature of the flood waters also serves to defeat the unexpected and unusual requirement for a special defect.” Id. It distinguished Miranda, observing that the court there “did not indicate what the weather conditions were on the date of the incident, or whether other factors existed that would have prevented an ordinary motorist from anticipating the two feet of flood waters at the Frio River crossing” and “decided [the case] long before, and without the benefit of, the supreme court’s decision in Kitchen.” Id. Consequently, the Corbin court viewed Miranda to stand for the proposition that “flood waters may constitute a special defect under certain circumstances when floods may be unexpected or unusual,” not “that flood waters on a roadway constitute a special defect under any and all circumstances.” Id. at 748.

Other cases have addressed the classification of lesser amounts of water present on a roadway. In a case that predated Payne and Kitchen, the Beaumont Court of Appeals held that evidence that a temporary roadway repair had created a “dam-like” effect that caused water to pool between three and seven inches deep over a roadway supported the submission of a special defect theory. State Dep’t of Highways & Pub. Transp. v. Zachary, 824 S.W.2d 813, 814-19 (Tex.App.-Beaumont 1992, writ denied). The court emphasized testimony from an eyewitness who recounted that, while it had been raining “on and off” that day, water had not ponded three inches deep elsewhere on the roadway, and that he had previously noticed ponded water at the location even when lesser amounts of rain had fallen. See id. at 815-16. The court also noted expert testimony to the effect that a motorist proceeding at the roadway’s 40 m.p.h. speed limit would not notice the depth of the ponded water due to its “sheeting” effect and would likely lose control when they hit it. Id. at 816.

More recently, the Beaumont court reached the opposite conclusion in another case involving water on a roadway — four inches of water on Interstate 10 — that had caused a motorist driving through it to hydroplane. Texas Dep’t of Transp. v. Fontenot, 151 S.W.3d 753, 761-62 (Tex. App.-Beaumont 2004, pet. denied). The court emphasized eyewitness testimony that “it had rained hard that morning,” though the precipitation had relented by the time of the accident and there was *474evidence that only one-half inch of rainfall had accumulated. See id. at 761. Relying on Kitchen and Corbin, the court held that the pooled water was not a special defect, reasoning that:

[a]lthough the rain had lightened, on a rainy day standing water on the road is neither outside the ordinary course of events nor contrary to routine expectation. The accumulated water did not constitute an obstruction, as the vehicles traveling on Interstate 10 were able to pass through the area. Although dangerous, the temporary presence of four inches of water on the highway did not present an unexpected or unusual road hazard during inclement weather.

Id. at 761-62. The court distinguished its earlier Zachary decision as involving “a ramp installed as a temporary road repair [that] prevented rainwater from flowing to the drain and created a standing pool that obscured the buckled concrete beneath.” Id. at 761. In its Corbin decision, the Fort Worth court similarly distinguished Zachary as turning on evidence that “an ordinary motorist would not have anticipated the large amount of water that had collected on the road as a result of the repair work” in light of the weather conditions. Corbin, 1 S.W.3d at 748 n. 4.

Finally, in Villegas, the San Antonio Court of Appeals held that pooled water on a state highway was not a special defect where the summary judgment evidence showed that “it had rained all day in the area on the day the accident occurred,” and “[w]ater on the road is not unexpected or unusual and something a motorist can and should anticipate when it has been raining all day.” Villegas, 120 S.W.3d at 32-33 (citing Kitchen, 867 S.W.2d 784 and Corbin, 1 S.W.3d 743). It reasoned, “[t]he water on the road was open and obvious and a condition that an ordinary motorist could have anticipated due to the weather conditions.” Id. at 33. The court distinguished Zachary as involving disputed underlying facts regarding the nature of the condition and evidence that the pooled water was not something that ordinary motorists would have anticipated. See id. at 33.10

Turning to the flooded intersection here, the size of the flood waters and physical barrier they presented — between 18 inches and four feet in depth, covering almost the entire block of Alcott between Gwendolyn and Riddle, and having a current through the Aleott-Gwendolyn intersection that moved vehicles and tore up pavement — resembled an obstacle in the roadway. See Eaton, 573 S.W.2d at 179; Miranda, 591 S.W.2d at 570-71; cf. Fontenot, 151 S.W.3d at 762 (noting that *475pooled water on 1-10 did not prevent ears from passing through it). However, guided by the supreme court’s analysis in Kitchen, we must conclude that these flood conditions were not objectively unexpected given the weather situation in Austin at the time. Although Leggett may have raised a fact dispute concerning whether the rainfall draining into Pond 342 on November 15, 2001, was precisely a once-in-500-year event or merely a once in 10- or 20-year event, there is no dispute that the rainfall across Austin was heavy and the weather conditions were severe. The heavy rainfall throughout the afternoon caused widespread flooding in Austin, accompanied by tornadoes. The conditions prompted the National Weather Service to issue multiple flash flood and tornado warnings for Austin, and the City to activate its Emergency Operations Center, which made its own public announcements warning of possible street flooding. Reflecting the obvious nature of these weather conditions and the public’s awareness of them, Ms. Leggett, during her deposition, acknowledged that she had discussed the tornadoes and possibility of flooding during a phone call with Nathan that afternoon.

Leggett emphasizes the testimony of City personnel that the City had no prior awareness or history of the Alcott-Gwendolyn intersection ever flooding, that the location was not included on the City’s database or list of flood-prone streets, and that the City had no prior awareness or history of Pond 342 ever overflowing its banks. These facts may distinguish this location from low-water crossings, see Cor-bin, 1 S.W.3d at 747, but they do not, in our view, distinguish it from the central rationale of that case and of Kitchen. A flooded street, even one that had never previously flooded, was not objectively unexpected to an ordinary motorist driving in Austin in the late afternoon and evening of November 15, 2001. See Kitchen, 867 S.W.2d at 786 (“Where there is precipitation accompanied by near-freezing temperatures, as in this case, an icy bridge is neither unexpected nor unusual.”). We also observe that the sheer size and depth of the flooded area presented an obstacle that would be open and obvious to ordinary motorists. See Corbin, 1 S.W.3d at 747. Although the record is inconsistent whether it was still daylight at the time of the incident — Lt. Reiner recalled that it was still daylight while Bissell though it had been dark — each indicated that the flood waters were readily apparent, were at least eighteen inches in depth, and had covered virtually the entire block of Alcott between Gwendolyn and Riddle.

In light of these considerations, we hold that the flooding in the Alcott-Gwendolyn intersection was not a special defect, but an ordinary premises defect.

Actual knowledge

Because the flooded Alcott-Gwendolyn intersection constituted an ordinary premises defect, Leggett can prevail only if she can prove, among other elements, that the City had actual knowledge of that condition at the time of Nathan Leggett’s accident. “Actual knowledge requires knowledge that the dangerous condition existed at the time of the accident, as opposed to constructive knowledge, which can be established by facts or inferences that a dangerous condition could develop over time.” City of Corsicana, 249 S.W.3d at 415 (per curiam) (adding that, in an premises liability case under section 101.022(a) involving a flooded low-water crossing, “the Legislature required that the City actually know that the crossing was flooded at the time of the accident”). Here, as discussed above, the City presented evidence that, at the time of Na*476than Leggett’s accident, it had no actual knowledge that the Alcott-Gwendolyn intersection had flooded.

To raise a fact issue regarding the City’s actual knowledge that the Alcott-Gwendolyn intersection had flooded, Leg-gett relies on the evidence that, in April 2001, the City had received a single citizen complaint that the drainage outflow from Pond 342 had become clogged with trash and debris and that water was rising behind it. Within a few days, a City crew cleaned the trash and debris out of the outflow grate. Leggett urges that “[a]t a minimum, [she] has raised a fact issue as to whether the City knew or should have known that the Trash Grate openings were so undersized as to render flooding not only possible but predictable.” As Leg-gett suggests, actual knowledge of an unreasonably dangerous condition can sometimes be proven through evidence of the premises owner’s knowledge of an antecedent condition that, when coupled with other facts, supports the reasonable inference that the owner actually knew of the ultimate dangerous condition. See Rodriguez, 931 S.W.2d at 537. In Rodriguez, for example, the supreme court held there was legally sufficient evidence that the City employee in charge of the basketball court had actual knowledge — based on his contemporaneous knowledge of the roof leaks and the fact it was raining — that there would be water on the court’s floor. See id. at 537.

However, such evidence can support an inference of actual knowledge “only when it ‘either directly or by reasonable inference’ supports that conclusion.” See City of Corsicana, 249 S.W.3d at 415. The mere fact that an unreasonably dangerous condition may tend to develop over time does not support an inference of actual knowledge that the dangerous condition existed at the time of the accident. City of Dallas v. Thompson, 210 S.W.3d 601, 602-03 (Tex.2006) (“[T]he fact that materials deteriorate over time and may become dangerous does not itself create a dangerous condition, and the actual knowledge required for liability is of the dangerous condition at the time of the accident, not merely of the possibility that a dangerous condition can develop over time.”); see also State v. Gonzalez, 82 S.W.3d 322, 330 (Tex.2002) (evidence that State knew that stop signs had been repeatedly vandalized “does not indicate, either directly or by reasonable inference, that [it] actually knew the signs were down before the accident occurred”). Under these principles, Leggett’s evidence that the outflow from Pond 342 had become clogged on one prior occasion seven months before the drowning (and was cleaned out after it was reported) falls short of supporting the inference that the City actually knew that the grate had become clogged again at the time of Nathan Leggett’s accident. More importantly, this evidence is insufficient to support an inference that the City actually knew of the ultimate dangerous condition that is the proper focus of Leggett’s premises-defect claim — the flooded intersection. See Marathon Corp. v. Pitzner, 106 S.W.3d 724, 728 (Tex.2003) (“[A]n inference stacked only on other inferences is not legally sufficient evidence.”).

We agree with the City that it has met its burden of conclusively negating its actual knowledge that, before Nathan Leg-gett’s accident, the Alcott-Gwendolyn intersection had flooded. And this holding, along with the others above, are decisive of whether Leggett has affirmatively invoked the trial court’s subject-matter jurisdiction by pleading facts within the limited waiver of the tort claims act; we need not reach the City’s other grounds for dismissal. Furthermore, our analysis demonstrates that these jurisdictional defects are not *477curable by re-pleading. See Texas A & M Univ. Sys. v. Koseoglu, 23B S.W.3d 835, 840 (Tex.2007) (instructing that when pleading defects cannot be cured, plaintiffs suit must be dismissed); see also City of San Antonio v. Hartman, 201 S.W.3d 667, 672-73 (Tex.2006) (section 101.055(2) of tort claims act shielded city from suit arising from motorists’ drowning in flooded roadway; city’s decision to prioritize responding to flooding at other locations “is precisely the kind of conduct this statute removes from judicial review”). Accordingly, we must dismiss Leggett’s suit for want of subject-matter jurisdiction.

CONCLUSION

We reverse the trial court’s order and render judgment dismissing Leggett’s suit for want of subject-matter jurisdiction.

Concurring Opinion by Justice PATTERSON.