delivered the opinion of the Court.

In this case, we must decide whether the evidence is legally sufficient to support the[*549] trial court’s determination that construction and operation of the Tarrant Regional Water District’s Richland-Chambers Reservoir caused a significant change in flooding characteristics that damaged the Gragg Ranch. If it is, we must decide (1) whether a taking resulted, (2) whether the trial court erred in trying the condemnation and compensation issues together, and (3) whether the compensation issue was properly submitted or the jury’s award properly supported. We hold that the evidence presented is legally sufficient to support the trial court’s findings that the reservoir caused recurrent destructive changes in flooding characteristics that directly impacted the Gragg property such that it was no longer usable for its intended purpose and was taken. We also hold that the trial court did not abuse its discretion in refusing to order separate trials on the questions of inverse condemnation and compensation. Finally, we hold that the trial court’s submission of the compensation issue is not reversible and the jury’s compensation award is supported by legally sufficient evidence. Accordingly, we affirm the court of appeals’ judgment. 43 S.W.3d 609.

I. Background

In 1990, the Tarrant Regional Water District began releasing water for the first time from the Richland-Chambers Reservoir on the Trinity River about eight river miles ^[1] upstream from the 12,516-acre Gragg Ranch. The Gragg family has owned the Ranch since 1949. The Gragg Ranch is one of the largest in East Texas, lying partly in Anderson County and partly in Freestone County. The Trinity River borders the Ranch in some places and bisects the property around the Ranch’s mid-point. In all, the Ranch has seventeen miles of river frontage. The Gragg property consists of 1,722 upland acres and 10,794 acres of bottomland within the Trinity River’s flood plain where cattle normally graze.

A number of factors have made the Ranch particularly suitable for cattle ranching. Regular flooding in the past contributed to the bottomlands’ exceptional fertility. A number of elevated levee roads traversing the bottomlands enabled large trucks to timely evacuate cattle in slowly rising flood waters. Because the Ranch’s acreage is contiguous and the bot-tomlands’ fertility produced a steady natural food supply, a very small staff could conduct a large-scale, highly profitable cattle-ranching operation. In 1989, O.L. Gragg leased the property to the Schwert-ner-Priest partnership. The partnership further increased the Ranch’s profitability by instituting the Schwertner Select program in which recently weaned calves, which are generally highly sensitive to stress and disease, are pre-conditioned for thirty to forty-five days before being transported to feedlots or stocker operations. The property’s open terrain, contiguity, and fertility, made it uniquely suited to this pre-conditioning program.

The Tarrant Regional Water District is a water control and improvement district created under article XVI, section 59, of the Texas Constitution. One of the District’s functions is to provide for the control, storage, preservation, distribution, conservation, and reclamation of water, including flood water. Tex. Water Code § 51.121(b)(1), (3). It may also control, abate, or change any shortage or harmful excess of water. Id § 51.121(b)(5). The District is authorized to acquire easements[*550] considered necessary, incident, or helpful to accomplish its purpose. Id. § 51.121(c). The District is a political subdivision of the State generally entitled to sovereign immunity, although it is subject to the strictures of article I, section 17, of the Texas Constitution, the so-called “takings” clause.

In 1987, the District completed construction of the 1.2 million acre-foot Richland-Chambers Reservoir, which it built to supply water to Tarrant County and surrounding areas. The 44,000-acre reservoir impounds the waters of Richland and Chambers Creeks, two tributaries of the Trinity River. The creeks’ watersheds comprise about sixty percent of an eighty-mile stretch of the Trinity’s watershed between the Trinidad Gauge, located near Trinidad Lake (between Corsicana and Athens), and the Oakwood Gauge (southwest of Palestine). The dam that holds the creeks’ waters back is located about a mile from the Trinity River. A narrow, steeply banked, straight discharge channel connects the dam and the river and blocks a large portion of the Trinity’s flood plain. The reservoir was not constructed to control floods but to supply water. Consistent with its intended function, the District keeps the reservoir as full as possible at a level only two feet below the overflow point.

In March 1990, extremely heavy rains caused extensive flooding throughout the Trinity Basin, and the District released water through the reservoir’s floodgates for the first time. For the first time in its history, the Gragg Ranch suffered extensive flood damage. That flooding breached levee roads in several places, and gouged large sections of land out of the Ranch’s bottomlands. O.L. Gragg and the Schwertner-Priest Partnership and its partners sued the District, alleging that its construction and operation of the reservoir had inversely condemned their property in violation of Article I, Section 17, of the Texas Constitution. ^[2] The District denied liability, but asserted a counterclaim asking the trial court to award it fee simple title, or alternatively, a permanent and perpetual flowage easement over any property the court might find it had inversely condemned.

The case was tried in 1998, by which time the Ranch had experienced a large number of floods similar in severity to the one in 1990 that prompted this suit. After a fifteen-day trial, the court held that the District had inversely condemned a flood easement on the Ranch, and submitted the case to the jury to determine just compensation. The jury found the difference between the market value of Gragg’s fee simple interest immediately before and immediately after condemnation to be $10,214,122, and the before-and-after value of the Schwertner-Priest Partnership’s leasehold estate to be $4,268,547. The trial court rendered judgment on that verdict and also awarded the District a permanent and perpetual flowage easement over the property. The court made numerous findings of fact and conclusions of law supporting its inverse condemnation holding. The court of appeals affirmed the trial court’s judgment. 43 S.W.3d 609. We granted review to determine whether the District’s construction and operation of the reservoir resulted in a taking of the Gragg property, and other related issues.

II. Inverse Condemnation

The District argues that Gragg failed to establish a taking for two reasons. First,[*551] it claims that Gragg failed to adduce any competent or reliable evidence that the reservoir’s construction and operation caused the flood damage that the Ranch experienced. Second, if Gragg established causation, the District claims that its actions were merely negligent and do not, as a matter of law, constitute a taking. See City of Tyler v. Likes, 962 S.W.2d 489, 505 (Tex.1997). We address each of these contentions in turn.

A. Causation

The District begins its attack on the causation evidence by challenging the reliability of Gragg’s experts’ opinions, specifically those of hydrologists Dwayne Stubblefield and Gary Pettit. The District contends that Gragg failed to show that the X-FOR computer model upon which these experts relied meets the standard for reliability we established in Merrell Dow Pharmaceuticals, Inc. v. Havner, 953 S.W.2d 706, 726-28 (Tex.1997), E.I. duPont de Nemours & Co., Inc. v. Robinson, 928 S.W.2d 549, 557-58 (Tex.1995), and related authorities. Gragg responds that (1) the District failed to properly preserve its objections, (2) the X-FOR model was merely cumulative of other evidence that was admitted without objection, (3) the X-FOR model was not central to Stubble-field’s and Pettit’s opinions on causation, which were based primarily on other evidence that has not been challenged as unreliable, (4) the X-FOR model was nonetheless shown to be rehable, and (5) causation was established by the District’s own records and witnesses irrespective of Stubblefield’s and Pettit’s testimony. We begin with the latter point because, if there is causation evidence from other sources that supports the trial court’s findings, the District’s reliability challenge is immaterial.

It is undisputed that before the reservoir’s construction, the Trinity River had regularly flooded the Ranch’s bottomlands. Gragg presented evidence, though, that the flooding that had occurred before the reservoir was built was not significantly destructive and was, in fact, a major factor in the cattle-ranching operation’s success. After the reservoir was constructed and operations began, Gragg claims, the nature of the flooding significantly changed and the Ranch suffered extensive damage that it had never before experienced. A number of the trial court’s fact findings describe the nature of that change:

1. ... the flooding on the Gragg Ranch has been far worse than would have occurred under natural conditions, and the flooding will continue to be far worse in the future.

2. ... the flood waters from the lake have arrived at the Gragg Ranch quicker and with less warning than would have occurred under natural conditions, and will continue to do so.

3. ... the flood waters from the lake have arrived at the Gragg Ranch with more force and velocity than would have occurred under natural conditions, and will continue to do so.

4. ... the floods have been deeper on the Gragg Ranch than would have occurred under natural conditions, and the deeper floods will continue.

5. ... the Gragg Ranch has been flooded for longer periods of time than would have occurred under natural conditions, and the longer floods will continue to be so.

The trial court found each of these exacerbated flood characteristics to be “[a]s a direct result of the construction and operation of the Richland-Chambers Reservoir.”

[*552] Gragg claims these post-reservoir flood effects made it economically infeasible to continue using the Ranch as a high-intensity cattle-ranching operation. Quicker inundation with shorter warning time and the destruction of levee roads made it impossible to safely evacuate herds from the bottomlands in floods. And the deeper and longer-lasting floods killed crops growing in the bottomlands that would normally have survived, reducing the land’s productivity. The District takes no issue with Gragg’s claim that the Ranch experienced these destructive flood effects; rather, the District contends there was no evidence to support the trial court’s findings that the reservoir’s construction and operation, as opposed to heavy rains, caused them.

In determining whether a finding is supported by legally sufficient evidence, we consider the evidence in the most favorable light, and indulge every inference in its favor. Formosa Plastics Corp. USA v. Presidio Eng’rs. & Contractors, Inc., 960 S.W.2d 41, 48 (Tex.1998). If the record contains more than a scintilla of evidence that supports the finding, we must sustain it. BMC Software Belgium, N.V. v. Marchand, 88 S.W.3d 789, 795 (Tex.2002) (citing Holt Atherton Indus., Inc. v. Heine, 835 S.W.2d 80, 84 (Tex.1992)). “More than a scintilla of evidence exists where the evidence supporting the finding, as a whole, ‘rises to a level that would enable reasonable and fair-minded people to differ in their conclusions.’ ” Burroughs Wellcome Co. v. Crye, 907 S.W.2d 497, 499 (Tex.1995) (quoting Transp. Ins. Co. v. Monel, 879 S.W.2d 10, 25 (Tex.1994)). Whether particular facts are sufficient to establish a taking presents a question of law that we review de novo, but we rely on the factfinder to resolve disputed facts underlying that determination. See Mayhew v. Town of Sunnyvale, 964 S.W.2d 922, 933 (Tex.1998).

Gragg produced evidence that the District released volumes of water through its flood gates that greatly exceeded what natural flows would have been in a manner that would impact the Gragg Ranch. It is undisputed that the Ranch begins to flood when the Trinity River’s flow reaches 13,-000 cubic feet per second (cfs). The dam’s operator, Ronnie Byers, acknowledged that on at least one occasion the District released more than 23,000 cfs in excess of what natural peak flows would have been from Richland and Chambers Creeks. The District’s gate release records show numerous additional releases in excess of natural conditions. In fact, the District’s records show hundreds of releases in an amount sufficient to cause flooding at the Gragg Ranch even if there were no other water in the Trinity River. John Rutledge, one of the District’s hydrological experts, acknowledged that the modeling his firm performed showed higher flooding than would have occurred under natural conditions at the Ranch in ten out of sixteen floods. And a 1991 internal District memorandum suggests that the District had recognized that changes in its gate-release procedures might be warranted to minimize flooding in the river and to “determine when our policies can be modified to permit higher releases earlier to avoid adding to flood levels in the Trinity.” The record contains more than a scintilla of evidence to support the trial court’s finding that the reservoir’s construction and operation intensified flooding at the Ranch.

There is also evidence to support the trial court’s findings that the reservoir caused quicker and more forceful flooding at the Ranch. There is evidence, for example, that the reservoir’s configuration creates a kind of bathtub effect in the river’s watershed so that rain falling at a point[*553] near the reservoir’s headwaters causes flooding at the dam site more quickly than it would if the creeks’ waters were not confined by the dam. The resulting force of the District’s releases causes flood waters to rise more rapidly and with less warning. The District contends that the reservoir’s stilling basin minimizes the force of these releases by significantly reducing the speed at which the water moves. But Rutledge acknowledged that water released from the reservoir’s floodgates moving at a high rate of speed into the stilling basin would cause a translatory wave, similar to a surge. While the water released through the gates might not itself reach the Ranch at the same speed it is released, Rutledge acknowledged that its impact would be transmitted almost instantaneously to the Trinity when the Trinity is out of its banks. Other District witnesses testified that the District attempts to release only when the Trinity is out of its banks. And another District witness confirmed that the characteristics of releases at the dam correlated with impacts at the Gragg Ranch. The District also maintains that water-release surges would be dissipated by a heavily wooded wildlife refuge adjacent to the reservoir and argues that, if the surge effects were as severe as Gragg claims, the refuge would have incurred greater damage than the Gragg Ranch. But this ignores evidence that the discharge channel directs gate releases almost directly into the river and evidence that the high banks of the discharge channel and spoils piles around it would have minimized any impact on the wooded area nearby.

Finally, several eyewitnesses testified that flood waters at the Ranch moved much more swiftly when the floodgates at the reservoir were open. More than one witness testified that water flowing from the discharge channel into the Trinity entered the river with such velocity that the flow of some of the river’s waters was reversed. Eugene Schwertner testified, for example, that he had flown over the area during several flood events. When the dam’s floodgates were open, he said, the water “was coming out so fast it had to go somewhere, so it was going backwards.” But when he returned the next day and only one or two gates were open, “the water would slow down, it wasn’t swift. Then, when they closed the gates, it was just like overnight; there wouldn’t be the fast movement. So you could see — I could see with my own eyes the gates opening and closing and what was happening.” Although the eyewitness evidence did not purport to, nor could it, explain how the reservoir’s construction caused what happened at the Ranch, it did verify water-flow characteristics that occurred when the floodgates were opened.

The District complains that Gragg failed to rule out the extremely heavy rainfalls of the early 1990s as the cause of damage to the Ranch. Undoubtedly, the Ranch would not have flooded without excessive rain. And damage to the Ranch caused by extremely heavy rain, even if reservoir releases contributed to the volume, could not in itself support a taking. Rather, the issue is significantly changed flooding characteristics that occurred despite similar circumstances so that it can be inferred that the reservoir was to blame. Here, the District’s own modeling showed that the number and duration of floods at the Ranch in the 1990s, after the reservoir’s construction, were higher than in the 1940s, a period of comparable rainfalls. Similarly, Troy Lovell, another of the District’s expert witnesses, also testified that flood-gauge data showed flooding in the Gragg Ranch’s vicinity in 1993 and 1995 lasting twice as long as floods in 1952, 1959, 1962, and 1965 that occurred in periods of similar rainfall. Although the Dis[*554] trict’s experts attributed the increase in flooding during the 1990s, as opposed to the 1940s, to the construction of various flood-control projects in the interim, those projects were in place before 1959.

It was not enough, of course, for Gragg to trace the damaging floods back to releases from the dam. Gragg was required to prove that the same damaging floods would not have occurred under the same heavy rainfall conditions had the dam not been constructed. Whether the reservoir’s construction and operation caused the exacerbated flood effects that the Ranch experienced was a sharply disputed fact issue. We must therefore defer to the trial court’s resolution of that issue as long as there is evidence to support it. Although we express no opinion on whether Gragg’s causation evidence “was far beyond a preponderance of the evidence ... almost to the level of ‘beyond a reasonable doubt,’ ” as the trial court concluded, the record does contain legally sufficient evidence to support the trial court’s causation findings. We next consider whether those findings support the legal conclusion that a constitutional taking occurred.

B. Legal Standard

Article I, Section 17, of the Texas Constitution provides:

No person’s property shall be taken, damaged or destroyed for or applied to public use without adequate compensation being made, unless by the consent of such person.

Tex. Const. art. I, § 17. At the heart of the takings clause lies the premise that the government should not “forc[e] some people alone to bear public burdens which, in all fairness and justice, should be borne by the public as a whole.” Steele v. City of Houston, 603 S.W.2d 786, 789 (Tex.1980) (quoting Armstrong v. United States, 364 U.S. 40, 49, 80 S.Ct. 1563, 4 L.Ed.2d 1554 (1960); Y.M.CA. v. United States, 395 U.S. 85, 89, 89 S.Ct. 1511, 23 L.Ed.2d 117 (1969)). A taking under this provision may be physical or regulatory. Mayhew, 964 S.W.2d at 933. A physical taking may occur when the government physically appropriates or invades private property, or unreasonably interferes with the landowner’s right to use and enjoy it. Westgate, Ltd. v. State, 843 S.W.2d 448, 452 (Tex.1992). When the government takes private property without first paying for it, the owner may recover damages for inverse condemnation. Id. But mere negligence that eventually contributes to property damage does not amount to a taking. Likes, 962 S.W.2d at 505.

The District claims that, to the extent the evidence supports the trial court’s conclusion that the reservoir’s construction and operation caused the harm that the Gragg Ranch experienced, it was attributable to mere negligence on the District’s part and not to intentional conduct, which the constitutional taking standard requires. Over the years, we have articulated the standard for a compensable physical taking in various ways. The cases reflect our efforts to account for several concerns in drawing the line between mere negligence and an unconstitutional taking. See, e.g., City of Houston v. Renault, Inc., 431 S.W.2d 322, 324-25 (Tex.1968). For one, we strive to avoid what would be an anomalous result if the State, an entity otherwise generally entitled to immunity for negligence, were subject to liability for something less than intentional behavior. More importantly, though, we seek to ensure that the public does not bear the burden of paying for property damage for which it received no benefit. See, e.g., Tex. Highway Dep’t v. Weber, 147 Tex. 628, 219 S.W.2d 70, 71-72 (1949). As we have noted, our Constitution provides for compensation only if property is damaged[*555] or appropriated “ ‘for or applied to public use.’ ” Steele, 603 S.W.2d at 792 (citing Davis v. City of Lubbock, 160 Tex. 38, 326 S.W.2d 699, 702-709 (1959)). “That is the factor which distinguishes a negligence action from one under the constitution for destruction.” Id. Accordingly, we have sought objective indicia of intent in particular contexts to determine whether property has been taken or damaged in furtherance of the public interest.

In City of Dallas v. Jennings, 142 S.W.3d 310, 314(Tex.2004), which we also decide today, we hold that the requisite intent is present when a governmental entity knows that a specific act is causing identifiable harm or knows that the harm is substantially certain to result. In the case of flood-water impacts, recurrence is a probative factor in determining the extent of the taking and whether it is necessarily incident to authorized government activity, and therefore substantially certain to occur. See Brazos River Auth. v. City of Graham, 163 Tex. 167, 354 S.W.2d 99, 107 (1961). While nonrecurrent flooding may cause damage, a single flood event does not generally rise to the level of a taking. See id. at 108. The recurrence requirement assures that the government is not held liable for taking property when a project’s adverse impacts, and by implication its benefit to the public, are too temporal or speculative to warrant compensation. See id. This is similar to the standard the federal courts have applied in determining whether the government’s actions have taken property affected by flooding, see, e.g., Turner v. United States, 901 F.2d 1093, 1095 (Fed.Cir.1990); Hendricks v. United States, 14 Cl.Ct. 143, 148 (Cl.Ct.1987); Anchor Estates, Inc. v. United States, 9 Cl.Ct. 618, 620-21 (Cl.Ct.1986); Singleton v. United States, 6 Cl.Ct. 156, 162-63 (Cl.Ct.1984), and it is the standard we apply in the present case.

The District claims the record establishes no more than mere negligence because if the reservoir is operated as intended, it would not add more water downstream than would naturally pass through the reservoir. Consequently, the District argues, “[o]nly if the District were negligent in operating the reservoir would it add more water than would have passed downstream naturally.” This argument, though, misapprehends the nature of Gragg’s complaint and the basis for the trial court’s judgment. All parties agree that the Ranch was subject to flooding before the Reservoir’s construction. But the trial court found that the property was rendered useless for its intended purpose because the reservoir’s construction and operation changed the character of that flooding — the water “arriv[ed] sooner, flowfed] faster, and [was] more forceful, deeper, and longer-lasting.” Although gate-release operations contributed to these effects, there was evidence that the reservoir’s physical characteristics, which we have described, were also significant and inevitably altered the characteristics of floods at the Ranch. While Gragg did introduce evidence of the District’s careless operations, that evidence was aimed largely at undermining the credibility of the District’s efforts to model the reservoir’s flood impacts at particular points on the property. And although there was evidence that the District’s gate-release procedures might mimic natural conditions over time, Gragg introduced evidence showing that the District’s releases actually resulted in unnatural surges of water.

We hold that the evidence in this case supports the trial court’s findings that the extensive damage the Gragg Ranch experienced was the inevitable result of the reservoir’s construction and of its operation as intended. In reaching that decision, we acknowledge the concerns raised by sever[*556] al amici curiae who contend that holding the District accountable in these circumstances will have significant repercussions on water suppliers in the state. ^[3] We note, however, that there was evidence that the design of the Richland-Chambers Reservoir is somewhat unique in a number of respects, including its comparatively limited excess storage capacity of about eight percent. According to various amici (Canadian River Municipal Water Authority, Lower Colorado River Authority, Lavaca-Navidad River Authority), water-supply reservoirs normally have excess capacity ranging from twenty-five to one hundred percent. Moreover, as we noted more than forty years ago,

[governmental agencies and authorities are necessities. They are capable of rendering great and beneficent public services. But any appeal to the tradition of our laws which omits a decent regard for private property rights is both inaccurate and distorted. It is because of this regard that our governmental agencies and authorities in acquiring properties for their public purposes are generally required to proceed under the power of eminent domain rather than under the police power. Such a policy has not resulted in a destruction of flood control and improvement agencies in the past and there is no reason to apprehend that the continuation of such policy will prove overly costly or inimical to the American way of life in the future.

Brazos River Auth., 354 S.W.2d at 105.

III. Bifurcation

The District argues that the trial court abused its discretion by declining to bifurcate the proceedings. According to the District, the trial court should have first conducted a bench trial on the takings issue, and then held a separate trial, if necessary, to allow a jury to assess reasonable compensation.

Texas Rule of Civil Procedure 174(b) allows a trial court to order a separate trial on any issue in the interest of convenience or to avoid prejudice. We review a trial court’s ruling on a motion for separate trial for an abuse of discretion. See Allison v. Ark. La. Gas Co., 624 S.W.2d 566, 568 (Tex.1981) (per curiam). The District complains that it was prejudiced in a number of respects by the trial court’s refusal to order separate trials. First, the District asserts that the failure to bifurcate left it confused about several issues, including “whether the court would find the District responsible at all,” whether the court would find a temporary or permanent taking which would in turn call for different damage measures, whether a fee or an easement was taken, and how much of the Ranch was affected. The District argues that the trial court’s failure to order separate trials was inherently unfair because it forced the District to present proof of damages while simultaneously contesting liability. We disagree.

In nearly every case a defendant must proceed with some amount of uncertainty of the type the District describes. Defendants typically proceed generally without knowing whether they will be found liable or on what, if any, theory. And it is not at all unusual for a plaintiff to present more than one damage measure during trial.[*557] For example, a plaintiff may allege breach of contract and quantum meruit, causes of action with different elements of proof and different damage measures. See, e.g., Murray v. Crest Constr., Inc., 900 S.W.2d 342 (Tex.1995)(per curiam). The District has articulated no reason why admitting damage evidence before liability is determined is any more prejudicial in the condemnation context than in any other.

On the other hand, the record supports the trial court’s findings that separate trials would have resulted in considerable and unnecessary evidentiary repetition. For example, the damage questions that were submitted instructed jurors “to consider only the differences in value caused by the construction and operation of Rich-land-Chamber Reservoir.” Whether and to what extent the Ranch was damaged by the reservoir’s construction and operation were also issues central to the trial court’s taking determination. Accordingly, it is likely that many, if not most, of the same witnesses would have been called to testify in both the liability and compensation trials had the trial court bifurcated the proceedings. As the trial court found, “[t]here were several weeks of common questions of law and of fact involved in the matters that would have been considered in the first phase and the second phase of a bifurcated trial.” And, as the court of appeals noted, this case had been pending in the trial court for almost seven years and had been through extensive discovery and numerous pre-trial proceedings by the time it was tried. Under these circumstances, we cannot say that the trial court abused its discretion in refusing to bifurcate the proceedings.

The District claims, though, that we mandated bifurcation in State v. Wood Oil Distributing, Inc., 751 S.W.2d 863, 865 (Tex.1988), and State v. Heal, 917 S.W.2d 6, 9 (Tex.1996). But those cases do not

establish such a bright-line rule. In Wood Oil, we considered a condemnee’s claim that the trial court erred in excluding evidence of impaired access to its property. We noted that “whether there has been a [compensable] material and substantial impairment of access or whether there exists merely the [non-compensable] issue of circuity of travel is a question of law, not of fact.” Wood Oil, 751 S.W.2d at 865. Stating that it was incumbent upon the trial court to make this determination before trial so that evidence could be admitted accordingly, we concluded that evidence concerning inconvenience damages resulting from circuity of travel was improperly admitted because it was an element that was non-compensable as a matter of law. Id. In Heal, we considered whether a landowner had established a right to compensation for impaired access, or whether the landowner had presented evidence akin to circuity. 917 S.W.2d at 9-11. Relying on Wood Oil, we concluded that the landowner had not established a material and substantial impairment. Id. at 9. Although it is often preferable or even necessary to bifurcate the trial of an inverse condemnation case so that takings issues are tried to the bench before damage issues are submitted to the jury, our statements in neither Wood Oil nor Heal support the District’s argument that separate trials must be conducted on taking and compensation issues in every case. Accordingly, the trial court did not abuse its discretion in denying the District’s bifurcation motion.

IV. Damages Issues

The District raises several issues regarding the damages the trial court awarded. The District contends that (1) the trial court erroneously submitted the case to the jury as a permanent taking and instructed the jury improperly about the easement the District was awarded, (2) Gragg presented no evidence of the[*558] Ranch’s value with and without the easement or of the bottomlands’ before-and-after value, and (3) Gragg adduced no evidence upon which the jury could apportion damages between those caused by the reservoir and those attributable to other causes. We address each of these arguments in turn.

A. Permanent v. Temporary Taking

The District argues that Gragg sustained only temporary injuries because the Ranch has not been “constantly and continuously flooded or injured by the reservoir.” According to the District, the proper measure of damages for a temporary taking would have been the cost of repair, and Gragg submitted no evidence of that cost. Thus, the District contends, the trial court erred in instructing the jury that the District had permanently taken a flood easement and in awarding damages based upon a permanent taking.

In Brazos River Authority, we considered the difference between a taking by flooding and temporary damage. 354 S.W.2d at 104-09. There, the Brazos River Authority’s construction of a dam caused siltation in the Brazos River that eventually caused flooding in a sewage-disposal plant and a water-treatment plant that the City of Graham owned. Because the evidence established that the dam’s construction would subject the sewage-disposal plant to repeated flooding that rendered its operations impossible, we held that the river authority had taken the property. Id. at 106. We upheld a compensation award based upon the difference in the plant’s value before and after the dam’s construction and operation. Id. at 101, 111. In contrast, the water-treatment plant had flooded only once, although it was possible that it would flood more in the future as siltation increased. Id. at 108. We thus concluded that the City was not entitled to the plant’s diminished value, but only to damages for injuries that resulted from the specific flood. Id. at 108, 111.

In this case, Gragg presented evidence of numerous instances in which the reservoir’s construction and operation resulted in significant, damaging changes in flooding characteristics. By the time this case was tried, the Ranch had been subject to damaging flooding on at least ten separate occasions. The trial court found that the reservoir’s construction and operation would “inevitably” cause these exacerbated flood characteristics, and that the damage caused was ongoing and would continue in the future. The record supports the trial court’s determination that the reservoir’s construction and operation caused a permanent taking. Accordingly, the trial court did not err in awarding permanent-taking compensation or in instructing the jury that the District was entitled to a permanent right to flood or overflow the Ranch’s bottomlands.

B. Before-and-After Property Value

The District further contends that the jury’s compensation award cannot stand because Gragg introduced no evidence of the property’s value with and without the flood easement. But Gragg presented the testimony of two appraisers regarding the property’s value both before and after it experienced the damage that the trial court found the reservoir’s construction and operation caused. While the appraisers did not use the word “easement,” the District inversely condemned the easement by causing that very damage. That the appraisers did not use the term “easement” does not mean that no evidence supports the compensation the jury awarded.

[*559] C. Apportionment

The District argues that no evidence was presented that would allow a jury to apportion damages between the bottomlands and the remainder of the property. In essence, the District argues that because the trial court awarded it an inundation easement across the Ranch’s bottomlands, the property was effectively severed and the trial court should have awarded damages based upon the value of the severed portion (the bottomlands) and the after-severance value of the remaining property. But the District did not raise this objection in the trial court. In fact, the District told the court that the jury should be asked, “What do you find was the difference in market value, if any, of the entire fee simple interest in the ranch (the Gragg lease fee and the Schwertner-Priest leasehold estate) immediately before and after March 7, 1990 (the date of the condemnation)?” Accordingly, it cannot now complain that the jury was not asked to apportion damages. Finally, the District argues that the damages should have been apportioned between those that its operations caused and those caused by the unusually heavy rainfalls that occurred in the spring of 1990. As we have noted, however, there is more than a scintilla of evidence to support the trial court’s conclusion that the exacerbated flood effects that damaged the Gragg Ranch were caused by the reservoir’s construction and operation. And in awarding compensation, the jury was instructed to consider only the difference in the property’s value that the reservoir’s construction and operation caused. Accordingly, the trial court’s judgment is not reversible on this basis.

V. Conclusion

We hold that the evidence is legally sufficient to support the trial court’s determination that the construction and operation of the Richland-Chambers Reservoir necessarily caused recurrent destructive changes in flood characteristics at the Gragg Ranch that rendered the property unusable for its intended purpose and resulted in a taking. ' We further hold that the trial court did not abuse its discretion in refusing to bifurcate the condemnation and compensation proceedings and committed no reversible error in submitting the case to the jury as it did. Accordingly, we affirm the court of appeals’ judgment.