Opinion

Article I, section 19 of the California Constitution (section 19) provides that when a public entity takes or damages property, it must pay the owner just compensation. (See, e.g., Locklin v. City of Lafayette (1994) 7 Cal.4th 327, 362 [ 27 Cal.Rptr.2d 613 , 867 P.2d 724 ] (Locklin).) In Belair v. Riverside County Flood Control Dist. (1988) 47 Cal.3d 550 [ 253 Cal.Rptr. *436 693 , 764 P.2d 1070 ] (Belair), we held that when a public entity’s design, construction, or maintenance of a flood control project poses an unreasonable risk of harm to property historically subject to flooding and causes substantial damage to it, the property owners may recover damages for inverse condemnation under section 19. ( Belair, supra, 47 Cal.3d at pp. 564-567.) Belair concluded that, if the public entity acted unreasonably, compensation “constitutes no more than a reimbursement to the damaged property owners of their contribution of more than their [proportionate share to the public undertaking].” (Id. at p. 566.) The question here is whether, in the narrow and unique context of flood control litigation, Belair's rule, as endorsed and refined by Locklin, supra, 7 Cal.4th 327 , should apply when the public entity’s efforts to divert water from a potentially dangerous natural course fail and cause property damage during a severe tropical storm. The Court of Appeal concluded the rule should apply.

We agree with the Court of Appeal and conclude that Belair’s reasonableness standard, as endorsed and refined by Locklin'% balancing principles (discussed below), should apply to flood control cases, including this one, where the failure of publicly managed flood control facilities causes property damage. 1 The Coachella Valley Water District (the District) here maintained dikes and levees that sought to channel water safely away from a potentially dangerous natural flow. When a water project fails, as this one did, causing flood damage, the issue is whether the system’s design, construction, and maintenance were reasonable. (Belair, supra, 47 Cal.3d at p. 565 .) If the public entity’s conduct is unreasonable and a substantial cause of damage, the entity “is liable only for the proportionate amount of damage caused by its actions.” (Locklin, supra, 7 Cal.4th at p. 368 .) This inverse condemnation rule invokes constitutional balancing principles and is not governed by tort concepts of fault or negligence. It requires a balancing of the public need for flood control against the gravity of harm caused by unnecessary damage to private property. (See Van Alstyne, Inverse Condemnation: Unintended Physical Damage (1969) 20 Hastings L.J. 431 , 489-490 (Van Alstyne).) Based on this rule, the Court of Appeal *437 correctly concluded the District is not liable for the flood damage to plaintiffs’ property. 2 We therefore affirm the Court of Appeal judgment.

I. Facts and Procedural Background

The District, a state agency, possesses flood control powers within a defined area of the Coachella Valley, including Magnesia Springs Canyon and the City of Rancho Mirage. Magnesia Springs Canyon is a natural watercourse that drains a watershed of about five and one-quarter square miles in the mountains south of Rancho Mirage. (See San Gabriel V. C. Club v. Los Angeles (1920) 182 Cal. 392, 397 [ 188 P. 554 , 9 A.L.R. 1200 ] [defining “natural watercourse” as a channel with defined bed and banks used by water naturally passing down as a collected body].) The discharge from Magnesia Springs Canyon watershed flows north, passes over Magnesia Falls, and forms the Magnesia Cove alluvial plain as it spreads toward Rancho Mirage.

Beginning in 1948, in an attempt to protect property from historical flooding in Magnesia Springs Canyon, a private developer constructed flood control facilities on the Magnesia Cove alluvial plain to divert canyon floodwaters northwest into the West Magnesia Channel. The District has owned the facilities since 1966. They consisted of: (1) a training levee to move the apex of the alluvial fan (a fan-shaped accumulation of sediment) to the north; (2) a diversion dike to intercept the waters and divert them to the northwest side of the alluvial fan; and (3) a channel to carry the waters to the Whitewater River. Constructed of locally quarried, uncompacted, and unreinforced sand and soils, the flood control facilities successfully diverted and channeled storm runoff away from the Bunches’ apartment building until a series of unusually severe tropical storms struck the area in 1976 (Kathleen), 1977 (Doreen), and 1979 (Dolores). According to the Bunches’ experts, their property was located in the sheet-overflow area of the Magnesia Cove alluvial fan, an area subject to flooding before the flood control facilities were built.

In 1976, the flood control facilities failed during Tropical Storm Kathleen at or near the point where they were intended to divert the floodwaters’ natural direction. Kathleen caused substantial damage throughout the entire Coachella Valley. Less than 2 percent of that damage, however, occurred on the Magnesia Cove and in Rancho Mirage. The storm did not damage the Bunches’ property. Local runoff, not failure of the District’s flood control facilities, caused the Rancho Mirage damage. From at least 1948 until Kathleen arrived, the facilities had successfully controlled all runoff from rain in the Magnesia Springs Canyon watershed.

*438 In the wake of Kathleen—whose floods dug a 10-foot-wide breach in the works, through which water flowed 50 to 150 yards beyond—the District undertook emergency repairs using additional natural materials. The repaired facilities withstood the runoff that Tropical Storm Doreen produced in 1977. At that time, the District authorized Bechtel Corporation to conduct flood control studies in Palm Desert, Indian Wells, and Rancho Mirage. The study was in progress in July 1979, when Tropical Storm Dolores struck.

Dolores was the most severe tropical storm in the recorded history of that region of the state; it is sometimes called the “300-year flood.” The floods Dolores caused overtopped the dike and levee at the point where the facilities were designed to divert the floodwaters, causing about $20 million in damage to the Magnesia Cove cities. Rains from Dolores led to the flooding of the Bunches’ apartment building, which was inundated with water, mud, and debris flowing from the point of the breach in the flood control facilities in a concentrated manner and at an abnormally rapid rate of flow. A wall along the southerly property line collapsed, and mud and debris buried automobiles parked on and adjacent to the property. The Bunches’ real property damage totaled $690,000.

In view of the damage Dolores caused, in November 1979, the United States Army Corps of Engineers agreed to participate in constructing flood control improvements. Their design, approval, and installation took nearly eight years to complete and cost the District about $7 million.

In 1982, the Bunches filed an inverse condemnation action against the District, seeking compensation for its physical invasion and destruction of their property. The court tried the liability issue, and a jury tried the damages issue. Applying then traditional standards of inverse condemnation, the trial court concluded the District was strictly liable for the Bunches’ property damage, without regard to the reasonableness of its flood control measures, because the breach in its facilities was a substantial cause of the damage. (See Albers v. County of Los Angeles (1965) 62 Cal.2d 250, 257 [ 42 Cal.Rptr. 89 , 398 P.2d 129 ] (Albers) [government strictly liable for damage caused by public improvement as designed and constructed].)

The jury awarded the Bunches $690,000 in damages. The District appealed from the judgment; the Bunches appealed from the judgment with regard only to the interest calculation included as part of the damages award. While the appeals were pending, this court decided Belair, which addressed the liability issue in the context of flood control facilities designed to protect against the “ ‘common enemy’ of floodwaters.” (Belair, supra, 47 Cal.3d at p. 565 .) Belair held that “. . . when a public flood control improvement fails *439 to function as intended, and properties historically subject to flooding are damaged as a proximate result thereof, plaintiffs’ recovery in inverse condemnation requires proof that the failure was attributable to some unreasonable conduct on the part of the defendant public entities.” (Id. at p. 567.)

In light of Belair, the Court of Appeal concluded that the issue for the trial court was not whether the District was strictly liable for the Bunches’ damages—the theory on which the case was tried—but whether the failed flood control facilities’ design, construction, and maintenance had been reasonable. (Bunch v. Coachella Valley Water Dist. (1989) 214 Cal.App.3d 203, 216-217 [ 262 Cal.Rptr. 513 ] (Bunch I).) Bunch I opined that Belair's analysis applied to all cases involving unintentional water runoff, whether they involved facilities designed to keep water within its natural course or designed to divert water safely away from a potentially dangerous natural flow. (Bunch I, supra, 214 Cal.App.3d at p. 213 .)

Following a second trial limited to the reasonableness question, the court ruled the District had acted reasonably under Belair, supra, 47 Cal.3d 550 , and found it not liable for the damage that occurred after floodwaters from Dolores breached its flood control facilities. The Bunches, contending that the trial court misunderstood and misapplied Belair, again appealed, arguing that the finding of reasonableness was erroneous as a matter of law.

The Court of Appeal affirmed the trial court judgment, applying much of the same reasoning it applied in Bunch I. On rehearing, the court rejected the Bunches’ argument that our recent Locklin decision restricted Belair's reasonableness rule to those cases involving activity immune from liability under the common law. The court concluded the District acted reasonably in designing, constructing, and maintaining its flood control facilities and should not be held liable because those facilities failed to protect the Bunches’ property from Dolores’s severe damage. As we explain, we agree with the Court of Appeal.

II. Discussion

A. Background

Historically, courts analyzed inverse condemnation liability issues by referring to traditional tort and property law concepts. This approach was due in part to the assumption that inverse condemnation liability was limited to situations in which a private party would be held liable for injury to property. (Belair, supra, 47 Cal.3d at p. 562 .) That general understanding changed with our decision in Albers, supra, 62 Cal.2d 250 , which “shifted *440 the focus in inverse condemnation cases from the common law to the Constitution.” (Belair, supra, 47 Cal.3d at p. 562 .)

The Albers case involved a landslide that county road construction caused. ( Albers, supra, 62 Cal.2d at pp. 254-255.) In finding the county liable for the landslide damage, Albers “ ‘rejected the notion that there need be a congruence between public and private liability in inverse condemnation actions.’ ” ( Belair, supra, 47 Cal.3d at pp. 562-563.) The Albers court made it clear that its liability rule did not derive from statutory or common law tort doctrine, but instead rested on the constitutional requirement of just compensation. (Albers, supra, 62 Cal.2d at p. 262 ; see Holtz v. Superior Court (1970) 3 Cal.3d 296, 303 [ 90 Cal.Rptr. 345 , 475 P.2d 441 ] (Holtz).) Albers relied on earlier precedent defining the relevant “policy” underlying our constitutional provision: “ ‘The decisive consideration is whether the owner of the damaged property if uncompensated would contribute more than his proper share to the public undertaking.’ ” (Albers, supra, 62 Cal.2d at p. 262 , quoting Clement v. State Reclamation Board (1950) 35 Cal.2d 628, 642 [220 P.2d 897](Clement)). In other words, as the Holtz court observed, Albers premised its inverse condemnation liability rule on its belief that “. . . the underlying purpose of our constitutional provision in inverse—as well as ordinary— condemnation is ‘to distribute throughout the community the loss inflicted upon the individual . . . .’” (Holtz, supra, 3 Cal.3d at p. 303 ; see Belair, supra, 47 Cal.3d at p. 558 .)

Thus, the Albers court held that, with two exceptions, our state Constitution’s just compensation clause requires that an owner of real property receive compensation for any actual physical injury to that property, whether foreseeable or not, that a public improvement, as deliberately designed and constructed, proximately caused. ( Albers, supra, 62 Cal.2d at pp. 263-264.) The Albers strict liability test was later restated in terms of “ ‘ “substantial” causation,’ ” so that landowners could establish inverse condemnation liability if the public improvement substantially caused the damage, although the improvement was only one of several concurrent causes. (Belair, supra, 47 Cal.3d at p. 559 ; see Souza v. Silver Development Co. (1985) 164 Cal.App.3d 165, 171 [ 210 Cal.Rptr. 146 ]; see also Ingram v. City of Redondo Beach (1975) 45 Cal.App.3d 628, 633-634 [ 119 Cal.Rptr. 688 ]; cf. Van Alstyne, supra, 20 Hastings L.J. at pp. 436-438 [noting that courts should express Albers’s proximate cause requirement in terms of substantial cause and effect relationship when eliminating foreseeability as an element of inverse condemnation].)

Albers recognized two historical exceptions to its strict liability inverse condemnation rule, “in which the urgency or particular importance of the *441 governmental conduct involved was so overriding that considerations of public policy [advised] against a rule rendering the acting public entity liable absent fault.” ( Holtz, supra, 3 Cal.3d at pp. 304-305.) The first exception involved damages a public entity inflicted in the proper exercise of its police power. (Id. at p. 305; see Gray v. Reclamation District No. 1500 (1917) 174 Cal. 622, 638 [ 163 P. 1024 ].)

The second exception to the strict liability doctrine occurred in the “unique” context of water law and derived from upper riparian private landowners’ limited common law privilege to defend themselves against the “common enemy” of floodwaters. (Archer v. City of Los Angeles (1941) 19 Cal.2d 19, 24-25 [ 119 P.2d 1 ] (Archer); Albers, supra, 62 Cal.2d at p. 262 ; Holtz, supra, 3 Cal.3d at pp. 304-306.) Under this traditional rule, a public or private entity’s liability for damage substantially caused by its flood control improvement efforts depended on whether its flood control system was intended to “improve” or “divert” the water’s natural flow. (Archer, supra, 19 Cal.2d at p. 26 .)

The “common enemy" doctrine granted private landowners the privilege of improving their property from external hazards such as floodwaters along a natural watercourse across that property. (Archer, supra, 19 Cal.2d at p. 26 .) As the Belair court explained in detail, under the “Archer exception” to the later Albers rule of inverse condemnation liability, public entities were granted unqualified immunity for activity that was privileged at common law, i.e., where the public entity had a common law right to inflict damage. ( Belair, supra, 47 Cal.3d at pp. 563-564.) The Archer exception, however, did not permit landowners to divert stream waters from their natural drainage across adjoining property. ( Clement, supra, 35 Cal.2d at pp. 637-638.) Thus, in cases involving stream water diversion as a flood control measure, the public or private landowners were held liable for any resulting damage without any specific determination of fault. (Ibid.)

The Archer exception to the Albers strict liability rule was soon criticized and later qualified by courts applying the doctrine in flood control cases. In Keys v. Romley (1966) 64 Cal.2d 396, 401 [ 50 Cal.Rptr. 273 , 412 P.2d 529 ] (Keys), Justice Mosk observed: “The courts which evolved and applied the extreme common enemy doctrine apparently acted from an exaggerated and uncritical respect for the rights and privileges of land ownership as expressed in the maxim cujus est solum [landowner owns everything above and below the land], together with an apparent belief that the only alternative would be to adopt the rule of natural servitude of natural drainage which would hinder the improvement of land and stultify economic development.” After reviewing the development of California water law, Keys adopted a *442 “reasonable use rule” with respect to the flow of surface waters from a dominant or upper landowner, (id. at pp. 407, 409.) This rule acknowledged California’s application of the civil law rule, under which owners of a dominant estate could, without liability, discharge surface water as it naturally flowed, but were required to act reasonably in using their property to avoid harming adjacent property. (Id. at p. 409.)

The Archer exception also created uneasiness among property law scholars, including Professor Van Alstyne, who observed: “The rationale of the ‘common enemy’ rule ... is of dubious validity when considered in the context of govemmentally administered flood control projects developed for the collective protection of entire regions. . . . HD Piecemeal construction, often an inescapable feature of such major flood control projects, creates the possibility of interim damage to some lands left exposed to flood waters while others are within the protection of newly erected works. Indeed, the partially completed works, by preventing escape of waters that previously were uncontrolled, actually may increase the volume and velocity of flooding .... The prevailing private law doctrine embodied in the ‘common enemy’ rule, however, imposes no duty upon the public entity to provide complete protection against flood waters .... Increased or even ruinous damage incurred by the temporarily unprotected owners, due to the inability of the improvements to provide adequate protection to all, therefore, is not a basis of inverse liability. The constitutional promise of just compensation for property damage for public use thus yields to the overriding supremacy of an anomalous rule of private law.” (Van Alstyne, supra, 20 Hastings L.J. at pp. 500-501, fhs. omitted; see also Holtz, supra, 3 Cal.3d at pp. 307-308, fn. 13.)

B. Belair

Belair reexamined the traditional distinctions in flood control law and attempted to reconcile the constitutional principles of Albers, supra, 62 Cal.2d 250 , with the unique problems flood control litigation created in an increasingly urbanized society. Belair involved the construction of public levees along a natural drainage course in an attempt to prevent flooding in an area historically subject to it. ( Belair, supra, 47 Cal.3d at pp. 555-557.) Relying in part on Justice Mosk’s historical analysis of property law in Keys, supra, 64 Cal.2d at pages 407-409, Belair recognized the dilemma modem courts increasingly faced when determining responsibility for damages caused by public works: “On the one hand, a public agency that undertakes to constmct or operate a flood control project clearly must not be made the absolute insurer of those lands provided protection. On the other hand, the damage potential of a defective public flood control project is clearly enormous.” (Belair, supra, 47 Cal.3d at p. 565 .) Belair concluded that these *443 competing concerns were best served in the flood control context by a rule of “reasonableness.” (Id. at pp. 565-566.) The court also suggested that in some cases an extraordinary storm may constitute an intervening cause that supersedes the public improvement in the chain of causation. (Id. at p. 560.) But because the Court of Appeal limited retrial to the reasonableness issue, significant here is Belair’s holding that “. . . the fact that a dam bursts or a levee fails is not sufficient, standing alone, to impose liability. However, where the public agency’s design, construction or maintenance of a flood control project is shown to have posed an unreasonable risk of harm to the plaintiffs, and such unreasonable design, construction or maintenance constituted a substantial cause of the damages, plaintiffs may recover regardless of the fact that the project’s purpose is to contain the ‘common enemy’ of floodwaters. [Citations.]” (Id. at p. 565, fn. omitted.)

Belair was careful to emphasize that its “reasonableness” rule would not discourage beneficial flood control projects, but would allow compensation for property that those projects unfairly damaged, creating a proper balance in constitutional takings jurisprudence. (Belair, supra, 47 Cal.3d at p. 565 ; see also Penn Central Transp. Co. v. New York City (1978) 438 U.S. 104, 124-125 [ 98 S.Ct. 2646, 2659-2660 , 57 L.Ed.2d 631 ] [citing reasonableness as a factor for consideration in inverse condemnation action]; Armstrong v. United States (1960) 364 U.S. 40, 49 [ 80 S.Ct. 1563, 1569 , 4 L.Ed.2d 1554 ] [takings clause does not obligate individual property owners to bear burdens that the general public should bear].) Belair also relied on Van Alstyne, who observed: “ ‘Mindful of the enormous damage-producing potential of defective public flood control projects, the courts have insisted that public agencies must act reasonably in the development of construction and operational plans so as to avoid unnecessary damage to private property. Reasonableness, in this context, is not entirely a matter of negligence, but represents a balancing of public need against the gravity of private harm.’ [Citation.]” ( Belair, supra, 47 Cal.3d at pp. 565-566, quoting Van Alstyne, supra, 20 Hastings L.J. at p. 455, fns. omitted.)

Belair acknowledged that the “common enemy” doctrine did not permit a public entity to divert water from its natural channel or drainage, and recognized that some post-Albers decisions may have endorsed a rule of strict liability. (Belair, supra, 47 Cal.3d at p. 566 ; see, e.g., Imperial Cattle Co. v. Imperial Irrigation Dist. (1985) 167 Cal.App.3d 263 [ 213 Cal.Rptr. 622 ] [irrigation system overflowed, diverting water onto plaintiffs’ property]; McMahan’s of Santa Monica v. City of Santa Monica (1983) 146 Cal.App.3d 683 [ 194 Cal.Rptr. 582 ] [water main diverting water under plaintiff’s property failed, damaging property]; Yee v. City of Sausalito (1983) 141 Cal.App.3d 917 [ 190 Cal.Rptr. 595 ] [storm drainage system that *444 diverted water across plaintiff’s property failed, and water seeped into plaintiff’s soil, causing subsidence]; Marin v. City of San Rafael (1980) 111 Cal.App.3d 591 [ 168 Cal.Rptr. 750 ] [underground drainage pipe diverting water under plaintiffs’ property failed, causing property damage]; Granone v. County of Los Angeles (1965) 231 Cal.App.2d 629 [ 42 Cal.Rptr. 34 ] [county strictly liable for public flood control works defectively constructed and negligently maintained].) This distinction for cases involving the intentional diversion or obstruction of a channeled flow of stream water from its natural course served as the foundation for the Bunches’ appeals in the present matter and in Bunch I.

Without expressly disapproving those Court of Appeal decisions finding strict liability for diverted surface waters, Belair observed in dictum that its “reasonableness” analysis would most likely apply to future diversion cases when the government sought to prevent flooding on land potentially subject to it. The court doubted whether unintended property damage caused by diversion of water “would elevate the test of inverse condemnation liability to absolute liability, rather than a reasonableness standard.” (Belair, supra, 47 Cal.3d at p. 567 .) Belair emphasized that in all flood control cases, “. . . the purposes of the Constitution, rather than the rules ‘emanating from the complex and unique province of water law,’ must fix the extent of a public entity’s responsibility. [Citation.]” (Id. at p. 567, quoting Holtz, supra, 3 Cal.3d at p. 306 .)

Thus, under Belair’s analysis, to establish whether a public entity acted reasonably in its flood control measures, the court must evaluate the facts, balancing the public need against the gravity of private harm. “The reasonableness of the public agency’s conduct must be determined on the facts of each individual case, taking into consideration the public benefit and the private damages in each instance. [Citation.]” (Belair, supra, 47 Cal.3d at p. 566 ; see also Van Alstyne, supra, 20 Hastings L.J. at p. 455.)

C. Locklin

Our next case involving property damage that public flood control measures caused answered some of the questions Belair left open. In Locklin , we held that Belair’s rule with respect to flood control projects and the “common enemy” doctrine had broader application to cases in which a public entity deliberately drained excess surface water into a natural watercourse, which eventually caused damage to downstream property. (Locklin, supra, 7 Cal.4th at p. 367 .) There, the owners of property along a creek and natural watercourse sued various public entities after public and private development eventually led to increased water flow in the creek, which damaged *445 their property. Although we observed that the trial court erred in exempting the public entities from liability under Archer's “natural watercourse” rule (Archer, supra, 19 Cal.2d at p. 26 ), we nonetheless concluded that liability was unwarranted because plaintiffs failed to prove that any unreasonable public entity conduct caused their property damage (Locklin, supra, 7 Cal.4th at p. 367).

In a thoughtful discussion of the complex nature of water law and the archaic rules that traditionally governed liability questions, Locklin explained that following Belair, which involved only flood control projects, some lower courts continued to apply the Archer exception to cases in which the public entity invoked the natural watercourse rule. (Locklin, supra, 7 Cal.4th at p. 366.) As we observed, however, Belair “signalled not the continuation of the Archer exception, but its demise. It survived only vestigially in the limitation of inverse condemnation liability for public flood control projects in natural watercourses to damage resulting from a public entity’s unreasonable conduct. Thereafter, a public agency that acted unreasonably in regard to its use or alteration of a natural watercourse might be liable in inverse condemnation for downstream damage.” (Locklin, supra, 7 Cal.4th at p. 366, original italics.) In holding that upstream and downstream riparian owners must act reasonably in their flood control efforts, Locklin opined that “. . . with respect to flood control projects, the public agency is liable only if its conduct posed an unreasonable risk of harm to the plaintiffs, and that unreasonable conduct is a substantial cause of the damage to plaintiff’s property. The rule of strict liability generally followed in inverse condemnation (see Albers, supra, 62 Cal.2d 250, 263-264 ) is not applicable in this context.” (Locklin, supra, 7 Cal.4th at p. 367 .)

Thus, Locklin reasoned that when a public entity uses a natural watercourse to drain surface waters from improved public property and makes storm drainage improvements that cause creek banks to erode and suffer landslide damage during a rainstorm, “[ijnverse condemnation liability ultimately rests on the notion that the private individual should not be required to bear a disproportionate share of the costs of a public improvement. Moreover, whether compensation must be paid for damages caused by alterations in the flow of a natural watercourse involves a balancing of interests.” ( Locklin, supra, 7 Cal.4th at pp. 367-368.) This was essentially Belair's rationale. (Belair, supra, 47 Cal.3d at p. 566.)

In light of Belair, Locklin recognized that to apply the reasonableness rule to cases other than those involving previously privileged conduct, courts must identify “standards for balancing the interests of riparian landowners and assessing reasonableness in inverse condemnation actions.” (Locklin, *446 supra, 7 Cal.4th at p. 368 .) Noting that Keys stated “the basic requirements of reasonable conduct” (Locklin, supra, 7 Cal.4th at p. 368; see Keys, supra, 64 Cal.2d at p. 410 ), and that Professor Van Alstyne proposed standards for balancing landowners’ various interests, Locklin identified several factors that could be used in assessing public entity liability: “(1) The overall public purpose being served by the improvement project; (2) the degree to which the plaintiff’s loss is offset by reciprocal benefits; (3) the availability to the public entity of feasible alternatives with lower risks; (4) the severity of the plaintiff’s damage in relation to risk-bearing capabilities; (5) the extent to which damage of the kind the plaintiff sustained is generally considered as a normal risk of land ownership; and (6) the degree to which similar damage is distributed at large over other beneficiaries of the project or is peculiar only to the plaintiff.” ( Locklin, supra, 7 Cal.4th at pp. 368-369.) 3

In addition, Locklin observed that “ [reasonableness in this context also considers the historic responsibility of riparian owners to protect their property from damage caused by the stream flow and to anticipate upstream development that may increase that flow. Keeping in mind the purpose of the constitutional right to compensation for damage caused by public works and improvements—that property owners contribute no more than their proper share to the public undertaking—plaintiff must demonstrate that the efforts of the public entity to prevent downstream damage were not reasonable in light of the potential for damage posed by the entity’s conduct, the cost to the public entity of reasonable measures to avoid downstream damage, and the availability of and the cost to the downstream owner of means of protecting that property from damage.” (Locklin, supra, 7 Cal.4th at p. 369 .)

Locklin thus provided courts with guidelines for applying the reasonableness rule we first recognized in Keys, supra, 64 Cal.2d 396 , and later applied to determine a public entity’s liability when its flood control project failed in Belair, supra, 47 Cal.3d at pages 565-566. Locklin emphasized that “. . . the reasonableness of a landowner’s action in discharging surface water runoff into a natural watercourse or in altering the watercourse itself cannot be determined in isolation. An owner in the lower reaches of a natural watercourse whose conduct has a relatively minor impact on the stream flow in *447 comparison with the combined effect of actions by owners in the upper reaches of the watercourse may not be held liable for any damage caused by the stream flow beyond the proportion attributable to such conduct. If the rule were otherwise, owners at the lowest reaches of a watercourse could preclude development of upstream property by imposing on a single upstream owner the cost of all damage caused by the addition of surface water runoff if that addition combined with the existing stream flow damaged the lowest properties.” (Locklin, supra, 7 Cal.4th at p. 360 .)

Thus, a property owner whose unreasonable acts cause damage to neighboring property is liable only for its proportionate share of the damage. When the upstream owner is a public entity, the court must also consider the nature of the improvement or public work, the degree to which its value offset the property damage, and all other relevant matters, including whether the utility of the possessor’s use of land outweighs the risk of damage to the property. ( Locklin, supra, 7 Cal.4th at pp. 359-360.) While the public entity must bear its proportionate share of liability, it is not subject to joint liability, but only to several liability. (Id. at p. 372, citing Mehl v. People ex rel. Dept. Pub. Wks. (1975) 13 Cal.3d 710, 718 [ 119 Cal.Rptr. 625 , 532 P.2d 489 ].)

D. Application of Belair and Locklin to Diversion Cases

The question before us is whether the reasonableness rule developed in the context of flood control improvements along natural watercourses should apply to cases in which a public entity diverts and rechannels water under a flood control system of dikes and levees that fail in a severe rainstorm, causing damage to properties historically subject to flooding. We conclude it should, and we disapprove those cases concluding otherwise.

The Bunches assert that the rule should not apply here because both Belair and Locklin expressly limited its application to conduct that would have been privileged under common law doctrines, i.e., the Archer “common enemy” exception. Because the diversion of waters from their natural course was not a privileged activity before the Belair decision, the Bunches assert the District should be strictly liable under the Albers rule for damage its flood control measures caused. ( Albers, supra, 62 Cal.2d at pp. 263-264.) The Bunches rely on numerous pre-Belair cases that have applied an absolute liability standard to flood control cases involving the diversion of surface or floodwaters from their natural channel or drainage. (See, e.g., Youngblood v. Los Angeles County Flood Control Dist. (1961) 56 Cal.2d 603 [ 15 Cal.Rptr. 904 , 364 P.2d 840 ]; Imperial Cattle Co. v. Imperial Irrigation Dist., supra, 167 Cal.App.3d 263 ; McMahan’s of Santa Monica v. City of Santa Monica, *448 supra, 146 Cal.App.3d 683 ; Yee v. City of Sausalito, supra, 141 Cal.App.3d 917 ; Marin v. City of San Rafael, supra, 111 Cal.App.3d 591 ; Granone v. County of Los Angeles, supra, 231 Cal.App.2d 629 .)

As the Bunches observe, these decisions hold that a public entity’s diversion of waters from their natural course that results in damage to private property requires payment of just compensation to the property owner. The Bunch I court rejected the Bunches’ argument that these pre-Belair cases should control the result here, concluding that Belair’s “far-reaching and elaborate analysis” should be applied “beyond the limited facts of that case.” (Bunch I, supra, 214 Cal.App.3d at p. 212 .) We discuss the Bunches’ argument below. 4

1. BelairILocklin rule in diversion cases

As noted, the Bunches assert that Belair’s reasonableness rule should apply only to those limited cases in which the public entity’s conduct would have been privileged at common law, and that the pre-Belair strict liability rule should apply to cases involving government diversion of water from its natural course. (See, e.g., McMahan’s of Santa Monica v. City of Santa Monica, supra, 146 Cal.App.3d at p. 698 .) But Belair warned against limiting application of its rule to cases involving previously privileged conduct. Thus, in response to the plaintiffs’ contention that the diversion cases endorsing a rule of liability without fault should also apply to nondiversion situations, Belair concluded it need not examine the validity of those diversion decisions because plaintiffs presented no evidence that the District affirmatively diverted or burdened their property with more floodwaters than would have escaped without the District’s flood control measures. (Belair, supra, 47 Cal.3d at p. 567 .) Belair doubted “whether evidence of an unintended ‘diversion’—an elusive concept to begin with [citation]—would elevate the test of inverse condemnation liability to absolute liability, rather than a reasonableness standard. As earlier discussed, the purposes of the Constitution, rather than the rules ‘emanating from the complex and unique province of water law,’ must fix the extent of a public entity’s responsibility. [Citation.]” (Ibid.)

Belair’s dictum indicates that the court believed its analysis could apply to all flood control cases involving unintended property damage. Nonetheless, *449 our reluctance to extend Belair beyond its facts is the basis for the Bunches’ assertion that the case has limited application, an argument aided, in some respects, by Belair’s invocation of tort concepts of “intent” in a case addressing a public entity’s inverse condemnation liability under section 19. The Bunches also rely on Locklin’s observation that Belair involved an activity formerly “privileged under the Archer doctrine.” (Locklin, supra, 7 Cal.4th at p. 354, fn. 16; see also Los Osos Valley Associates v. City of San Luis Obispo (1994) 30 Cal.App.4th 1670, 1680 [ 36 Cal.Rptr.2d 758 ] [dictum interpreting Locklin as creating exception to Albers in limited cases where the state at common law had privilege to inflict damage].)

As the District and its amici curiae observe, however, Belair’s policy reasons for imposing a reasonableness rule in common enemy cases—to avoid discouraging beneficial flood control improvements, while compensating losses unfairly incurred—extend logically to all cases involving flood control improvements affecting property historically subject to flooding, without regard to whether the activity was privileged at common law. A constitutional analysis for determining inverse condemnation liability in the flood control context should not include “a fruitless search for the somewhat artificial moral elements inherent in the tort concepts of negligence and intentional wrongs.” (Van Alstyne, supra, 20 Hastings L.J. at p. 495; see also Van Alstyne, Statutory Modification of Inverse Condemnation: The Scope of Legislative Power (1967) 19 Stan.L.Rev. 727, 771-776.) As Professor Van Alstyne recognized, “. . . it is arguable that strict liability for damage resulting from the diversion of water flowing in a natural watercourse may be reasonably sensible as applied to adjoining riparian owners; a contrary view would expose settled reliance interests to the threat of repeated and diverse private interferences that could discourage natural resource development. Stream diversions, however, may be integral features of coordinated flood control, water conservation, land reclamation, or agricultural irrigation projects undertaken on a large scale by public entities organized for that very purpose. Where this is so, the community may suffer more by general fiscal deterrents resulting from indiscriminately imposed strict liabilities than by specifically limited [liabilities] determined by the reasonableness of the risk assumptions underlying each diversion.” (Van Alstyne, supra, 20 Hastings L.J. at p. 502, fn. omitted.)

Moreover, contrary to the Bunches’ contention, nothing in Locklin or its progeny precludes application of Belair’s reasonableness rule to cases involving the failure of flood control measures designed to divert potentially dangerous natural water flow. Pre-Belair cases considering inverse condemnation liability under the just compensation clause addressed whether owners of damaged property would contribute more than their proper share to the *450 public undertaking if not compensated for the damage. (Ante, at pp. 439-440; see, e.g., Clement, supra, 35 Cal.2d at p. 642 .) In light of Belair and Locklin , this principle is balanced by the possibility that imposing open-ended liability on public entities charged with creating and maintaining flood control improvements will discourage the development of needed public works. (Belair, supra, 47 Cal.3d at p. 565 .)

As Professor Van Alstyne explained, “Plan or design characteristics that incorporate the probability of property damage under predictable circumstances may later be judicially described as ‘negligently’ drawn; yet, in the original planning process, the plan or design with its known inherent risks may have been approved by responsible public officers as being adequate and acceptable for non-legal reasons. For example, the damage, although foreseeable, may have been estimated at a low order of probability, frequency, and magnitude, while the added cost of incorporating minimal safeguards may have been unacceptably high in proportion to available manpower, time and budget. . . . The governmental decision ... to proceed with the project under these conditions thus may have represented a rational (and hence by definition non-negligent) balancing of risk against practicability of risk avoidance.” (Van Alstyne, supra, 20 Hastings L.J. at pp. 489-490, fns. omitted.)

Thus, the placement, design, and construction of even the most effective system inherently involve a complex balancing of interests and risks. Whatever choice the responsible agency makes will necessarily affect the patterns of flooding in the event the project fails, and will almost certainly increase certain risks in order to reduce others. The dangers posed to individual lands by the failure of any public flood control project are “potentially enormous” and sometimes deserve compensation. However, strict and “open-ended” liability for the failure of a project whose overall design, construction, operation, and maintenance was “reasonable” would unduly deter the development of these vital bulwarks against common disaster. (See Belair, supra, 47 Cal.3d at p. 565 .)

In the context of inverse condemnation, therefore, a flood control agency does not necessarily exact “disproportionate,” and thus compensable, contributions from particular landowners simply because it constructs adjacent flood control improvements that may alter how floodwaters will affect those landowners if the improvements fail to contain the flow. When a public flood control system fails to protect land from historic periodic flooding, the only way to determine whether a damaged private landowner has thereby been forced to contribute a compensable “disproportionate” share of the public undertaking is to determine whether the system, as designed, constructed, operated, and maintained, exposed him to an “unreasonable” risk of harm, either individually or in relation to other landowners.

*451 Therefore, when a public flood control improvement designed to divert or rechannel potentially dangerous water flow is a substantial cause of property damage, courts must balance “ ‘public need against the gravity of private harm’ ” in determining whether to compensate the landowners for that damage. (Belair, supra, 47 Cal.3d at p. 566 , quoting Van Alstyne, supra, 20 Hastings L.J. at p. 455, fn. omitted.) In balancing these interests, courts must weigh the factors set forth in Locklin, supra, 7 Cal.4th at pages 368-369. (Ante, at pp. 445-446.) We derive from Belair, and from the Locklin factors applying its rule, the principle that compensation in cases involving the failure of flood control improvements to protect property adequately against historic periodic flooding should rest not on antiquated notions of fault, or common law labels defining the type of waters requiring flood control measures, but rather on the balancing of interests that section 19 requires. This balancing of interests serves both the private sector and public improvement efforts by addressing the cost-spreading objective of the just compensation clause while protecting public entities from unlimited, undeserved liability that could well inhibit further construction of public works.

2. Applying the Belair/Locklin test

The Bunches next assert that even if we do apply a Belair/Locklin balancing test pursuant to the objectives of section 19, the trial court should have considered only the District’s reasonableness in constructing the particular facility that failed, rather than conducting an assessment of all District flood control facilities. Bunch I had remanded the case for a determination of the District’s reasonableness in light of Belair. The Bunches assert that, on remand, the trial court impermissibly relied on the District’s limited financial resources as the principal ground for its finding of reasonableness and did not properly consider the District’s design, construction, maintenance, or operation of the failed facilities. The Bunches rely on several decisions limiting reliance on financial considerations in determining constitutional rights. (See, e.g., Watson v. Memphis (1963) 373 U.S. 526 [ 83 S.Ct. 1314 , 10 L.Ed.2d 529 ] [city could not refuse to desegregate public parks on theory that it was less expensive to deny constitutional rights than to afford them]; In re Grimes (1989) 208 Cal.App.3d 1175 [ 256 Cal.Rptr. 690 ] [officials could not use budget shortage to deny prisoners reasonable telephone access to counsel]; Inmates of Sybil Brand Institute for Women v. County of Los Angeles (1982) 130 Cal.App.3d 89, 102 [ 181 Cal.Rptr. 599 ] [budget constraints could not justify denial of equal protection to female inmates].)

As the District observes and the record indicates, the trial court may have considered evidence of the District’s limited budget and its allocation of *452 funds among all its activities. This evidence was relevant and admissible. (Locklin, supra, 7 Cal.4th at p. 369 .) As Professor Van Alstyne observed, fiscal constraints are part of the careful balancing of the public and private interests involved in the initial determination of the government’s reasonableness in its shifting the risk of loss from flooding to private resources. He wrote: “Assuming foreseeability of damage, the critical factors in the initial stage of the balancing process relate to the practicability of preventive measures, including possible changes in design or location. If prevention is technically and fiscally possible, the infliction of avoidable damage is not ‘necessary’ to the accomplishment of the public purpose. ... [f] On the other hand, if the foreseeable type of damage is deemed technically impossible or grossly impracticable to prevent within the limits of the fiscal capability of the public entity, the decision to proceed with the project despite the known danger represents an official determination that public necessity overrides the risk of private loss. The shifting of the risk of loss to private resources is not sought to be supported on grounds of mere prudence or expedience but on the view that the public welfare requires the project to move ahead despite impossibility of more complete loss prevention. In this situation, an additional variable affects compensation policy. The magnitude of the public necessity for the project at the particular location, with the particular design or plan conceived for it, must be assessed in comparison to available alternatives for accomplishing the same underlying governmental objective with lower risk, but presumably higher costs . . . .” (Van Alstyne, supra, 20 Hastings L.J. at pp. 491-492, fns. omitted.) Thus, although fiscal constraints are never alone determinative of the government’s reasonableness in its flood control measures, they are a relevant consideration in the overall balancing test and reasonableness determination.

As the District also points out, however, the trial court did not base its finding of reasonableness solely on the District’s financial limitations. Instead, the trial court properly considered other relevant evidence, including substantial expert testimony on the reasonableness of the District’s flood control measures and remedial action following Tropical Storm Kathleen. Based on all the evidence presented, the trial court issued a statement of decision that concluded the District acted with “commendable promptness” in repairing the breach Kathleen caused. The court believed the District acted promptly to obtain professional information and expertise in contacting Bechtel within days of Kathleen’s damage. The court also concluded the District acted reasonably in its effort to provide the Coachella Valley with effective flood control.

As the Court of Appeal observed, the Bunches’ expert, consulting engineer Wilbur Lockman, testified that in his opinion the construction of the *453 flood control facilities was unreasonable for their intended purpose. Nonetheless, Lockman conceded the District was not on notice of the need for repair or redesign of the flood control facilities until Tropical Storm Kathleen in 1976.

Dr. Simons, also a consulting engineer, testified that better remedial measures after Kathleen would have included realigning the angle at which the levee diverted water toward the dike and stabilizing the channel with rock riprap grouted with concrete or soil cement.

Several experts testified on the District’s behalf, including Joseph Countryman, a consulting engineer. Countryman testified that Dolores was twice as large as a 100-year storm and larger than the “standard project flood” the United States Army Corps of Engineers used in designing flood protection measures. Countryman stated that Bechtel had reviewed the corrective measures Lockman and Simons recommended and had rejected them because they would not have worked had they been implemented. In Countryman’s opinion, the District’s actions following Kathleen in 1976 comported with sound engineering practice and were reasonable.

Thomas Levy, the District’s general manager and chief engineer, testified that the District retained Bechtel in 1976 within days of Kathleen to prepare a flood control plan for the communities of Indian Wells, Rancho Mirage, and Palm Desert. As Levy noted, Bechtel’s plan took several years to implement because it required the District to address environmental issues.

In sum, the record shows the trial court’s conclusion the District acted reasonably and promptly in repairing post-Kathleen damage and in taking steps to develop a more permanent flood control plan was sound. Expert testimony indicated the detailed steps the District undertook to secure expert technical and engineering information on repairing its facilities and in managing those repairs represented sound engineering practice. Expert testimony was in conflict as to the engineering feasibility of the short-term solution of repairing the breach in the West Magnesia Springs Canyon stormwater facilities to restore the preexisting level of flood protection, allowing the District to devote its remaining resources to developing a comprehensive flood control plan. But as the Court of Appeal observed, the trial court reasonably relied on the District’s expert testimony that any attempt to “quick fix” the problem by realigning the diversion levee and lining it with stronger material would have failed during Tropical Storm Dolores. Thus, the trial court based its finding of reasonableness on a careful analysis of several factors, including the project’s purpose to divert the potentially dangerous natural flow in order to prevent flood runoff from *454 fanning out over the Magnesia Cove alluvial plain, the public need for the project, the risk of private harm, the alternatives available to reduce the risk of harm, and the District’s overall fiscal and flood control responsibilities.

The Court of Appeal concluded that this evidence satisfied Belair's reasonableness rule by applying “the basic requirements of reasonable conduct” that the Keys court identified (Locklin, supra, 7 Cal.4th at p. 368 ; see Keys, supra, 64 Cal.2d at p. 410 ) and that the Locklin court applied (Locklin, supra, 7 Cal.4th at p. 369 ). Although the Court of Appeal did not specifically refer to the six factors Locklin identified for use in assessing liability, factors which we hold courts should apply uniformly in all flood control cases, we nonetheless conclude the trial court adequately assessed the reasonableness of District’s flood control measures by using a comparable test.

The Bunches also contend that Belair and Locklin should not apply to cases in which the storm that breached the flood control facility exceeded its design capacity. Here, the District’s improvements were designed to withstand a “25-year flood.” According to the Bunches, because the tropical storm that broke the levee was considered a “300-year flood,” the District should be strictly liable for its inadequate improvements that were designed to protect against a “25-year flood,” notwithstanding the Belair/Locklin reasonableness rule. But as we discussed, even Belair observed that a public entity that undertakes to construct improvements to protect property from potentially dangerous flood damage cannot be made the “absolute insurer” of those lands it seeks to protect. (Belair, supra, 47 Cal.3d at p. 565 .) Henceforth, the focus in flood control cases should be whether, under the factors discussed above, the public entity acted reasonably in designing, constructing, and maintaining its flood control systems, not whether it could insure that no flooding would ever occur on that same property. Thus, the Bunches’ final argument has no merit.

III. Conclusion

The Belair/Locklin reasonableness test applies to cases involving public flood control works that cause physical damage to private property. As the Court of Appeal concluded, the trial court properly understood and applied the Belair reasonableness test. It used many of the balancing factors that Locklin subsequently applied. We conclude that courts should use these factors in cases where a public entity’s flood control measures, designed to protect against potentially dangerous periodic flooding, cause property damage. Although the Court of Appeal did not directly apply those factors, we find nothing in Locklin that requires us to modify the court’s disposition. We therefore affirm the Court of Appeal judgment.

*455 George, C. J., Mosk, J., Kennard, J., Baxter, J., Werdegar, J., and Brown, J., concurred.

As in Belair, this case does not present, and we do not decide, the question whether the reasonableness standard applies when flood control measures cause flood damage to land that was not historically subject to flooding. (See Belair, supra, 47 Cal.3d at p. 565 .) Of course, if the government, by works it constructs on its own property or elsewhere, diverts or dams natural waters, thereby permanently submerging previously dry private land in order to provide benefits to the public at large, a compensable direct “taking” of the submerged land may occur no matter how “reasonable” the government’s conduct. Moreover, cases under the federal Constitution have held that a “taking” may occur when a government dam or flood control project subjects certain previously dry lands to inundation that is less than permanent, but is frequent and inevitably recurring. (Barnes v. United States (1976) 538 F.2d 865, 870 [ 210 Ct.Cl. 467 ], and cases cited.)

Plaintiffs are Kenneth and Deidre Bunch (the Bunches).

The Locklin factors contemplate that the inquiry into “reasonable” design, construction, operation, and maintenance is not limited to a narrow examination whether the system’s technical specifications, intended capacities, materials, workmanship, and repairs were adequate under all the circumstances. ( Locklin, supra, 7 Cal.4th at pp. 368-369.) Instead, the inquiry should include specific consideration whether the location and configuration of the system, and its purpose to divert the natural flow, were themselves “reasonable.” In other words, the “reasonableness” of the system encompasses a determination whether, all things considered, it should not have diverted, and thus reallocated, the risks of periodic or sudden flooding as it was designed to do.

Initially, the District contends that the law of the case doctrine precludes the Bunches from raising questions about the effect of pre-Belair cases because Bunch I rejected those issues. That doctrine affects the rights of the parties, but does not preclude us from discussing Locklin’s effect on application of Belair’s reasonableness rule in this case to clarify the law. (See DiGenova v. State Board of Education (1962) 57 Cal.2d 167, 179-180 [ 18 Cal.Rptr. 369 , 367 P.2d 865 ].) It is unnecessary for us to decide whether the doctrine precludes the Bunches from challenging the decision in Bunch I because our decision is consistent with the holding in Bunch I.