A Response to the "Section 216 Review of Completed Works Reconnaissance Report, Coralville Lake" (dated May 1997, released August 1997)
By Larry Molnar and Charles Newsom; September 15, 1997
In August, a letter was sent by the Army Corps of Engineers notifying the residents of the Iowa River Valley of the findings of the Review of Completed Works Reconnaissance report concerning Coralville Lake: "The study concludes that there is no Federal interest in further study of flood damage reduction measures at the Coralville Lake project." Despite enormous physical and economic changes, the old management plan turns out to be the optimal plan for future operations. The letter also stated that the study was coordinated with the various government entities in the Iowa River valley.
However, close inspection of the letter and of the report reveals a number of alarming inconsistencies that call into question the commitment and the competency of the Corps to honor the promises made following the flood of 1993 and to manage Coralville Lake. In the following response, we treat the following five key issues.
We therefore recommend that the Corps reconsider this report and their future plans for revising lake operations. We urge all affected people in the river basin, urban and rural, to express their concerns to the Corps about both the procedures and the content of this report.
1) Public Review and Coordination
The standard policy of the Corps of Engineers before finalizing a policy is to solicit both public input at the outset of the development of a policy and public review of the conclusions the Corps recommends. This is a natural and desirable policy for a service-oriented government entity. Even an ideal policy will meet with public resistance if there is no mechanism to obtain public approval. The most recent case in which this procedure was followed for Coralville Lake was the 1991 study that led to a change in the seasonal pool elevations.
The present study was commissioned to study future operation of Coralville Lake under the greatly changed economic and climate conditions described in the Initial Appraisal and in the Damage Survey reports. In finding that the old operating plan is the optimal plan and by recommending no further study, the present study sets the policy for future management of the lake. The public input for this study was sought in the standard manner (as is documented in Appendix C). The promise of public review of the conclusions was set out in Col. Kraus's May 3, 1994 letter (omitted from Appendix C) to the Iowa City mayor in response to her March 2, 1994 letter (p. C-1) requesting a timetable for review of dam operations. The need for public review was also implied in the August 1996 letter when it stated that public coordination had been done. However, the only coordination mentioned in the report was the letter sent March 7, 1997 (p. C-56) which dealt with environmental issues only, not operational ones. (The Iowa City city engineer, Rick Fosse, confirmed the intent of this letter in a phone call made in March.)
With regard to the study of the operation of the lake, the first and only information made public before the report was finalized as to what options were being considered and what methods were being used was the cryptic description given in the December 1996 newsletter (p. C-48). The newsletter ends with the promising paragraph: "We encourage open communication throughout our study process. If you have questions or comments about the Coralville study, or wish to suggest an alternative to be considered in this study, please call our Coralville Lake study manager, ..." Acting on this message, Dr. Molnar called both the study manager and the planning division point of contact, Martin Hudson, to ask for clarification about the mention of flowage easements in the newsletter and about what recommendations the study would make. He was informed this information could not be released to the public until the study had been reviewed at all internal levels and a final version was published. While this contradicted the newsletter, we still assumed that there would be an opportunity for public review upon publication of the report. But no policy review process was mentioned in the report or the announcement letter. Dr. Molnar contacted Martin Hudson on 6 August to inquire about the public review procedure or any future study plans. He was unaware of any definite plans for public review or further study, but promised to get in touch when he found information on this. No further information has been received at this writing.
A possible reason for the lack of public review at this time is that the Corps does not consider the report findings cited in the letter to be the current policy. In the "Recommendations" on p. 67, it states that although revision of the Water Control Plan is not in the Federal interest, it would nonetheless be an "aid" to future operations. Also, in his July 1995 letter (p. C-16), District Engineer McCully indicated that an interim regulation plan would be developed by the end of FY97.
If in fact a thorough study of the water control plan is still being planned, as we hope it is, it would be helpful to public coordination to send a second letter correcting the statements of the first. A brief explanation of the relationship between the conflicting policy statements would also be informative.
2) The Need for Quantitative Uncertainties
Unlike all scientific and engineering literature, the Reconnaissance Report lacks any estimation of the uncertainties of its quantitative results. As noted on page 2 of "Guidelines for Determining Flood Flow Frequency" (a manual the Corps is supposed to use to guide study methodology): "User decisions must be based on properly applied procedures and proper interpretation of results considering risk and uncertainty." No valid conclusions can be drawn without an estimate of the uncertainty. The central conclusion of the Reconnaissance Report is that no further investigation of operations alternatives is warranted. The criterion to warrant further investigation is that the benefit-to-cost ratio must exceed 1.0. The conclusion is based on the finding that the maximum benefit-to-cost ratio for alternative operation plans (i.e., the larger of the two values they computed) is 0.03, which is less than 1.0. However, if the uncertainty on 0.03 is a factor of 34, the benefit-to-cost ratio exceeds 1.0 and investigation is warranted after all.
To be sure, even when not explicitly stated uncertainties are sometimes indicated implicitly by the number of nonzero digits used. Furthermore a factor of 34 may seem an implausibly large uncertainty. Leaving the specifics of the Reconnaissance Report to the next sections, we cite an example from the 1991 report to show the number of digits is not used at Rock Island to indicate implicit uncertainty, and that a factor of 34 is not an implausible uncertainty in principle. Urban damages between the dam and the English River in a 100 year flood were estimated in the 1991 report to be $32,700 (p. F-5 of that report). Estimates of the same quantity in the Reconnaissance Report (based on the actual damages that occurred just two years after the 1991 report) are $16,705,900 (p. B-4). By the number of significant digits, one would have expected accuracy to within a factor of 1.01, but the actual error was a factor of 510.
3) Ad hoc nature of recommended plan
On p. A-5 the recommended plan is described as being summarized in Table A-4. This table calls for maximum possible outflow whenever the lake exceeds the height of the spillway. It is clear from the further calculations that what is being recommended is not this, but rather the procedures actually in use since 1993: the Rock Island engineers negotiate the outflows with the Division Commander (then in Chicago now Vicksburg) on an ad hoc basis (p. C-9). An operational difficulty with this approach found in 1993 is that it requires some days to get approval of the first variation. A methodological difficulty with including such an option in the report is that it is ambiguous what outflows to use in simulating high flow events.
4) Methodology of Flood Control Optimization
According to the enabling legislation for the Coralville dam, "The reservoirs planned for the upper Mississippi River Basin have been designed primarily for local flood control" (House Report 2353-19, May 1938). Likewise, the reservoir operation should be "... designed to produce the maximum practicable total benefit to all interests concerned" (Public Law 761, June 1938). Determining the "maximum practicable benefit" requires an optimization of the many parameters that go into an operating plan.
We consider below 5 key problems with methodology that make it impossible for the Reconnaissance Report to determine an optimal plan. The first is a general consideration, the next three consider estimation of benefits, and the last (and most decisive) point considers estimation of costs.
4.1) Searching parameter space
Nearly 50 years ago, when the first operating plan for Coralville Lake was being designed, a full optimization was not possible due to computational constraints. So a relatively small number of alternatives were computed in detail. Personal computers can now fully evaluate a single plan in just a few minutes time, so computational constraints are no longer relevant. Modern optimization techniques would involve computing benefit-to-cost ratios of thousands of alternatives and automatically selecting the most promising for closer scrutiny. For this Reconnaissance Report, however, only two alternative operating plans were computed! No sample of just two alternatives can be considered a serious exploration of the possibilities, and certainly not a determination of maximum practicable benefit.
4.2) Including key parameters: downstream thresholds
What is more, neither of the alternatives actually computed included any thresholds for protection of the lower reaches of the river. Neglect of this important parameter led to the trivial conclusion that neither of the plans were optimal. As we are aware of nobody who has suggested removing downstream constraints altogether, the choice of study alternatives is baffling. Dr. Molnar's written input at the outset of the Reconnaissance Study explicitly suggested determining optimal downstream thresholds (p. C-30). By including this parameter a large negative term in the computation of benefits of an optimized operation plan would be eliminated.
4.3) Trends in river flow
A second characteristic of old ways of designing an operations plan is to assume the absence of climate cycles, and therefore to treat all experience in the period of record equally. It is now recognized that beyond the oscillation between wet and dry years, long term correlations exist and are helpful in making predictions for coming years. For their forecasts, the National Weather Service uses a 30 year period to establish a norm. In the specific case of the Iowa River, the Corps is well aware (p. C-27) that the USGS documented in 1990 that the Iowa River has had a 50% increase in its flow since the dam was built compared to the 55 years preceding. The computations used in the Report ignore this trend, and in so doing dilute the importance of the 1993 event by a factor of about 3. The benefits of a plan that does well in such an event are therefore also underestimated by the same factor.
4.4) Inclusion of extreme events
Finally, since only the period of record is considered in the damage calculation, the performance of alternative models in a flood more severe than the flood of 1993 is not taken into consideration. One striking result in that event was the great degree to which FEMA regulations requiring new construction to be one foot above the 100 year flood line minimized damage. Actual river levels reached within a few inches of that line in Iowa City, damaging some homes, but leaving the great majority (those built to code) unharmed. However, a further increase of a few inches might have increased damages by almost tenfold. Post-1993 floodproofing efforts magnify the importance of this step still more. This distinction is evident in the table on page B-2 of the Report. Including such a flood in an estimation of benefit-to-cost ratios is just as important as including a flood like 1993 as the lower likelihood is offset by the greater impact. The benefits of a plan that does well in extreme events is therefore underestimated by a factor of about 2 because of this omission.
4.5) Costs of an optimized plan: flowage easements
The major costs listed in the Report for an alternative operating plan were possible flowage easements. No details of the legal requirements for such easements were mentioned in the Report, so we asked the authors in charge of hydraulics, legal counsel, and real estate what were the actual criteria used to estimate these costs. We were told that there was no legal basis for the costs that were used. They felt easements were necessary for the alternatives considered precisely because they were demonstrably not optimal. No flowage easements would be required for an optimal plan. By including a nonsensical cost, one can get any ratio one wishes, as is this case. By removing the spurious costs, the total costs diminish by a factor of 147, a great enough factor to change the fundamental conclusion by itself.
5) Omission of important conclusions from recommendations
5.1) Revision of downstream thresholds
On p. 30, the Report states that "the constraints have a noticeable effect on the regulation of the reservoir" and that such changes "could, with further study and coordination with local interests, justify raising the constraints." We noted in Section 3.2 above our agreement with these conclusions about the importance of additional study to determine optimal thresholds. The one downstream farmer represented at the input meeting on July 30, 1996, also voiced a desire to reconsider how these thresholds are managed. Curiously, however, this conclusion is omitted from the study conclusions and recommendations (p. 66-67) without comment.
Adding to the urgency of this conclusion is the absence of any written record of how the current levels were established. (This absence was established by the Corps a few years ago in response to a request for information by Dr. Molnar.) On p. 29, the general rationale for the existing downstream thresholds is described: concern about crops in season and physical structures out of season. However, as Dr. Molnar noted in his input (p. C-30), despite the regulation policy the non-crop thresholds are exceeded in 1 of every 5 years, while the crop thresholds are exceeded in 2 of every 5 years (virtually every wet year!). Hence it is questionable what structures still survive at the current thresholds after so many dousings, and at what level crops are currently planted.
5.2) A written flash flood contingency plan
As described in several places, automated gages were installed by the US Geological Survey in 1994 (with financial underwriting from Iowa City, Coralville, and the University of Iowa) to monitor the development of flash floods, and a standard operating procedure (SOP) was written by the Corps (in coordination with these entities) regarding dissemination of this information. The Corps has also installed a website which posts these numbers hourly (not every 15-30 minutes, as is erroneously stated on p. 65). On page A-12, it further recommends development of a written contingency plan that considers travel time of flood waters, etc., and can be used to guide operation of the dam in a flash flood. This recommendation repeats the promise made by District Engineer McCully in 1995 (p. C-15) and repeated by Planning Division Chief Hanson in 1996 (p. C-45). Curiously, however, this recommendation is also omitted from the study conclusions and recommendations (p. 66-67) without comment.
It is also disturbing that on p. 27 and p. 61 it refers to the informational SOP as an "SOP for operating the Coralville Dam in the event of flash flooding". This is in direct contradiction to p. A-12, where it is clear the SOP has no operational aspect as yet. The importance of the distinction between information and operation was underscored in 1993 when the high water mark in Iowa City was reached on August 10 precisely because of the lack of a written contingency plan. The ability to control the water was not an issue on that day as the lake was more than 5 feet below the spillway. Real time information was also not an issue as the dam manager (John Castle) was well aware of the 4 inch rainstorm that began at 11 pm the previous evening. However, having no written guidelines as to how to act, the dam gates (which were wide open at the time) were not lowered until after 5 am, too late to diminish significantly the crest of the flood in Iowa City.
A good faith effort using available methods to perform a serious study has not been made by the Corps. They have violated their own procedures for public review and for study methodology. Their incomplete and inconsistent conclusions are therefore inaccurate and should not be taken seriously, but rather as an indication of their abilities to carry out a serious study.
The standard procedure for the Corps is to include public review as an Appendix in a report. We are therefore taking the initiative to distribute this response to the distribution list given in Appendix D (i.e., to those noted as having been sent a copy of the Report). In the December newsletter, it was announced that a summary of the report will also be posted on the Rock Island website, with the address being furnished with the announcement of report completion. We will also make this response available through our "Iowa River and Coralville Dam Information" (http://www.physics.uiowa.edu/~cnewsom/iowa_river) site.