ESHMC Meeting Notes September 21St and 22Nd, 2010

Total Page:16

File Type:pdf, Size:1020Kb

ESHMC Meeting Notes September 21St and 22Nd, 2010

Final

ESHMC Meeting Notes September 21st and 22nd, 2010

Item 1 - Introductions were made, and an attendance list was circulated. The following were present at the meeting:

- David Blew - Bryce Contor - Rick Raymondi - Allan Wylie - Sean Vincent - Mike McVay - Jennifer Johnson - Harvey Walker - John Lindgren - Chuck Brockway - Willem Schreuder - Greg Sullivan -Hal Anderson -Lynn Tominaga -Rich Rigby*

*Present but did not sign attendance sheet.

John Koreny, Chuck Brendecke, Rick Allen, Lyle Swank, and Stacey Taylor joined the meeting via Polycom.

Rick Allen offered comments on pages 9 and 10 (Item 12) in the minutes of the June 24th & 25th meeting.

Item 2 – Mike McVay summarized what has been done to update the calculation of non- irrigated lands recharge (NIR) for ESPAM version 2. Various problems and concerns with the current approach were discussed, and Rick Allen provided explanations and insight into the data and calculations. He suggested a quality control review of the precipitation data and thought there may be some weakness in the ET Idaho information. Rick and Mike agreed that the calculation of infiltration in dormant turf may not be a good proxy for what is happening on the ESPA. They both recommended that the method used in ESPAM 1.1 to derive NIR also be used for ESPAM 2, and that Mike and Rick would continue to work on developing a new method. Rick added a concern that variable textures in lava rock could impact residual moisture evaporation, and that new calculations would be needed to take this into account.

Willem Schreuder wanted to know the significance of not being able to derive better or more accurate numbers for NIR. Bryce Contor said that most NIR occurs on lava rock, and he suggested investigating BSU’s capacity to assist in the calculations using the VIC model. Willem asked about the premise of the calculations in the VIC

1 Final

model, and Bryce was not sure. Willem asked the committee what is the amount of recharge to be expected on lava rock. Rick Allen responded that at least 50% of the precipitation on lava rock percolates, and he recommended investigating the evaporation potential from lava rock. He went on to say that it is good to compare independent methods such as the simulated time series weather data that is an input to the VIC model that BSU is using. He concluded that the same precipitation data that is incorporated into the METRIC model be used for NIR calculations so that the two systems are consistent.

Chuck Brockway asked what the input volume is for NIR, and Bryce responded approximately 700,000 AF. Willem expressed the concern that we are working with a large recharge term in the center of the ESPA that PEST uses to adjust the distribution of transmissivity and water levels. Willem then asked how the NIR calculations affect the temporal distribution of recharge, and Bryce responded that the ESPAM version 1.1 and the ET Idaho algorithms were about the same. Chuck Brockway then said that he agreed the committee should go back to the ESPAM version 1.1 method for now. Bryce added that he thought there would be value in pursuing a refined calculation because of the way ET Idaho treats the timing of precipitation.

Item 3 – Mike McVay then presented work he had completed on the synoptic ground water level measurements. He began by reviewing and refining the work done by Nathan Erickson for 1980, 2001, and 2002. He developed a refined database of wells used in development of the potentiometric surface maps from the synoptic water level measurements. Mike’s database is transparent so all users can see his approach. He used survey data and the 10 meter National Elevation Dataset (NAD) to obtain elevations. The database includes water level data, well logs, a set of data and logs used in the contouring, elevation logs, and a set of wells removed. He contoured the data on 50 ft intervals, removed perched water levels, and pulled out other “bull’s eyes” or anomalies.

Chuck Brockway asked what the width of the band was for the synoptic analysis, and Mike responded 2 months, but that most data fall within a couple of weeks. Willem asked if the wells removed for the maps were consistent, and Mike said yes. Hal Anderson asked what the criteria were for the bull’s eyes. Mike said that it usually was an anomaly created by one well, but it could be several wells, and that he tried to obtain the picture of regional ground water flow. Mike showed a PowerPoint of the wells used to construct the contour maps. Willem asked if Mike incorporated a “drift” function in the construction, and Mike said no. Willem then suggested that Mike consider a first order drift function especially for the upper end of the aquifer.

Mike then showed the water level change maps and explained his methodology in developing them. Although he used filtering in the development of the potentiometric surface maps, he did not carry the filtering into the change maps. Bull’s eyes were shown including one in the Oakley fan and one near Wendell. Chuck Brendecke asked what the cause was for the bull’s eye in Wendell, and Mike

2 Final

did not know. Chuck Brockway asked if Mike looked into the construction of the well near Wendell, and Mike said he did not. Willem suggested that Mike look at the construction and the hydrograph for that well. Mike said that he did look at the hydrograph for the deep wells in the Oakley fan, and the data showed a long term decline. Mike added that the fall synoptic and change maps were much more complicated that the spring maps. Chuck Brockway asked if Mike performed volume calculations, and Mike indicated that he would. Bryce recommended that he use the calibrated model storage coefficient, and Willem said use an approximation for this term. Chuck Brendecke recommended that IDWR perform volume changes for all increments when synoptic measurements were taken.

Lyle Swank asked why the water level change map for the period from fall 2001 through fall 2008 showed an increase, and Bryce responded that it could be from the increase in precipitation in 2006 and in 2008. Lyle indicated that a severe drought was experienced in the fall of 2001, and that recent years were more stable. He agreed with Bryce. Chuck Brendecke asked if the fall of 2001 was not a good starting point for a change map, and Lyle said that severe drought was not a good starting point. Bryce said that the data is limited by when the synoptic were funded.

Chuck Brendecke asked about the wells in the synoptic database. Mike said that the measurements were performed by IDWR, the USGS, and consultants. There are all types of wells in terms of use, and it is hard to determine ownership. Willem asked if this well database is a superset of the National Water Information System database (NWIS), and Allan Wylie responded yes plus other wells. Bryce asked why the Water Management District wells were not included, and Mike said that there was redundancy in those wells.

Allan Wylie said that in ESPAM version 1.1, the wells that Nathan Erickson used for the development of potentiometric surface maps were based on the synoptic measurements and were then used for model calibration. Willem commented that we calibrate to the well data not the maps. Allan responded that for any well that Mike includes, the data goes into calibration. Pumping or recently pumped wells are excluded. Willem said that where there only a few wells that each measurement makes a big difference in calibration. Chuck Brockway commented on the representation of the change in the Oakley fan area. Mike indicated it was two wells that were creating the severe drawdown contours. Chuck questioned whether it is representative of the change in this area. Mike conceded that it was very difficult to contour the area. There are faults, and the geology is complex. Chuck Brockway also pointed out that the southwest part of the A & B Irrigation District shows a large decrease. Willem said that it is not as big as the negative change in other areas. IDWR agreed to post the database that Mike used for the maps. The ESHMC agreed that the water level calibration targets for ESPAM version 2 should come from the wells used in the change maps.

3 Final

Item 4 - Jim Brannon briefed the committee on his progress in reviewing the MKMOD processing tool for ESPAM version 2 input data files or datasets. The programming for MKMOD consists of approximately 1300 lines of code in the PERL programming language. Jim believes the code is very reliable but different than other languages that he is familiar with. Bryce pointed out that MKMOD is being used for the first time by the ESHMC in the calibration of ESPAM version 2. Jim went on to say that he has been going through the programming logic with a high level of interest and has been concentrating on the important parts and components. He is also creating diagrams of what the tool is doing. Finally, he is putting the datasets to tests to ensure their accuracy and checking the logic to make sure all cases are handled. Jim reported that his effort is in progress and that he has seen no problems or anything suspicious or unusual. He had not performed any data testing yet. Finally, Jim recommended that the committee allow him time to finish the high level diagram for review and comparison to the conceptual model.

Allan Wylie asked if Jim was able to look at soil moisture, and Jim said yes and there is a block of code (about 20 lines) within the On-farm budget. The On-farm budget will be the heart of the high level diagram. Bryce pointed out that there are changes in the NIR component that should be considered during the comparison to the conceptual model. Willem said there is not anything to compare the new code with unless one performs hand calculations.

Willem went on to explain aspects of the PERL and PYTHON codes and what can be done to shorten PEST runs. He said the PYTHON programming may not result in faster runs than PERL because there are lots of nested do-loops. Willem added that the output calculations and error trapping reduce memory usage. Willem asked Jim if the core section of MKMOD was converted to PYTHON, would that help make it easier to read the code. Jim said he was comfortable in PERL. Rick Raymondi asked Willem if he will continue his effort to write the code in PYTHON, and Willem said yes. Willem said he is trying to solve the performance bottle neck. He said PYTHON is less complex and easier to read than FORTRAN and that he hopes his work will help the overall effort.

Jim asked if a graphical diagram is agreeable to the committee even though his final product is taking a little longer than he thought, and Allan and Rick Raymondi said yes. Willem said the committee needs to understand the algorithms better and there are lots of devils in the details. Jim said it doesn’t look like that there are significant changes in MKMOD versions 3 and 4. Jim asked Allan if he will use PYTHON, and Allan said yes if it is faster. Jim ended the conversation by saying that he hopes the flow chart will help anyone who wants to rewrite the code, and Willem added that it will be good for documentation.

Item 5 - Jim Brannon also updated the committee on his effort to complete the quantification of spring discharge at the National Fish Hatchery (NFH). Chuck Brendecke had made Jim aware of a data problem in 2005, and Jim said the errors resulted from a construction bypass at Main Spring. Jim said that the NFH had collected data during

4 Final

the bypass effort, and the correct data had been inserted to fix the database. He showed the committee the web site and where the new data had been updated and assured the committee that all graphics had been updated. Rick Raymondi asked if the discharge to the slough had been added to the database, and Jim said he had not yet heard from the NFH regarding if the data exist.

Item 6 - Mat Weaver briefed the committee on gage installations completed at the Bridal Veil spring and the ABC complex at the SeaPac facility. Mat show photographs of Bridal Veil springs, summarized the water rights, and discussed the historical flow data. The springs discharge from 2 culverts into the hatchery raceways. Doppler flow meters were installed at the outlets and the stage is read in one channel. The flows are combined. Mat reported that it is difficult to provide a QC check on the flows, but a location above the falls was measured with good accuracy ( within 2 – 4.2%) and shows good correlation with the combined flow meters. Mat said that more work is needed to verify a good QC measurement location.

Mat also mentioned that the installation at the ABC spring complex is complete with a broad crested weir installed last January. Installation braces were used to tie down the weir. Mat showed raw and corrected flow measurements from the ABC springs and explained the correction used for the discrepancy between the weir and the flow tracker measurements.

Item 7 - Allan Wylie presented a discussion of the progress in calibrating model spring flow targets. He indicated that in the NFH model cell there are 17 springs. Some are NFH springs (9 of 17), and some emerge at SeaPac. Allan showed the range in elevation data for the springs from Covington and Weaver (USGS, 1990). In an adjacent model cell, there are 8 springs in the Thousand Springs complex, and two are used by the power plant with the addition of flow from Sand Springs which emerges in an adjacent model cell. The flows were not separated in the power consumption data received from Idaho Power. Allan also showed the range in elevation of these springs. Allan pointed out that in the ESPAM version 2 cell output, there are three drains available at different elevations, and the conductance for each one of the three is calibrated separately. He added that MODFLOW outputs the total cell discharge of for the three drains added together, thus, PEST calibrates total cell discharge, not individual spring discharge. Allan proposed three options:

1) Scale the data up to represent all the springs in the cell; 2) Modify MODFLOW to output drain discharge not discharged by model cell; and 3) Develop utility to compute discharge by drain cell in selected model cells.

A debate followed. Willem asked if we have the resolution to distinguish between low, medium, and high elevation springs. Allan indicated it becomes complicated because the SeaPac springs discharge into 2 different model cells and we are scaling up the discharge based on Covington and Weaver data to represent the output from all springs in the cell. Chuck Brockway thought that the Covington and Weaver measurements were “random” estimates and that we need to do better in calibrating

5 Final

spring discharge. Greg Sullivan said we have the steady state contribution from the river gages for calibration. Allan Wylie said we don’t have enough data to do a mass balance in any reach because there is so much discharge into talus and directly to the river. Allan then said we need to scale up to account for additional springs. Willem suggested adding the three drain elevation discharges to get the cell discharge. Chuck Brendecke said take the three drains and apportion spring discharge by elevation, and he indicated it could be a mixture of different springs from different complexes. Jim Brannon asked Allan if option two was difficult. Allan said he did not want to modify MODFOW, and that he liked option three better. Willem said that option two could be done if we used the MODFLOW streamflow routing package. He then asked Allan if there are data for Majic Springs, and Allan said we have data for Thousand Springs, the NFH, but not SeaPac.

John Koreny asked Allan to explain the first option. Allan said for example, if we have a transient target for the NFH, and Covington and Weaver have NFH and SeaPac discharging each at 50 cfs, the NFH discharge represents half of all spring discharge in a cell. Therefore we increase the target discharge by a factor of 2. Then the total cell discharge is scaled up. Allan went on to say that, in the end, PEST will match both individual springs and the reach.

Allan asked the committee if everyone is comfortable with scaling up the cell discharge. Willem said that unless we can distinguish between high and low elevation springs in a cell, he may not be totally comfortable. Allan briefly discussed the A, B, and C spring categories and that the lowest weight is given to the C springs. Chuck Brockway said that he still would like to see an emphasis for modeling to individual springs, but because of data limitations, he suggested option 1 as the best. He recommended further exploring option 2. Willem said that this appears to be a data decision, not a model decision. Chuck Brockway asked if option 2 would work on any cell. Allan responded that he didn’t think we have the data, but that he could perform a run to extract that information. Chuck Brendecke recommended option 1, and the committee was in agreement. Allan said that he will show the cells presented and how they will be scaled up in the next meeting. He said that in all other cells, the complex is treated as one discharge so there is not an issue similar to these two cells.

Item 8 – Allan Wylie presented the results of his most recent model calibration run with the A, B, and C springs as targets. He provided his assumptions, the data sets used, the water budget (P100827A) and MKMOD version (MKMOD4.exe), and the On-farm adjustable parameters incorporated. Allan then went through the results for transmissivity, storage, riverbed and drain conductance, perched river seepage, tributary underflow, non-irrigated recharge, ET for lands irrigated by sprinkler and gravity using surface water, the efficiencies for lands irrigated by sprinkler and gravity using surface water, efficiencies by surface water irrigated entities, canal seepage, DPin and DPex results, a comparison of measured vs. modeled results for transient head calibration for wells throughout the ESPA, the modeled vs. measured gains for Snake River reaches and spring reaches, the modeled vs. measured

6 Final

discharge comparison for 11 individual springs, and finally the modeled vs. measured discharge comparison for the Group C springs combined.

Allan suggested taking out the model cell where Lower White Spring emerges since that discharge is likely related to a block of Quaternary basalt that is separate from the ESPA. He mentioned that the tributary underflow for the Henry’s Fork and the Camas and Beaver Creek basins bumped the upper limit of allowable flow volume, and the results for NIR did not make any sense. The discussion then focused on ET for lands irrigated by sprinkler and gravity using surface water, and the efficiencies for lands irrigated by sprinkler and gravity using surface water, and what can be done with PEST using regularization. Allan commented that is seemed that for a greater amount of freedom that PEST was granted, the results seemed less believable. Greg Sullivan suggested that the parameters should be set instead of giving PEST the freedom. Chuck Brockway and John Koreny said that they agreed with assigning parameters. Willem commented that what Allan did was a valid test by letting PEST set parameters, but now the runs should be performed with PEST locked down. Allan then said that the efficiencies for lands irrigated by sprinkler and gravity using surface water did not come into play in the calibration runs and that the efficiencies should be adjusting other factors.

Chuck Brockway made a general comment that the committee should zero in on calibrating the model. Bryce agreed. Mike McVay asked if PEST was finished in terms of calibrating, and Allan said he wasn’t sure where PEST was going. Willem said either tell PEST not to change or tell PEST why we disagree. Allan said he believed that the adjustments by PEST were done because it did not have soil moisture. The committee then agreed that IDWR should make a calibration run with a fix on the On-farm algorithm for soil moisture. Chuck Brockway added that PEST could be allowed to adjust parameters in June through August, and Willem said he will check into it with John Doherty. Allan said he thought that in terms of efficiency, a soil moisture number should be selected and set for the calibration run, then involve the soil moisture algorithm and let PEST run. Willem agreed because he thought there was a need to determine what parameters have the largest impact. Allan suggested that the parameters were transmissivity and river bed and drain conductance. Willem asked how much surface water is being applied through sprinkler, and Bryce said a lot. Bryce then said that ET is the greatest water budget parameter, and canal seepage is the next largest, and it is fixed for the coalition.

Allan then commented that the model reach gains in the early and late data period are out of phase for the Near Blackfoot to Neeley reach. He added that the model is over- shooting the Blue Lakes spring discharge for the recent data. Allan pointed out that a large portion of the Crystal spring discharge is unmeasured. Chuck Brockway said there have been measurement problems at Crystal. Allan went on to say that the discharge from most springs is over predicted for the last couple of years of the dataset. Sean Vincent asked why PEST is over estimating the seasonal amplitude of some springs. Allan said he observed that if PEST matches the head in wells, it over estimates the springs, and that he thinks that it the fixed transmissivity assumption is

7 Final

part of the problem. Allen added that there is an excellent match for the spring discharge at Rangen. Chuck Brendecke asked if Rangen is the highest elevation spring, and Allan said that he thought so and would check to make sure. Allan noted that there are some unexplainable spikes in the Malad River discharge, and Dave agreed to look at the data.

Item 9 – Allan Wylie provided an On-farm subcommittee report. He said that the committee decided to fix canal seepage for the coalition, and that there would be manual adjustments for other surface water entities. Bryce added that there is a design document and two memos that discuss how canal seepage was developed for the other entities. He said that in the On-farm budget, that there is no difference between canal seepage and deep percolation. Greg Sullivan said that canal seepage is a knowable number for all canals and that the committee could obtain such data from recharge efforts. Allan said that the scaling factor for canal seepage will be started at 1 and adjusted between .95 and 1.05, but PEST would be told to keep adjustable scaling factors as similar as possible. John Koreny said that letting PEST figure canal seepage is dangerous; he said it is a physical process, and it would be better to get canal lengths and ask for seepage rates from canal managers. Chuck Brockway said a best estimate should be made and canal seepage should be fixed at that number. Chuck Brendecke said that there is not a big difference between the .95 and 1.05 scaling factor. Allan was asked how the canal seepage was calculated, and he said Bryce determined how much water was required to satisfy crop consumptive use and have some water left for deep percolation, the remainder was assigned to canal seepage. The committee agreed that the scaling factor should be set at 1 and PEST should adjust between 0.95 and 1.05.

The next subject discussed in the On-farm subcommittee report was irrigation efficiency. Allan said that it is a function of crop mix, irrigation method, and soil type. He went on to say that the quality of the crop from alfalfa and grass are not sensitive to drought and can be grown by deficit irrigation, potatoes are sensitive to drought and cannot be grown by deficit irrigation, and grain is somewhere in between. Allan added that Bryce has provided a spreadsheet with various crops to help determine irrigation efficiency. Allan said that the subcommittee debated letting PEST adjust efficiency, starting at .8 and allowing the adjustment to run between .75 and .9. Allan said that he thought efficiency should be fixed, and Willem agreed except for Dietrich and Richfield. Bryce said the soil algorithm should take care of those areas. Chuck Brockway asked what if we fix efficiency for all areas, would it be a big mistake. Willem said not a big mistake at first. Greg recommended that we fix them all, but that it did not have to be the same number. Willem conceded that these are small knobs, and we could adjust the values when calibration is fine tuned. The committee decided to fix irrigation efficiency in model calibration but that it did not have to be the same number for all entities, as a starting place the committee decided to fix gravity at 0.8 and sprinklers at 0.85.

Allan then said that there is a disconnection between return flows calculated by the On-farm water budget and the measured returns and gave the Heise to Shelley reach

8 Final

where returns are a significant percentage of the gains as an example. Allan reported that the subcommittee decided to tightly constrain the On-farm parameters (DPin and DPex) so that calculated flows better reflect actual return flow measurement data or that returns needed to be a calibration target. The committee agreed that returns need to be a calibration target and Willem agreed to modify MKMOD to output returns. Finally, Allan said that the subcommittee discussed the effects of soil moisture and decided that the soil moisture algorithm should be included in the On-farm budget. Item 10 - Bryce Contor reported that the ESPAM version 2 water budget was completed, and all required input to MKMOD had been prepared. He indicated that there are 15 input files, and briefly discussed the files.

Item 11 - Bryce provided the committee an update on the water budget training that is being planned for committee members and other interested parties. He first gave an overview of the modeling process and showed how PEST modifies parameters to calibrate the model and create calibrated parameters. Then he discussed the purpose of the training was to provide full transparency of the water budget, show how to build model input files, how to run MKMOD and MODFLOW, and how to summarize output data. Chuck Brendecke asked who would receive the training. Bryce responded the ESHMC, and Chuck Brockway added that consultants would want the training. Bryce added that a better format was needed for the input files, and that a decision was needed on format before going forward with developing new professional tools.

Harvey Walker expressed his desire that the model and tools should be user friendly. Bryce said that tools have been built for ESPAM version 1.1 and for CAMP, and that he believed useable tools can be built. Greg Sullivan expressed caution that tools cannot be built for every potential use. Bryce concluded the discussion by saying that two levels of training are needed, one for the model, and one for tools.

Item 12 - Rich Rigby provided an update on the ESPA CAMP Implementation process. He said that the committee is trying to create areas of common interest with smaller entities working together in a “grass roots” approach. Rich added that the next meeting is scheduled for October 14, 2010 and that progress is necessary. He said that they have a few years to accomplish goals and that there is interim funding for committee proposals. The Idaho Attorney General has indicated that any money spent on projects is in the form of a loan. The projects being considered are the NSCC reuse, the Egin Lakes recharge, Idaho Irrigation District recharge, and AWEP projects.

He indicated that model runs show that 49% of the NSCC reuse will benefit the Buhl to Thousand Springs reach and 26% will benefit the Devil’s Washbowl to Buhl reach. Next he said that the Egin Lakes recharge will be undertaken by the Freemont- Madison Irrigation District and that engineering work is being done to expand delivery capacity by 30,000 A-F/yr and to create additional canal system improvements. The total cost is projected at $440,000, and most of the benefits would accrue to the Ashton to Rexburg reach, with 20% accruing to the Near

9 Final

Blackfoot to Neeley reach. The model predicts that 30% of the recharge will remain in the aquifer after 5 years. Rich indicated that the projected costs for expanding and improving the facilities in the Idaho Irrigation District to accommodate recharge would be about $69,000. This would include improving the capability to monitor and measure recharge which would primarily benefit the Shelley to Near Blackfoot and Near Blackfoot to Neeley reaches. Finally, Rich discussed the AWEP projects that receive both Federal and IWRB financing and involve contracts with individual land owners or spring users.

Rich then informed the ESHMC that the IWRB wants a modeling committee to assist in decision-making by providing a rigorous review of aquifer impacts by proposed Board actions. The committee will use ESPAM version 2 when completed, and Rich noted that the monthly time step is a desirable aspect of the model. Rich commented that since the Near Blackfoot to Neeley or the Near Blackfoot to Minidoka reaches are implicated in the Delivery Call, it is important to determine how much and the timing of the benefit that accrues to those reaches. Rich said that the committee would include Chuck Brockway, Chuck Brendecke, Roger Warner, Allan Wylie, and an individual from Idaho Power Company and that committee consensus is not always necessary.

Chuck Brockway asked what the Bureau’s capability is for modeling in the upper Snake River basin, and Jennifer Johnson said that for ground water it is ESPAM and for surface water it is MODSIM, but there is an initiative to convert to RiverWare. Rich Rigby said that IDWR is also looking at RiverWare. Chuck Brendecke asked who will perform model runs, and Rich said that will be the responsibility of the Department. Then Chuck Brendecke asked if the ESHMC would inform the “model process” of the Board committee, and Rich said that has not been defined and that the Board wants a comfort level that benefits are considered for projects. Chuck Brockway said that the ESHMC is technical, its function is to provide expertise and determine impacts from modeling, and it has been chastised for delving into the administrative aspects of issues. Chuck then added that this new committee cannot be divorced from administration. Bryce asked who will fund the committee, and Rich said the constituents. Bryce then cautioned that he who has resources will benefit, and Rich said he is aware of that potential.

Chuck Brockway then asked Rich to explain the issues with incidental recharge. Rich said that the surface water users provide benefits by their operations via incidental recharge. Part of this group feels that the price of their participation in CAMP is offset by their incidental recharge and that they should be compensated for this contribution to ground water. Other surface and ground water entities disagree, and the result is that a new funding approach needs to be developed.

The ESHMC set the next meeting dates for November 15th and 16th, 2010. (Note – the dates were later changed to November 22nd and 23rd, 2010.)

10 Final

Item 13 - Allan Wylie initiated a discussion on the subject of model validation. He presented two options:

1) Leave a few years out of the calibration dataset to see how well the model predicts those years. 2) After a few years, conduct a scenario incorporating a few years not in the dataset.

Allan added that in the past, the paradigm was that you leave some data out. Now, model validation sounds good but is almost never done. Chuck Brockway said that it is good to do, and seldom done, but if you undertake validation you have more confidence in the model. Then Chuck said that the current model is calibrated 1980 – 2002 and asked what if we used 2003 to 2010 for verification to see how the model predicts what actually happened. John Koreny asked what the time period is for the new model, and Allan said 1985 through 2008 and asked Allan to explain the difference between calibration and validation. Allan said that in calibration, the actual dataset is compared to model output, and parameters are tweaked to improve the match; whereas, validation is done after calibration to compare output to recent data, but you don’t adjust the model.

Willem thought that the model should be built with all of the data to obtain the best calibration. He added that for verification, you could ask the model to make a prediction that we know the answer to and then observe if the model predicts accurately. Chuck Brockway said that you could try to see if the model predicts a new stress it hasn’t seen. Greg Sullivan and Chuck Brockway agreed with Willem that the committee should use the full dataset to calibrate the model, and Chuck added that you then investigate whether the new model will simulate what happened. Chuck Brockway said that within the ESPA conditions are not the same because the USGS data show declines, and when we attempted to validate the springs, the model did a poor job of predicting the decline. Bryce Contor said that the spring dataset was flawed.

Greg Sullivan said if we performed validation with the most recent dataset, and if it goes poorly, we just calibrate to the new data. Greg concluded that we should just calibrate the model and skip validation. Willem again said use all the data to calibrate and find another way to verify the model. Chuck Brockway said people normally don’t validate the model because it is like throwing out the data, but it would be nice to validate the model in some way. Willem said that if the model calibrates well to spring flow, then it is good. Then you look at a totally new large stress to see how it behaves.

Allan Wylie said that with ESPAM version 1.1, we have a calibrated model. We asked the model what would happen if the drought continued. The model results made sense in our opinion. Then we asked the model what would happen if the drought ended. The model said the aquifer would recover but it would take a long time. The committee felt that it made sense, and we were able to tell constituents that

11 Final

we liked the model. Chuck Brockway asked Allan if he compared 2008/2009 with the drought scenario. Allan said no, what we decided to do is recalibrate the model with the recent data.

Willem said that through calibration, you try to determine if the model did the right thing, then you formulate a question to see if the model behaves appropriately. Sean Vincent said that we should try to maintain the status quo and see how the model predicts. Willem said rather you need to impose change to see how it behaves. Sean then asked if we can check if we are at equilibrium with no change in stress. Willem said equilibrium is whether inflows equal or balance outflows.

Chuck Brendecke asked if we can go backwards where we were in the 1970’s, and Willem agreed with the idea if spring flows and water levels are available. John Lindgren asked if we can take the first 4 years of the model period out and see if we simulate accurately. Allan said then we could see what the 1970’s or 1960’s would have looked like. Jim Brannon said he agreed with Chuck Brendecke, and he would like to see if the mode reacts to stresses in the way we said it would. Dave Blew asked if we can see if the model predicts a 1% decline year after year. Willem went back to his contention that the only way to validate the model is to make big changes and see how it does. Chuck Brockway asked if we could see what would happen if the A & B Irrigation District wasn’t there. Greg Sullivan said that the validation results do not have to be 100% accurate because we still can learn from the exercise.

Sean Vincent said that the issue is equilibrium, and the only way to verify it is at the end. Chuck Brockway said we still have residual impact from the past. Willem said that the question is can we agree on what the stress should be. Chuck Brockway said that the model says we are in equilibrium, and Greg said maybe with respect to pumping. Chuck Brockway agreed with Greg and said that maybe there are other reasons besides pumping that result in non-equilibrium. Allan said that there were people that didn’t understand what we meant by equilibrium. Harvey Walker said there were people that don’t agree or don’t trust the model.

John Koreny said there is lots of work ahead and that a well planned out calibration is the way to go. He added that we should not leave out anything in the effort. Greg Sullivan said so the choice is whether to leave out or leave in any data. Sean said maybe we should do both. Chuck Brockway said there is public relations work to do regarding what the model can and can’t do. The committee agreed that no data should be left out of calibration for later validation, although Sean Vincent had his reservations.

Sean Vincent cautioned that there could be problems with not doing validation. Allan said we should do scenarios to accomplish validation. Sean said we should make model runs without tweaking parameters. Willem said we have to give it a problem not seen before. Chuck Brendecke said we should only use data through 2000 exercising the model without the stress of drought. Dave Blew said a good test is to see if the model predicts subtle changes because that is what we deal with regarding

12 Final

spring discharge. Greg Sullivan thought that there is a problem with expectations because the changes in ET and diversions are large. Chuck Brockway said spring users are getting into a risk analysis approach, and the owners need to know confidence limits. Willem said the problem is that you tell spring owners that if everything else is the same, you should have a 2% increase in flow, but everything else is not the same. Chuck Brendecke concluded the discussion saying that the ability to predict is a function of knowing all other influences, not the ability of the model.

Item 14 - Sean Vincent led a brainstorming session of the design features of ESPAM version 3. He reviewed the previous brainstorming meetings, and then led the committee in a listing of current model uses. Bryce said that the model was also used as a predictor, Sean said to run scenarios, Rick Raymondi said to predict flow to spring reaches, and Chuck Brockway said to perform multiple transfers. Sean then showed a comparison of ESPAM versions 1.1 and 2 attributes. Willem made the point that we should not refer to the ESPA as a confined system. Then Sean reviewed what was agreed upon in previous meetings, and he indicated that stakeholders want predictions at a local scale. Chuck Brockway asked the question regarding a Class A spring target and whether we feel that ESPAM version 2 can be used directly to determine impacts to that spring. Allan said probably. Greg said we don’t have a completed ESPAM version 2 yet, and Allan said that is why I said probably. Sean summarized the point by saying that it appears that the goal is at the scale of a spring.

Sean continued the discussion of model uses and said that one requirement is to facilitate an uncertainty analysis and provide additional tools for certain uses. Bryce said that more effort is needed in the conceptual model to include the geologic underpinnings of the Rexburg Bench. Chuck Brockway said the connection between the Bench and the South Fork of the Snake River may not be accurate because it doesn’t reflect faulting that may impact that connection. Allan suggested that the new model include the Menan gage. Lyle said the gage has been in operation from 2001 to the present. Allan said from the combination of the Heise and Menan gages could help PEST in determining impacts to the South Fork. Chuck Brockway said the geology should be reviewed. Willem said that the model can always be improved to represent the system better, but he asked where we would get the data. Allan said that he recommends that we find unused wells on either side of the suspected fault and equip the wells with transducers. Chuck Brendecke said that there are numerous places to improve the model representation of the system, and he recommended a review of well logs and preparation of fence diagrams. Chuck Brendecke went on to say that there may be places where 2 model layers is appropriate and recommended that the committee look at this for ESPAM version 2 design components.

Jim Brannon suggested that the Department look at data management systems and develop a generic layer to make more data available to those who need it. He suggested a web-based data distribution system. Chuck Brockway said that he would like to query water rights along with the hydrology. A discussion followed regarding the deficiencies of the Department’s databases.

13 Final

Chuck Brendecke said he would like to be able to select wells to retire based on where they are rather than the priority date and gave diagrams of capture areas an example. Sean suggested that what Chuck Brendecke is requesting comes under the area of model tools. Bryce said that tools to assess management actions that are targeted to individual uses would be beneficial.

Sean began a discussion of model uncertainty. Greg said that one number is not always good and that it varies with location and type of effort. Chuck Brockway said there should be an error bar on spring flow and that model simulations of the impact to water levels is desirable. Allan that this type of simulation is doable and similar to modeling the impact to a spring but not easy and requires a lot of CPU. Bryce said the water level response to a change in stress should be a model use requirement. Dave Blew asked if the uncertainty will change with the magnitude of stress. Allan said that it depends because with distance there is more uncertainty, but if you pump a well harder, the impact will be more pronounced. Jim Brannon said uncertainty has to be at the resolution of the model, and Willem added that uncertainty is also determined by the resolution of the data. Allan Wylie said that for predicting water levels, we need to keep in mind that the model is 2D.

Chuck Brockway said that a problem in the CAMP process is defining all the beneficiaries, and he gave the domestic water users as an example. Chuck then asked if we should consider contaminant transport in the new model. Finally, Chuck said we need to look at the Oakley fan and asked if we are satisfied with the Shelley to Near Blackfoot reach and the changes in the hydraulic connection with water levels. Willem said that river stage is a crude representation in the model and asked if we can represent secondary effects of actions.

Sean then asked if the committee would consider tools for the layman. Bryce said that the committee might consider something like a version of the transfer tool for the layman, perhaps a tool for the consultants, and one or two other tools for the public. Dave Blew said the surface and ground water interactions depend on the tool that is used.

14 Final

DECISION POINT SUMMARY

The following was agreed upon:

1) The method used in ESPAM version 1.1 to derive NIR would also be used for ESPAM version 2. The data set would be extended, and PEST would be allowed to adjust the NIR input. Mike McVay and Rick Allen would continue to work on developing a new method, and when it is complete, it would be inserted into ESPAM version 2. 2) IDWR will perform volume calculations for the change in water levels demonstrated by the maps produced by Mike McVay. IDWR will also perform the calculations for all increments when synoptic measurements were taken. 3) IDWR will look at the construction and the hydrograph for the well near Wendell that is responsible for an anomaly in the water change maps. 4) IDWR agreed to post the database that Mike McVay used for the potentiometric surface and water level change maps that were based on synoptic ground water level measurements. 5) Jim Brannon will finish the high level flow chart which will facilitate code review and comparison to the conceptual model by other members of the committee, assist anyone who wants to rewrite the code, provide documentation MKMOD. 6) The committee agreed that IDWR should proceed with Option 1 - Scale the data up to represent all the springs in the cell. 7) IDWR agreed to show the three cells presented by Allan in the September 2010 meeting and how they will be scaled up in the next meeting (November 22nd & 23rd, 2010). Note - Idaho Power will take part in the presentation. 8) The ESHMC agreed that the water level calibration targets for ESPAM version 2 should come from the wells used in the change maps. 9) The committee agreed that IDWR should make a calibration run with a fix on the On- farm algorithm for soil moisture. 10) Dave Blew agreed to look at the Malad River discharge to determine if the spikes in the data could be explained. 11) The committee agreed that the scaling factor used to adjust canal seepage should start at 1 and only be allowed to adjust between 0.95 and 1.05. 12) The committee decided to fix irrigation efficiency in model calibration but that it did not have to be the same number for all entities. Starting values are 0.8 for gravity and 0.85 for sprinkler. 13) The committee set the next meeting dates for November 15th and 16th, 2010. (Note – the dates were later changed to November 22nd and 23rd, 2010.) 14) The committee agreed that no data should be left out of calibration for later validation, although Sean Vincent had his reservations.

15

Recommended publications