ESHMC Meeting Notes September 24Th, 2013
Total Page:16
File Type:pdf, Size:1020Kb
Draft
ESHMC Meeting Notes September 24th, 2013
Item 1 - Introductions were made, and an attendance list was circulated. The following were present at the meeting:
- Rick Raymondi - Allan Wylie - Harvey Walker - David Blew - Jennifer Johnson - Jennifer Sukow - David Hoekema - Sean Vincent - Chuck Brockway - Jon Bowling - Neal Farmer*
* present but did not sign the attendance sheet
Bryce Contor, Greg Sullivan, Chuck Brendecke, Dave Colvin, Willem Schreuder, Lyle Swank, Stacey Taylor, and Sophia Sigstedt joined the meeting via polycom.
Item 2 - Rick Raymondi began the meeting by discussing items of committee business. The first was the recent IDWR reorganization with the most significant change being the new Deputy Director position now held by Mat Weaver. A current Department organization chart was provided to the committee. Next the committee was reminded that letters were sent in August to Senator Crapo and Congressman Simpson requesting their support for the planning and funding of Landsat 9. Finally, the committee was notified that IDWR would be jointly holding a Drought Forum with the Pacific Northwest Climate Impacts Research Consortium on October 18th, 2013. At the conclusion of business items, Chuck Brockway asked that Eric Powell be added as a member of the ESHMC.
Item 3 - Bill Kramber provided the committee with a METRIC progress report. Bill gave credit to Dr. Ricardo Trezza, Carlos Kelly, and Clarence Robison for their work in processing ET data using the METRIC model. All three work for Rick Allen at the University of Idaho Kimberly Research Institute. Bill showed a list of years that have been processed and years that are in progress and indicated that the committee needs to decide whether 2012 or 2013 should be process next by the Kimberly Research Institute. Chuck Brockway asked how long it takes to process a year, and Bill said about 1 year. Then Bill showed where the METRIC data could be found on the ftp site in the special projects folder. He also showed a list of the years that could potentially be processed using METRIC between 1984 and 2013.
Bill then went into a detailed review of the difficulties with processing Landsat images for 1986 that are associated with cloud cover at the beginning and end of the irrigation
1 Draft
season. He showed Landsat images for paths 39 and 40 and discussed cloud masking and interpolation. Then he discussed using AVHRR satellite images for 1986 that could be used to develop NDVI ET to fill in for cloudy Landsat images. Bill said that AVHRR data are acquired every day but there is lower spatial resolution compared to Landsat images. He said that the useful data will be limited to the middle 1/3 or center of the AVHRR. Bill showed a summary chart of the % clearness for days during the 1986 irrigation season that combined the AVHRR and Landsat data for paths 39 and 40. He summarized that there are too many cloudy days and that ET Idaho data may have to be used to complete the 1986 processing.
Bill then reviewed the Landsat images for 1992, another year for which he is attempting to process ET. He showed the images for paths 39 and 40 and the cloud masks that affect the analysis. He indicated that biweekly AVHRR NDVI composites are available from 1989 to present and that he planned to use the highest quality NDVI data to quantify the seasonal ET.
The next topic that Bill discussed related to providing background information to aid a committee recommendation to select either the 2012 vs. the 2013 irrigation season to process ET. David Hokema mentioned that 2013 was probably a drier year. Bill indicated in summary that neither year is 'great', but Path 40 can probably be processed via METRIC with some supplement for both years. Path 39 will have challenges for both years, but we may be able to develop a robust means to use METRIC-MODIS and METRIC-VIIRS to complete missing periods while keeping the 30 m sharpness of Landsat. Bill noted that for 2013, both Landsat 7 and Landsat 8 data are available. Chuck Brockway recommended that the 2013 data be processed so that the METRIC data as well as the other entire water budget could be available for the next version of ESPAM.
Bill completed his presentation indicating that during the next ESHMC meeting, Rick Allen and Carlos Kelly would discuss potential improvements to METRIC including automating certain aspects of the processing and quantifying the ET measured on rangelands.
Item 4 - Jennifer Sukow discussed new gage installations that would allow winter seepage measurements on the Little Wood River. She began by briefly revisiting how the water budget is addressed in ESPAM2.1 in the irrigated areas along the Big and Little Wood Rivers. She indicated that the overall water budget calculations could be improved with the measurement of irrigation return flows and more winter flow measurements on these two rivers.
Jennifer then discussed two sites on the Little Wood River (WM#10 and WM#54) where the flows are currently measured by the WD 37 Watermaster during the irrigation season with a current meter to maintain the rating curves. She indicated that IDWR had recently equipped these sites for year-round measurements. These measurements would allow the calculation of perched river seepage during the winter in the reaches from Carey to Richfield and Richfield to Shoshone.
2 Draft
Jennifer added that if winter measurements could be obtained on the Little Wood River at Gooding (WM#17), then perched river seepage could be quantified in the reach from Shoshone to Gooding. Finally, if winter measurements could be obtained on the Big Wood River at Gooding (WM#9), then perched river seepage could be quantified between below Magic Reservoir to Gooding.
Item 5 - David Hoekema updated the committee on the recharge to the ESPA that occurred during the release of approximately 53,000 AF from Magic Reservoir between October 29 and December 7, 2012. This topic was previously discussed in the April 3, 2013 meeting. David summarized the release and indicated that some of the water flowed to the Snake River. He began a comparison of the modeled vs. observed results.
Neal Farmer said that some of the flow during the release was diverted into the Richfield Canal. A portion of this flow was directed towards Devil’s Headgate, and a portion was directed into the Cottonwood Slough. David said he would obtain the flow data for the diversion into the Richfield Canal form Neal Farmer. Chuck Brendecke asked David if he would update the recharge analysis with the Richfield data, and David said yes.
David showed the modeled head changes in the ESPA vs. the measured data. He also showed model runs of the impact on the aquifer with calculations of the volume remaining after defined time periods. Chuck Brockway commented that there are significant water level changes associated with the 32,000 AF of recharge. Chuck asked David if the 10% of recharge that remains in the aquifer after ten years is a volumetric difference, and David said yes. David also said that the added spring discharge will increase the flow in the Snake River measured at the Swan Falls gage. Chuck Brendecke asked if David could provide the modeled vs. measured data for spring discharge. David said he could once the data are available. John Bowling and Chuck Brockway expressed support for Chuck Brendecke’s request. Then Jon Bowling asked if David had looked at the gage data from the Malad River at Bliss for evidence of a change in discharge, and David said no.
Item 6 - Allan Wylie gave an overview of the prioritization of candidate improvements to ESPAM2.1. He indicated that he compiled suggested improvements from the June 2013 meeting, obtained committee feedback on the rankings, assigned scores to the rankings, and made his recommendation to committee. Allan said that during the voting, most of the respondents suggested additional improvements, but, since no one else was aware of the newly suggested improvement, the only person voting for it was the person making the nomination.
Allan showed the ranked results, and he identified the improvements IDWR was going to make regardless of the voting. He said that in the end, there were only six respondents or voters on the improvements. Allan recommended that IDWR pursue all low effort recommendations on the list with a bias toward the recommendations receiving the most votes. He added that resource intensive recommendations should
3 Draft
come from the higher scoring candidates. Bryce endorsed Allan’s recommendations. Willem said that he could easily change the coding for the names of irrigation entities. Chuck Brockway asked where IDWR would focus the updates to tributary underflow calculations, and Allan said it would be done in a piece meal fashion as new data became available.
Item 7 - Allan gave a brief overview of the MODFLOW USG, a finite-difference groundwater model that was developed to support a wide variety of structured and unstructured grid types. He said that USG was free and had undergone extensive review by the USGS. Allan said that USG works well with the existing ESPAM input files, although there were minor changes needed to the BCF and the WEL files. He also said that it contains a new Sparse Matrix Solver which provides several methods for solving equations and that it works well with PEST.
Allan said that he compared objective functions using two runs (MODFLOW 2000 and MODFLOW USG) and obtained almost the same answer, but USG produced results in about ¼ the time. Regarding calibration with USG, he said that he experienced no problems when in a fixed transmissivity mode, and it was more stable than MODFLOW 2000 in a variable transmissivity mode, although he noted that USG fails to converge from time to time. Allan showed various configurations for model layering schemes, grid options, and approaches for incorporating the connected linear network. He concluded that the switch to USG can be done, and he also felt that converting to an unconfined representation of the ESPA could also be achieved. He recommended converting to MODFLOW USG for ESPAM2.2, and further experimenting with variable transmissivity and other options included with USG.
Chuck Brockway asked if USG prevented cells from going dry. Allan responded yes. He said he currently turns that option off so that the cell is not active when it goes dry. Neal Farmer asked if USG could be used in association with his tracer tests, and Allan said yes and that it could represent linear features in the model.
Willem asked why Allan recommended USG as the model framework, and Allan said it provides a better solver and allows the user to represent variable transmissivity in the model. Allan added that he did not realize USG would be faster with the new solver. Willem agreed with Allan, but he said he is not convinced that all the bugs have been worked out of USG. He added that he is hesitant to use too much grid refinement in modeling before the bugs are worked out. Allan said he is somewhat skeptical of the “ghost node” approach that is used to interpolate heads for non-uniform model grids.
Greg Sullivan asked if we could wait and test USG at the right time. Willem said yes, and he added that USG needs to be tested. . He then said that when you represent flow across a node interface, it goes from a head perspective to a flux perspective. He added that it seems awkward, and he recommended a delay before using the model. Allan said that he was not interested in employing the unstructured grid options, but was interested in the faster and more stable solver available with USG. Willem said he doesn’t know how much more accurate the model would be using the “ghost node”
4 Draft
Allan said that so far he is impressed with the experiments that he is performing with MODFLOW USG using a uniform rectangular grid. Greg said that it seems like researchers will publish papers where they have incorporated unstructured grids. Allan agreed. Sean Vincent asked if USG employs different solvers. Allan said yes. Greg said that setting runs up with different conditions might help. Allan said that he wanted to try superposition runs. Willem said that if you stay with the current grid you should get similar answers. Allan said that if he adopts USG and incorporates the simpler model improvements suggested in the previous talk, the conversion to USG should not be contentious.
Chuck Brockway asked if there were any problems with the well file, and Allan said that the well file needed some minor reformatting and he wrote a program to accomplish that on the fly during calibration runs. The reformatting could be accomplished more efficiently with a new version of MKMOD. Willem said he had not started that process yet. Chuck Brockway asked if this new version of MKMOD would be considered a new model, and Allan said no. Chuck asked if it would be perceived as a new model. Allan said that new calibrations are considered new model versions. Chuck stated that the committee has said that it would embrace new model versions. Allan said that we hope to complete ESPAM2.2 much quicker than the time required to develop ESPAM2.0/2/1. Chuck Brockway said that it is perplexing that ESPAM1.1 is still being used for transfers. Dave Colvin asked if Allan performed any tests with USG and upstream weighting. Allan said that he had not, but he felt it was not needed. Allan added that he had not tried Richard’s equation for representing the movement of a wetting front through the vadose zone either.
Item 8 - Allan began a discussion regarding developing an unconfined representation of the ESPA using MODFLOW USG in a variable transmissivity mode. Allan first discussed changes that were made to ESPAM in order to calibrate the model. This involved minor changes to the “Block-Centered Flow” flow package, and changes to the bottom of the aquifer or modeled aquifer thickness, and a minor modification to the well file.
Allan showed the committee the sequence of runs and checks that he performed in running USG in the variable transmissivity mode and discussed the results in terms of convergence and model output values. He indicated that he occasionally experienced convergence problems. Willem asked what the model was doing when a run was crashing, and Allan said it was dividing by 0, and it was always during the first steady state stress period. He added that he was not sure why it didn’t work. Willem asked if Allan could send him a crashed run and added that it has to be something peculiar with the solver. Willem asked if Allan ever tried SatMin. Allan asked if it involved modifying the MODFLOW code to prevent any cell from ever going dry. Willem said yes. Allan said he tried it and it worked, and it was the only way that he was successful calibrating any model with variable transmissitivity with earlier versions of MODFLOW.
5 Draft
Allan went on to say that he was not sure where the convergence problems are, but he thought that some parameter pairings just will not work. He showed a diagram of the convergence between parameter estimates during calibration. Chuck Brockway asked if you don’t know where the problem is, how can you convince me that we don’t have a problem with the model? Allan said that if you steer away from the problem, and PEST does not want to go back to that portion of parameter space, then you don’t need to worry. Willem said that the bottom line is that you cannot prove we do have a problem and that the “blow ups” are not important to the solution. Willem added that when the aquifer gets thin, you run into convergence problems, but the solutions may be correct.
Chuck Brockway said he is worried that we do not know what is going on in the model, and it is difficult to tell a client or an administrator that the model does not converge. Chuck Brendecke asked if Allan could see if the convergence problem is where you had to choose a minimum thickness. Allan said that we would know the model cell where it crashed, and it would be the cell with the biggest head change. Allan said he believed that he saved the first run, and he would post it. Chuck Brockway said that he hoped that Allan’s experiments do not result in recommendations. Allan recommended that the ESHMC convert to MODFLOW USG as the model framework. He said we should not convert to an unconfined representation with variable transmissivity or implement a refined or variable grid at this time.
Allan said that he could implement many of the improvements on the list in developing ESPAM2.2. Greg said that an evaluation of a refined grid could be done in the mean time. Allan agreed and said that he would continue to review the literature regarding MODFLOW USG for updates since there may be bugs that need to be corrected. Jon Bowling said there are many M.S. and PhD graduate students working on this problem. Chuck Brockway said that if we expect bugs, he is uneasy with converting to and using MODFLOW USG. Jon Bowling said that we should convert to and test the new version. Willem said that there are parts of the code that are not ready for prime time, and they will not be used.
Chuck Brockway said that the issue needs to be broken down and explained to the layman. He asked if the spring discharge was the same using MODFLOW 2000 and MODFLOW USG. David Hoekema asked if Chuck was referring to the variable transmissivity MODFLOW USG representation. Allan said that the fixed transmissivity representation for MODFLOW 2000 and MODFLOW USG were almost exactly the same, although he was not finished with the USG calibration. Chuck said that it was his understanding the USG could take care of the cell wetting and drying issue. Allan said cell drying and rewetting was supposed to be more stable with MODFLOW USG .
Item 9 - Neal Farmer presented an overview of the managed recharge sites at Mile Post 31 and near Lake Walcott. He began showing a map of the recharge canals and selected recharge sites in the western portion of the ESPA, and then he focused on Mile Post 31. He said that a pilot test had been performed on the site in 2010. Then a new turnout structure was constructed, and recharge was directed into the basin on March 27, 2013 at a rate of 200 cfs. The total volume recharged was 1,200 AF. Neal showed
6 Draft
hydrographs for 2 wells at a distance of 200 feet and 1.25 miles from the Mile Post 31 site. The water level data spanned the period before, during, and after the recharge event, and the impact on ESPA water levels was evident.
Next Neal discussed the Lake Walcott recharge site. He showed N – S and E – W geologic cross sections that ran through the site. Then he discussed the drilling and testing of two wells to determine if recharge via well injection was feasible above the water table. Neal said that the injection tests indicated low hydraulic conductivity in the unsaturated zone. Chuck Brockway asked what the goal of recharge at this site was, and Neal said a rate of 100 cfs. Chuck recommended getting away from the reservoir and sediments. Neal said the advantage of this site is that recharge could occur during the winter. Chuck Brockway asked if there were any treatment requirements, and Allan said that the permit allows direct injection of Lake Walcott water. Jon Bowling asked what the diversion location was, and Neal said the pump station is planned to be southeast of the test wells on the edge of Lake Walcott.
Neal then presented a “snapshot” of the hydrogeology in the Egin, Rexburg, Great Feeder, and Market Lake areas using geologic models developed with a software program called Rockware. He showed well locations and discussed geologic trends with respect to deposits of sediments including clays and basalt flows. He noted a rise in the clay deposits near Menan Butte, and said that near the Egin Bench, basalt occurs on top of sediments. Bryce asked Neal to describe the depositional scenario for the clays. Neal said that during tectonic and volcanic events, damming features developed on the Snake River, and clays were deposited in the backwaters. He added that once the dams were broken, coarser sediments were deposited on top of the clays. Allan said that the vertical exaggeration in the geologic models enhanced the dip of the clays. Neal agreed with Allan and redeveloped the model after the exaggeration was removed.
Chuck Brockway asked Neal if he had depicted faults with the software. Neal said not yet, but the software was capable of this. Chuck then asked Neal if he had extended the geologic model to the Rexburg Bench. Neal said that he had not been asked to do that yet. Chuck Brendecke said that if we become comfortable with refining the ESPAM grid, then we may want to create geologic models like this. Allan said that we probably will want this type of work at Egin, American Falls and in the Burley – Paul areas.
Item 10 - Kevin Boggs presented the results of his research related to improvements in forecasting the ESPA discharge from springs in the Thousand Springs area. The title of his presentation was: Improved Baseflow Forecasting. He indicated that Gary Johnson, Jerry Fairley, Rob Van Kirk, and Jim Bartolino were on his dissertation committee and also contributed to his work. Kevin presented analytical, statistical, and numerical approaches to forecasting and the overall conclusions.
Regarding the analytical approach, Kevin said that Gary Johnson and Rob Van Kirk, in particular, helped derive equations that were the basis for his analysis. He started with the governing equation, described the region of flow, and established the boundary and initial conditions. Kevin said that he Used Darcy’s law to develop a general form of a
7 Draft
recharge/discharge equation for total discharge from aquifer. His method of solution was the Fourier Series for aquifer discharge. He used equations to describe recharge from instantaneous events and recharge that is periodic in time. The conclusions to his analytical approach are that there is a relationship between lag, attenuation, and distance (aquifer time scale control), and short-term changes in aquifer stresses near the discharge location account for most of the annual and decadal-scale aquifer discharge variability. The seasonal variation in discharge is derived from a pulse that is close to the spring.
Then Kevin described his statistical approach. He listed predictor variables that included the Big Lost River flows, Big Wood Canal Company diversions, the Big Wood River flows, ET, Milner-Gooding Canal Company diversions, Northside Canal Company diversions, precipitation, groundwater pumping, sprinkler percentages, aquifer storage, and the snow water equivalent (SWE) at the Soldier Mountain snotel site. He recognized that predictors potentially lagged in time, and their impact was averaged over several years. Using an ARIMAX model, he showed that spring discharge is a combination of last year’s spring discharge plus a combination of exogenous predictors at an optimum lag and moving average window (MAW) for each predictor.
In refining his statistical approach, Kevin utilized a calibration dataset (1950 – 1999) and a validation dataset (2000 through 2010). In an approach he defined as Model 1, he only used historic data to forecast spring discharge (data available each April). In an approach he defined as Model 2, he found that recharge sources impact spring discharge in the same year the recharge occurs. For both models, Kevin used an AICC ranking and indicated it is an effective way to evaluate the relative strength of predictors. He concluded that Model 2 explained nearly twice as much variability in discharge, and the model with the highest AIC weight included streamflow, two irrigation diversion variables, and “second tier” variables defined as aquifer storage and the SWE at the Soldier Mountain snotel site.
The third part of Kevin’s presentation was his numerical approach which was based on ESPAM2.0. He said that the primary spring discharge components are the decay of spring discharge from initial head conditions and the contributions to spring discharge from future recharge and pumping events. He developed a forecast tool that is a spreadsheet (Excel –VBA coded) that combines the effects of starting heads and indi- vidual recharge components over time.
Kevin’s conclusions were that through his analytical work, he developed an under- standing of the relationship among lag, attenuation, and distance. His statistical mod- eling is consistent with his analytical results, and he was able to develop an annual spring Q forecast (applied each April). Finally, his numerical too produces an accurate monthly forecast (applied each January).
Item 11 - The committee agreed that the next meeting should be Friday, December 13, 2013.
8 Draft
DECISION POINT SUMMARY
The following was agreed upon:
1) IDWR agreed to add Eric Powell as a member of the ESHMC.
2) IDWR agreed to obtain the portion of the release from Magic Reservoir during the fall of 2012 that was diverted into the Richfield canal and update the recharge analysis.
3) IDWR agreed to review spring discharge data for evidence of an increase in flow due to the release from Magic Reservoir during the fall of 2012.
4) For development if ESPAM2.2, IDWR recommended that it pursue all low effort recom- mendations on the prioritized list of improvements, with a bias toward higher ranking rec- ommendations.
5) For the development of ESPAM3.0, which would involve more comprehensive revisions, IDWR recommended that resource intensive updates should come from the higher scoring recommendations on the prioritized list of improvements.
6) IDWR agreed to distribute the list of improvements intended for ESPAM2.2 vs. ES- PAM3.0.
7) IDWR agreed to post the initial model runs that failed when using MODFLOW USG in the variable transmissivity mode where convergence problems developed between initial pa- rameter estimates.
8) The committee agreed that the next meeting should be Friday, December 13, 2013.
9