Model Accuracy regarding Aquifer Levels and Base Flows

The software that much of the world uses to simulate aquifers was written by Michael McDonald.  Nebraska hired McDonald/Morrissey Associates as Nebraska's experts in the dispute with Kansas. Nebraska has hired many experts and attorneys and almost all of them are under non-disclosure agreements.  In April of 2006, McDonald/Morrissey Associates released a review of the accuracy of the Model they helped developed. 

The full report can be read at: http://www.waterclaim.org/presentations/problems/mcdonaldmodelreview_060411.pdf

The report reviews 73 locations and compares the Model calculated results with the observed results.  The report does not quantify the percentage of locations that are different however summing the results up shows that 56% of the well locations have trends that are different than what the Model predicted.  That is a 56% error rate.  You can verify the error rate for yourself by clicking on the dots in the first map and comparing them to reality.  This makes it easy to identify the problem areas and see how much the errors are.   Keep in mind to look at the y axis and note the scale you are looking at.

WaterClaim has color coded the wells to show the change in aquifer level since the 1950's.  Red denotes a decline in the aquifer.  Green denotes an increase.  Blue denotes a relatively stable aquifer.  There is insufficient data or an error in the data at the yellow dots.  You can click on each dot and see the observed data and the predicted data. 

Map1  (interactive)

As can be seen, most of the Republican River Basin does not have a declining aquifer.  The problem is primarily in the three western counties. 

This second map shows the accuracy of the Model.  If the Model accurately predicts the trend of the aquifer increase or decrease then the Model is considered to be helpful.  If the Model does not accurately predict then the Model needs more adjustments.  Each red and yellow dot on the second map indicate a problem with the Model.

There are 78 dots.  By our estimation 5 of these data points have errors or insufficient data to make a determination. Each red dot or green dot with a red center means we believe the Model is off in its prediction.  Approximately 40% of the data points are incorrectly predicted by the Model.  According to the maker of the software the Model is only wrong 15% of the time.  You can make up your own mind by reviewing the data.  To find where we disagree with McDonald/Morrissey click on the green dots with the red center or the yellow dots.  You will need to click on the top map to see the data.   If the observed trend and predicted trend since 2001 are in agreement then McDonald says the Model is doing its job.

You might notice that the accuracy of the Model is better in the west.  In the east and along the main stem, the accuracy of the Model is substantially worse.  The Model generally predicts that things will be or are worse than they really are.  This would mean that the Model is saying there is more of a problem than there really is.  Since there is a huge amount of money depending on the accuracy of the Model, this is worth investigating more and making sure it is as accurate as possible.

The dots on the above map are clickable.  If you notice a dot on the bottom map you want to look at, note its number and click on it on the above map.

Map2

The report concludes that the individual well records are not important but what is important is the accuracy of the observed vs the calculated base flow numbers at the 24 locations that the Model uses to measure compliance.  In other reports, these points are called the accounting points. In this report, they are called stream gages. And what is the report says happens at these 24 points determines if Nebraska is compliance or not. 

I have found two different maps that show where the accounting points might be.  They do not agree with each other but they are similar.

This is a very poor quality map but is the only one that I have been able to find that shows the accounting points.  Even though multiple people have asked the DNR for more details on the location of the accounting points, the DNR refuses to release the information, even to the Legislature.

Please refer to the Base Flow concerns for a very closely related review.