Truckee CMIP5 LOCA, RCP 4.5 and 8.5

25 policies are chosen from the archive of 463 policies using the Kennard Stone sampling algorithm. The decision variables, including shortage tier elevations, volumes, and surplus distances, are scaled before the algorithm is implemented. Each vertical bar is a representation of the shortage tiers for a policy. The y axis shows elevation, while color and labels show shortage volume. The x axis shows satisficing rank, where rank 1 (left) is the most satisficing. Satisficing calculations are based on the 500 member SOW ensemble created with cLHS. Policy ID and SOM node group are listed above each policy.

25 policies are chosen from the archive of 463 policies using the Kennard Stone sampling algorithm. The decision variables, including shortage tier elevations, volumes, and surplus distances, are scaled before the algorithm is implemented. Each vertical bar is a representation of the shortage tiers for a policy. The y axis shows elevation, while color and labels show shortage volume. The x axis shows satisficing deviation rank, where rank 1 (left) is the most negative satisficing deviation. Satisficing deviation calculations are based on the 500 member SOW ensemble created with cLHS. Policy ID and SOM node group are listed above each policy.

25 policies are chosen from the archive of 463 policies using the Kennard Stone sampling algorithm. The decision variables, including shortage tier elevations, volumes, and surplus distances, are scaled before the algorithm is implemented. Each vertical bar is a representation of the shortage tiers for a policy. The y axis shows elevation, while color and labels show shortage volume. The x axis shows satisficing rank, where rank 1 (left) is the most satisficing. Satisficing calculations are based on the 500 member SOW ensemble created with cLHS. Policy ID and SOM node group are listed above each policy.

52 solutions are above the 90th percentile satisficing percentage (40.79%). They are clustered into 9 nodes with SOM.

476 Lake Mead solutions simulated in 76 SOW each. Satisficing percent uses Mead1000 <10%, Powell3490<5%, and LBShortVol<600KAF criteria

AGU Fall Meeting 2020

Note that the colors indicate the experiment in which the alternative was found. 1: 5K FE. 2: 10K FE. 3: 5K FE Dominated by solution(s) in the 10K archive.

Note that the colors indicate the experiment in which the alternative was found. 1: 5K FE. 2: 10K FE. 3: 5K FE Dominated by solution(s) in the 10K archive.

Objectives for 18-member archive found with Borg-RW. The RW model of was modified as follows. The Mead 1000 objective was changed to Mead 1100, and two objectives to minimize AZ and CA average annual shortage were added. Each alternative is colored according to the objective LB Shortage Frequency.

Decision Variables for 18-member archive found with Borg-RW. The RW model of was modified as follows. The Mead 1000 objective was changed to Mead 1100, and two objectives to minimize AZ and CA average annual shortage were added. Each alternative is colored according to the objective LB Shortage Frequency.

Wet year defined as annual flow exceeding 19.1 MAF, which is one standard deviation above observed mean. The max gap is maximum years between wet years in an SOW, and avg gap is the average gap between wet years in a SOW.

Frequency defined as the count of years exceeding wet/dry thresholds. Duration defined as the max consecutive dry/wet years in the SOW. Wet/dry thresholds defined as the mean observed natural flow plus/minus one standard deviation, 19.1 and 104 MAF, respectively

Deficit defined as annual flow in N year window below 15 MAF. Deficit is positive, surplus is negative. In this figure, the cumulative deficit in a N year window is normalized by the window length, resulting in the annual average deficit in a N year window.

After obtaining N year average flow in every year, take the max and min values of every trace

Deficit[surplus] calculated as cumulative flow below[above] 15 MAF each year in a N year window. The deficit[surplus] is then normalized by the window length, resulting in the average annual deficit [surplus] during the N year window. The maximum deficit [surplus] observed in each SOW is taken.

Calibration using 4 VIC model soil parameters. Tradeoffs of two objectives shown.

Note for Nathan: this is the same plot you already have in your RPubs account, but with a red-blue colorscale. You should switch all to use these

Y axis shows the N year window size in which Mead 1000 is calculated. X axis shows the Year-Month the N year window begins. Color is the percent of months in the N year window where Mead is below 1000 feet. This figure shows the average, as Mead 1000 takes a unique value for every SOW in the uncertainty ensemble. This example analysis is for policy R1 from Alexander 2018.

Y axis shows the N year window size in which Mead 1000 is calculated. X axis shows the Year-Month the N year window begins. Color is the percent of months in the N year window where Mead is below 1000 feet. This figure shows the 75th percentile, as Mead 1000 takes a unique value for every SOW in the uncertainty ensemble. This example analysis is for policy R1 from Alexander 2018.

Probability is the number of traces where Mead drops below 1000 feet at least one time in the next N years divided by the total number of traces. Note, this is not the same as the Mead 1000 metric used in Alexander 2018.

15 MAF as threshold Positive is deficit Negative is surplus

policy R1 Boxplots of the difference between the max and min pool elevation in every year. Considers six supply and demand scenarios