In this task, we will choose a constant selected threshold value and test the methods we use to separate out independent flood events. First, we will use a simple time-based separation criterion with a small and a large value of separation (we have already tested the drainage-area based guideline as one point). Then, we will use a more advanced separation technique that requires the flow between adjacent events to recede beneath a threshold.

Get a Baseline

First, we will set a control for the next few tests. We will get a summary for the default time separation criteria (10 days computed using the ceil[5+ln(DA)] rule) and a threshold of 2,000 cfs. 

• Right click on the "SanLorenzo_1EventPerYear" analysis and select Save As....
• Enter "SanLorenzo_Base" as the analysis name.
• Open the new analysis and select the Filtering tab.
• Enter 2000 in Selected Threshold field. 2,000 cfs is in the middle of the "stable" range for the shape plot and mean excess plot and should produce a similar result to other thresholds in this region.
• Make sure that the Time Separation Between Peaks panel still appears as in the image below:
  
  
• Click the Compute button.
• Investigate the results:
  • Navigate to the Tabular Results tab.
  • Click the Plot Curve button.
  • Click the Plot Event Count Histogram button.
• Click OK to close the "SanLorenzo_Base" analysis. 
• Save the SSP study.

Time-Based Separation

We will test two additional separation durations: 3 and 21 days, and see how the results change relative to the base time separation of 10 days (holding the selected threshold constant at 2,000 cfs). 

• Create 2 new Peaks Over Threshold Analyses, named "SanLorenzo_3days" and "SanLorenzo_21days."
• Change the Time Separation Between Peaks criterion to User-specified and enter 3 and 21 as the Duration of time between peaks (days), respectively.
• Click the Compute button.
• Save the study.

QUESTION 7: How do you feel about the performance of 3 and 21 days as time separation criteria?

Time-Based Separation with Recession

Finally, we will test the effect of separating events not only by time, but also by requiring that flow goes below a certain recession threshold between events in order for them to be considered independent. Here is a link to the USGS daily flow page for this gage showing the wet part of WY1982 which had a number of large peaks: https://waterdata.usgs.gov/monitoring-location/11160500/#dataTypeId=daily-00060-0&startDT=1981-12-01&endDT=1982-05-01

Take a look at the behavior of the hydrograph between the peaks. The recession is based on how high the prior peak is and how much time has passed. We will try a couple of recession thresholds based on this plot, a higher (more permissive) threshold at 500 cfs, and a lower (less permissive) one at 100 cfs, and also combine the 500 cfs threshold with a shorter time separation criteria to reduce the influence of timing in favor of the recession threshold. 

• Duplicate the "SanLorenzo_Base" analysis and enter "SanLorenzo_500cfs_10days" as the Name.
• On the Filtering tab | Time Separation Between Peaks panel, check that the the second radio button is selected and the gage's drainage area of 106 (square miles) is entered in the field.
• In the Method for Computing Magnitude Differential panel, select the box next to Defined Threshold Value. Enter 500 in the field.
• Click Compute.
• Repeat the previous steps.
  • Duplicate the "SanLorenzo_500cfs_10days" analysis.
  • Name the analysis "SanLorenzo_500cfs_3days." 
  • On the Filtering tab | Time Separation Between Peaks panel, select the first radio button and enter 3 days.
• Repeat the previous steps. 
  • Duplicate the "SanLorenzo_500cfs_3days" analysis
  • Name the analysis "SanLorenzo_100cfs_3days." 
  • On the Filtering tab | Method for Computing Magnitude Differential panel, change the Defined Threshold Value to 100.

QUESTION 8: Identify the differences between these three choices and the baseline model we started with.

Return to Peaks Over Threshold Analysis: Flood Separation and Threshold Choice for the final workshop files.