Michael Schaffner, Alexander Tardy, Jayme Laber, Carl Unkrich, and David Goodrich, A collaborative effort between NWS Western Region Headquarters, NWS San Diego, NWS Oxnard, and the USDA-ARS Southwest Watershed Research Center ALERT User Group Symposium, May 8, 2014
Opening Questions What tools at a Weather Forecast Office are needed to effectively identify ungaged basins under a flash flood threat? Assuming the ability to accurately define a flash flood warning polygon and mention a given basin in the text of the warning, is that enough? How about predicting the magnitude of the event? The impacts of the flash flood? The timing of the flood? How can the NWS warn recreational users and in particular those traveling in the streambed?
Overview of Fish Creek Fast responding 50-square mile watershed. Ungaged. Located within Anza Borrego State Park in Southern CA. Receives heavy recreational use. Stream channel is a state unmaintained road. The basin contains the Wind Caves which is a popular hiking destination.
Split Mountain Road Confluence of N. Fork Model outlet point Wind caves
Fish Creek Images Image source: http://www.panoramio.com/photo/46761615 Image source: http://www.panoramio.com/photo/14546598
Forecast Challenges for Fish Creek While the Yuma radar provides good coverage for the basin, it has not been adequate to consistently detect flash flooding. Even if radar and Flash Flood Guidance worked perfectly, they do not provide details on the magnitude or timing of flash floods. Not only is there a requirement to detect flash flooding in Fish Creek (first step), but it is essential to decision support services (DSS) with the state park and its visitors to predict timing and the magnitude of a flash flood (second step).
Flood of Record July 30, 2012 Two vehicles were washed downstream. Occupants had to climb to higher ground and be rescued by helicopter. Eyewitness reports indicated a wall of water along portions of Fish Creek.
Decision Support Service Engagement with Anza Borrego State Park This flood was the catalyst for the Weather Forecast Office to engage state park rangers. Led initially to an all-day tour of Fish Creek. Subsequent interaction with state park ranges has taken place. This event marked the recognition of the high flash flood threat for Fish Creek and the surrounding canyons.
Flood of Record Rainfall: Rainfall covered the entire basin. 1.60 inches = basin average rainfall. 8.00 inches = maximum rainfall pixel within basin from radar.
Fish Creek Channel Overview High water mark Overbank area Main channel of Fish Creek Image source: NWS San Diego
Flood of Record Pictures Image source: http://www.desertusa.com/mb3- photos/fish-creek1.jpg Image source: Anza Borrego State Park
Flood of Record Discharge Estimation Discharge estimate was a rough estimate. Not from a formal USGS slope-area measurement. Cross sectional area of main channel and overbank area estimated. Discharge = 13,600 cfs. Results were compared to output from USGS regional regression equations.
Peak Flows calculated from USGS regression equations PK2 (2-year peak flow) PK5 (5-year peak flow) PK10 (10-year peak flow) PK25 (25-year peak flow) PK50 (50-year peak flow) ESTIMATED PEAK FLOW PK100 (100-year peak flow) PK500 (500-year peak flow) Peak Discharge 148 cfs 1,080 cfs 2,799 cfs 6,170 cfs 9,720 cfs 13,600 cfs 14,970 cfs 29,980 cfs Peak flow statistics calculated from USGS California StreamStats Program
How does one go about modeling an ungaged basin? Collect a series of flow events. Magnitude Timing Radar rainfall data Preferred to have a several high flow and if possible a flood of record event. Develop a synthetic peak flow rating curve. Develop critical stages (Action Stage, Minor Flood Stage, Moderate Flood Stage, and Major Flood Stage).
KINEROS2 Model Distributed and event based. Designed to simulate flash floods in semi-arid regions. Produces a forecast hydrograph. Produces a new forecast as radar data is received. Runs on a PC. Requires initial set-up of the model using a USDA GISbased program (AGWA).
KINEROS2 Model Components for Fish Creek Overland flow planes Open channel elements Basin outlet = where hydrograph is generated
Model Calibration The model was calibrated manually. Using a full range of events including null events. Events used in calibration: Major flood (3) Moderate flood (1) Minor flood (0) Action (1) Null (3) The calibration test for a flood event is that it fell into the correct flood category (e.g. moderate flooding). The calibration test for a null event was to ensure that the simulated peak flow remained below Action Stage.
Model display for flood of record
Model Results - Summary Calibration scheme was simple with two sets of calibration parameters. One for low flow through low-end major flood events Another for higher-end major flood events through the flood of record. Simulated peak flow for each event fell within appropriate flood category it was assigned to (action, minor, moderate, major, etc.). Null events produced minimal flow and remained below Action Stage.
Model Results Lead Time Lead time is constrained by the fact that radar QPE is being used and no QPF. Lead time is largely a function of the time of concentration for the outlet point. The model provided an average lead time of: 98 minutes for action stage 63 minutes for minor flood stage 50 minutes for moderate flood stage 48 minutes for major flood stage.
How the model fits into Weather Forecast Office Operations The forecast hydrograph will first be used as guidance to determine if a flash flood warning is needed. If a warning is issued, wording specifically mentioning Fish Creek would be placed in the text of the warning. Will be used in conjunction with other tools available. As skill is gained in the use of the model, information on the magnitude, timing, and impacts associated with the flash flood peak will be included in the flash flood warning and DSS provided to the state park.
Conclusions Collecting data in the field and from other sources allowed the reconstruction of not only the flood of record, but also several other significant flood events. When well calibrated, the model can provide valuable information across the full spectrum of the flash flood warning process. The model is capable of simulating peak flow in terms of both magnitude and timing. The approach employed here allows locations without a stream gage to be modeled.
Questions and Discussion