A Slinky-Space Streamflow Estimation Method for TMDLs: TIme Series Analysis Meets Geostatistics
Location
Eccles Conference Center
Event Website
http://water.usu.edu/
Start Date
3-28-2006 10:20 AM
End Date
3-28-2006 10:40 AM
Description
Water quality analyses under the Clean Water Act Total Maximum Daily Load (TMDL) program typically require estimation of total pollutant load in a water body using simultaneous observations of pollutant concentration and streamflow. However, in small rural watersheds, streamflow measurements can be sparse and may not correspond in time with pollutant concentration data. Because of this, a robust streamflow estimator is required that uses existing streamflow observations—however sparse—to generate instantaneous streamflow estimates coinciding with pollutant concentration data. This paper presents a streamflow estimation method that operates on a single time series in a time domain termed “slinky space” to produce instantaneous streamflow estimates from sparse data. The approach involves “wrapping” the streamflow time series data into a coil or “slinky”-shape where each loop of the slinky represents one year of data; and corresponding calendar days (e.g. June 12) for each year are spatially near each other on the surface of the slinky. One then models the surface of the slinky using nearest neighbor inverse distance weighting, kriging, or other geostatistical methods to spatially interpolate the needed missing streamflow data. This approach is shown to be effective at modeling one of the strongest indicators of streamflow—persistence—in both a sequential days, and across years of similar (e.g. drought) conditions.
A Slinky-Space Streamflow Estimation Method for TMDLs: TIme Series Analysis Meets Geostatistics
Eccles Conference Center
Water quality analyses under the Clean Water Act Total Maximum Daily Load (TMDL) program typically require estimation of total pollutant load in a water body using simultaneous observations of pollutant concentration and streamflow. However, in small rural watersheds, streamflow measurements can be sparse and may not correspond in time with pollutant concentration data. Because of this, a robust streamflow estimator is required that uses existing streamflow observations—however sparse—to generate instantaneous streamflow estimates coinciding with pollutant concentration data. This paper presents a streamflow estimation method that operates on a single time series in a time domain termed “slinky space” to produce instantaneous streamflow estimates from sparse data. The approach involves “wrapping” the streamflow time series data into a coil or “slinky”-shape where each loop of the slinky represents one year of data; and corresponding calendar days (e.g. June 12) for each year are spatially near each other on the surface of the slinky. One then models the surface of the slinky using nearest neighbor inverse distance weighting, kriging, or other geostatistical methods to spatially interpolate the needed missing streamflow data. This approach is shown to be effective at modeling one of the strongest indicators of streamflow—persistence—in both a sequential days, and across years of similar (e.g. drought) conditions.
https://digitalcommons.usu.edu/runoff/2006/AllAbstracts/30