2002 Annual

Thursday, 17 January 2002: 9:30 AM
Improving flood prediction using Kalman Filter, mesoscale atmospheric model forecasts and radar-based rainfall estimates
Ashutosh S. Limaye, USRA, Huntsville, AL; and K. Stellman
Poster PDF (110.3 kB)
In this study, we are incorporating radar-based rainfall estimates to improve Quantitative Precipitation Forecasts (QPFs) from mesoscale atmospheric models using a Kalman filter. We focus on two QPFs from different mesoscale atmospheric models and two radar-based rainfall estimates to estimate flood forecasts in flood-prone Amite River watershed in southeastern Louisiana. The Stage III rainfall estimates, generated by Lower Mississippi River Forecast Center (LMRFC) for the LMRFC domain, are produced by merging WSR-88D radar data and rain-gage measurements. The StageIV estimates are generated by the National Hourly Precipitation Analysis by combining the national mosaic of WSR-88D data and hourly rain gage measurements. StageIII has a larger rain gage network (hourly as well as cooperative rain gages) and is used in LMRFC river stage forecasts. The nested mesoscale atmospheric model MM5 is used to provide updated hourly forecasts every 12 hr over a forecast period of 48 hours and at a spatial resolution of 12 km. In addition to the MM5 forecasts, we are using LMRFC runs of workstation version of the mesoscale atmospheric model Eta. It is run at a spatial resolution of 10 km.

We are focussing our attention of this study on a large rain event associated with hurricane Allison, which flooded parts of Louisiana in June of 2001. We will present some results from the intercomparison of different mesoscale model QPFs for the rain event as well as radar rainfall products and their utility in Kalman filter context for streamflow estimation.

Supplementary URL: http://www.ghcc.msfc.nasa.gov/regional/limaye/ams2002_extended_abstract.html