Mapping the course of satellite precipitation estimates for flash floods into the 21st century

This paper briefly describes a "roadmap" of Satellite Precipitation Estimation (SPE) development from the late 1970s into the 21st century. For most of the past 20 years, the production of SPEs for flash floods has been a manual/interactive process for alerting forecasters and hydrologists of the potential for heavy precipitation and flash floods. Precipitation estimates for flash floods are GOES driven products due their (GOES data) high spatial and time resolution characteristics. As the 21 st century approaches the challenge has been to automate SPEs for flash flood application. Automation will allow an increase in SPE application from the already established "alert tool " (mentioned above) to infusion into the NWS/OH stage 3 Product (gauge + WSR 88D + satellite (planned), hydrological flow models that are run at local RFCs and cloud models (that will inturn provide precipitation estimates).

The NOAA/NESDIS Satellite Analysis Branch (SAB) computes operational QPEs using the Interactive Flash Flood Analyzer (IFFA) technique to provide real time SPEs to field forecasters and RFCs. SAB's SPEs are sent out via AFOS and AWIPS. However, due to the manual/interactive nature of the IFFA methodology, precipitation estimates cover limited areas for limited periods of time and can take a siginificant amount of time to produce. In order to improve the spatial and temporal coverage of SPEs while improving timeliness, NESDIS/ORA has developed an automatic SPE algorithm called the Auto-Estimator. The experimental Auto-Estimator provides in real time the following precipitation information: (a) instantaneous rainfall, (b) average hourly rainfall rates, (c) 3 hourly rainfall accumulations, (d) 6 hourly accumulations, (e) 24 hour accumulations. Also 15 minute GOES imagery is used to compute the estimates (IFFA mentioned above, only uses 30 minute imagery). The Auto-Estimator is being tested by SAB and current plans are for this and other algorithms to replace the IFFA system.

In order to improve the accuracy of the Auto-Estimator and make this algorithm more robust, other channels from GOES (GOES has 5 imager channels) will be exploited for precipitation information. In addition, especially over water and other remote areas, microwave precipitation estimates from SSM/I and AMSU will be used to continuously update GOES precipitation measurements both with respect to precipitation intensity and location. Updates will become quite frequent (every 2 or 3 hours) when both the SSM/I and AMSU passes are utilized. Microwave derived measurements are more physically/directly related to precipitation than those form GOES. However, microwave is only available on POES platforms and are thus much less timely than GOES and have lower spatial resolutions. Where WSR 88D data is available, the SPE algorithm will use such information to delineate rain/no rain areas and eventually to calibrate and adjust the GOES estimates (both with respect to location and intensity).

The ultimate is to integrate the GOES with the SSM/I and AMSU, WSR 88D and in-situ rain gauge measurements to come up with a multi-spectral/multi-sensor algorithm for estimating precipitation form all types of precipitation systems (convective, stratiform, hybrids). Integrated precipitation data sets are a necessity for imrproving the prediction of stream/river flows and crests. Finally, artificial neural network techniques for estimating heavy convective rainfall have show great promise. Our greatest hope is to have by the 2005 - 2010 time frame, microwave "flying" on operational geostationary platforms that cover the globe.