Improving Real-Time Precipitation Estimation over the Mountainous Regions of the Western United States, using Multi-Sources Remotely Sensed and Lightning Data

Improving satellite-based rainfall retrieval algorithms for estimating high resolution (up to hourly 4km x 4km) precipitation, particularly over mountainous regions, where ground-based sources (e.g. radar and gauge networks) cannot cover, is the objective of this study. High-resolution cloud-top infrared brightness temperature (IR-T_b) from geostationary satellite (GOES) and cloud-to-ground lightning (CGL) are used for precipitation estimation. Rainy cloud with colder top temperature (T_b) and stronger lightning generally produces heavier rainfall. CGL-rainfall studies have demonstrated that lightning (L) can estimate rainfall more accurately than remote sensing-based estimates, particularly for thunderstorm events, because CGL-R relationship is highly correlated (cc= 0.72). The occurrence of CG lightning varies with topography and seasonality. An artificial neural networks algorithm has been applied for estimating rainfall from cloud-top IR in conjunction with lightning. Preliminary results demonstrate that using lightning in addition to cloud T_b could increase the accuracy of rainfall estimates up to 0.23. The results are for a short time storm in the July 2002 over an area with latitude 32-38^oN and longitude 104-112^oW in a mountainous region.