84th AMS Annual Meeting

Wednesday, 14 January 2004: 8:45 AM
Automatic Gage-Adjusted Radar Rainfall Estimation for Real-Time Storm Water Control
Room 6B
Leiji Liu, OneRain Inc., Orangevale, CA; and B. Hoblit, R. Jensen, and D. Curtis
Poster PDF (309.9 kB)
Accurate rainfall estimation is an important issue in research and engineering fields. The traditional approach is to use a number of rain gages to create the spatial and temporal rainfall distribution using different interpolation algorithms. Using this technique, rainfall estimates are marred by the sparse distribution of rain gage network and events that occur between gages are not captured by the gages. Radar rainfall estimates, on the other hand, can capture the entire rainfall events but physical characteristics of radar system and operational practices have determined that radar rainfall estimates are not always consistent with ground truth. Gage adjusted radar rainfall estimates combine the advantages of both radar and gages, resulting in better rainfall estimate.

OneRain Inc. (formerly NEXRAIN Corp.), on behalf of City of Louisville/Jefferson County Metropolitan Sewer District (MSD), developed an automated gage-adjusted radar rainfall system that is integrated with MSD’s existing rain gage network, computer network, telemetry system, web page, combined sewer overflow (CSO) program and emergency response initiatives. The CSO program includes an innovative real-time control (RTC) system to manage sewer flows. The system is a hardware/software package designed to ingest gage and radar data, adjust the radar data with the local gage data, serve rainfall images to the Internet and Intranet, and create output data files. These output data files integrate seamlessly with MSD’s geographic information system (GIS), the real-time control (RTC) system, and with MSD’s Oracle database for future use. The system also includes a radar rainfall forecast feature that provides predicted rainfall in five minute increments over the next 60 minutes.

Single parameter adjustment is used due to its effectiveness for small area and its computation efficiency for real time adjustment. Algorithm used is robust enough to fix missing radar or gage files by interpolation in temporal and spatial domains.

The hardware includes a SunFire V100 server with OS Solaris 8 for processing gage and radar data and a Dell Pentium 4 machine with OS Windows 2000 Professional for image display. The machines are attached to a local network so files and images created can be transferred to specified locations for different purposes. Another completely redundant and parallel system operates at OneRain’s home office in California to ensure continuous output and backup capability.

Supplementary URL: http://