STAR Satellite Rainfall Estimates - Current Work
The Self-Calibrating Multivariate Precipitation Retrieval (described here and in in Kuligowski 2002) has run in real-time at NESDIS / STAR since November 2004. Some relatively minor improvements have been made to the algorithm since then, but some significant changes have completed offline testing and should soon be implemented into the real-time processing. These include:
- Adding Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) and Microwave Imager (TMI) data to the calibration data set;
- Matching the SSM/I and AMSU CDF's to the TMI distribution (intercalibration);
- Larger calibration regions (found to improve performance and reduce processing time);
- Separate calibration for deep convection based on difference between GOES IR (band 4) and water vapor (band 3) brightness temperatures;
- Additional of GOES visible data for improved daytime rain / no rain discrimination.
For additional details, please refer to the slides from a presentation at the 23rd American Meteorological Society Conference on Hydrology. A conference paper can be found here.
GOES-R Algorithm Development
The next-generation of NOAA's Geostationary Operational Environmental Satellites (GOES) is scheduled for launch in early 2015, and it will feature advanced sensors that will be highly useful for remote sensing of precipitation. These include the Advanced Baseline Imager (ABI), which will feature more spectral bands, finer spatial resolution, and more rapid scanning than the current-generation Imager, and the GOES Lightning Mapper (GLM) which will provide the first-ever operational monitoring of lightning from geostationary orbit.
Rainfall rates retrieved from SEVIRI data at 12:45 UTC January 7, 2005 using the GOES-R Rainfall Rate algorithm
Forecast of 3-hour rainfall accumulation for 1800-2100 UTC January 5, 2005 retrieved from SEVIRI data at 1745 and 1800 UTC using the GOES-R Rainfall Potential algorithm
To prepare for GOES-R, the Algorithm Working Group (AWG) has been commissioned to provide recommended, demonstrated, and validated algorithms for processing GOES-R observations into products which satisfy user requirements. As part of this effort, the Hydrology Algorithm Team (AT) is responsible for three products:
- Rainfall Rate;
- Rainfall Potential during the next 0-3 hours;
- Probability of Rainfall during the next 0-3 hours;
For the first two algorithms, the Hydrology AT selected and evaluated several candidate algorithms using METEOSAT Spinning Enhanced Visible Infra-Red Imager (SEVIRI) data as a proxy for ABI. The results of this testing were that the Self-Calibrating Multivariate Preciptation Retrieval (SCaMPR) algorithm was selected for the rainfall rate retrieval, and the K-Means technique developed at the NOAA / OAR / National Severe Storms Laboratory was selected for the rainfall nowcasting.
Since there are no legacy satellite-based probability of rainfall algorithms, the algorithm has been developed "from scratch" using statistical matches between intermediate rainfall forecasts from the Rainfall Potential algorithm and observed rainfall.
The source codes for all three algorithms are in various stages of implementation into a demonstration operational framework at this time.