Multi-sensor Precipitation Reanalysis

The network of Weather Surveillance Radar - 1988 Doppler (WSR-88D), commonly know as the Next Generation Weather Radar (NEXRAD), consists of approximately 140 sites in the Continenal U.S. Most radars have been operational for approximately 10 years and have been providing radar reflectivity estimates for the NEXRAD Precipitation Processing Subsystem (PPS) which produces radar-derived rainfall products in real time for forecasters in support of the National Weather Service's mission and external users. The PPS and follow-on multisensor precipitatoin estimation applications compute rainfall estimates in stages and, historically, the Stage III products are generated at the River Forecast Centers (RFCs). Recently, the Multisensor Precipitation Estimation (MPE) algorithm has replaced the Stage III algorithm. The MPE algorithm is an improvement on Stage III in several areas. The MPE algorithm delineates an effective radar coverage area based on seasonal radar climatology. The mosaicking scheme uses data from adjoining radars based on the radar sampling geometry. The algorithm provides analysis over the entire RFC, rather than mosaicking of radar-by-radar analysis. The algorithm has an improved mean-field bias correction, and it includes a new local bias correction procedure. Each of these improvements is designed to reduce or eliminate biases that are inherent in the radar rainfall estimates, but the algorithm is geared toward real-time operational implementation.

The NEXRAD data are available for an approximately 10 year record, which provides for a long term data set at high resolution both spatially and temporally. We have implemented the MPE algorithm with the historical NEXRAD data, the Digital Precipitation Array (DPA) products, in a reanalysis mode to develop a data set that is suited for long term climatological applications. We perform the reanalysis with the goal of reducing biases that continue to plague operational products. Reanalysis allows for several improvements to the historical radar rainfall products. One of the main improvements included in the reanalysis is to incorporate more in-situ measurements of rainfall which are important for the bias correction procedures. Higher quality and higher density rain gauge measurements will help improve the multisensor rainfall estimates. Further the reanalysis allows for detailed experiments for parameter tuning. All of these experiments will allow us to improve current estimates such that they are more suited for long term water resources and climate applications.

Radar-based Precipitation Product

1. High Resolution (spatially and temporally)

2. Unbiased

3. Continental U.S. (Current Project Domain is North and South Carolina)

Figure 1
click on image for larger view

DelGreco S., D. Kim, and B.R. Nelson, "Operational Issues from NCDC Perspective," Q2 - Next Generation QPE and QPF Workshop, Norman Ok, June 27-30, 2005.

Nelson, B.R, DJ Seo, D Kim,J.J. Bates, 2006: "Multisensor Precipation Reanalysis." American Meteorological Society 2006 Annual Meeting, Paper #4.6, Atlanta, GA, January 29 - February 2, 2006.

Nelson, B.R. and J.J Bates, 2004: "Rainfall variability studies based on a long term radar-rainfall data set." AGU Fall Meeting, 13-17 December 2004, San Francisco, CA, Washington, D.C.,American Geophysical Union, pages (H33C-0479 1340h) (Dec. 2004).