ASSESMENT OF MULTISENSOR QUANTITATIVE PRECIPITATION ESTIMATION IN THE RUSSIAN RIVER BASIN

The use of weather sensing radar measurements along with corresponding gauge data seek to provide reliable estimates of rainfall rate and accumulation which is essential to achieve accurate measurements and therefore recognize intense precipitation and issue warnings. Radar rainfall estimators have a number of advantages over gauges including the ability to observe precipitation over wider areas within shorter timeframes and providing advanced warning of impending precipitation events. The radar reflectivity-rainfall (Z-R) relations are traditionally used for quantitative precipitation estimation (QPE). The primary area of interest in this study is concentrated around the National Oceanic and Atmospheric Administration (NOAA) Hydrometeorology Testbed (HMT) in the Russian River basin north of San Francisco, CA (figure). This watershed covers approximately 1500 square miles and has an annual average discharge of around 1,600,000 acre-feet. In this mountainous terrain, the challenge of obtaining reliable QPE's is limited by beam blockage and overshooting, and orographic enhancement. Even if a perfect empirical Z-R relation can be applied, the accuracy is subject upon factors such as: radar calibration, ground clutter, attenuation, beam blockages, bright bands and anomalous propagation, etc. In development of Z-R algorithms, rain gauges provide ground truth to the estimation of Z-R coefficients for a given region. In this study, radar data is taken from the surrounding NEXRAD WSR-88D radars (KMUX, KDAX, KBHX, and KBBX) as well as the C-band TV station radar KPIX. The rain gauges used for ingesting and for independent comparison come from the Automated Local Evaluation in Real Time (ALERT) and the NOAA HMT gauges. The rain events used occurred throughout the year during months of March, April and December of 2012. The analysis evaluates the performance of precipitation estimation from the National Mosaic and QPE (NMQ) algorithm package, which is developed by the National Severe Storms Laboratory (NSSL) and the Multisensor Precipitation Estimator (MPE) developed for use within the Advanced Weather Interactive Processing System (AWIPS). These results are compared to a simple analysis performed using only KPIX reflectivity to rainfall amounts.

2. METHODOLOGY

The QPE precipitation fields, generated by NMQ, evaluated in this analysis include: gauge only, multiple radar-only, multiple radar with vertical profile of reflectivity (VPR) correction, and multiple radar with VPR and gauge correction. Along with radar input, 180 gauges are used by NMQ for gauge analysis. The computed QPE output are gridded into common latitude-longitude coordinates and compared to an independent HMT gauge set consisting of 17 gauges. The domain of interest for this study is shown in the figure. The independent HMT gauge QPE is created by gridding hourly gauge data into lat-lon using inverse distance weighting (IDW). Statistics are calculated by comparing the common grid points from the NMQ output with the independent gauge IDW QPE grid using b=2 and a 2km radius of influence. In addition to the NMQ QPEs derived using the four NEXRAD radars (KDAX, KMUX, KBHX, and KBBX – see Figure), the QPE was derived using the reflectivity fields from a T.V. broadcast C-band radar (KPIX) that scans the same area of interest but, unlike the NEXRADs, is closer and has a nearly unobstructed view of the Russian River basin. The Martner Z-R relation (Z=44R^1.91) was used since it was derived in the CA coastal mountain range in the vicinity of the analysis region.

MPE QPE fields were generated using Digital Precipitation Array (DPA) files from the same radar set and gauge data input is mentioned above for NMQ. The PRISM (Parameter-elevation Regressions on Independent Slopes Model) climate mapping system was also used for scaling the multisensory estimates in MPE. The MPE output generated is a 4km by 4km grid in Hydrologic Rainfall Analysis Project (HRAP) grid system, which is then converted to lat-lon and then compared to the independent gauge data. The same IDW scheme was used to re-interpolate the MPE products to 1km by 1km grids for cross comparison with NMQ measurements.

To gain a simple insight into rainfall over the Russian River watershed, the KPIX radar reflectivity fields generated over the domain and the Martner relation are used to create QPE fields, and then statistics are generated and compared along with NMQ and MPE results.

3. ANALYSIS

Several scenarios were evaluated in the analysis of NMQ and MPE QPE outputs. These were to consider the use of radar input into these systems. One case was to use only the NEXRAD radars, another is to only consider the impact from the nearest radar with the best coverage (KPIX) and lastly to consider all radars. In the simple sense, the QPE fields were then calculated using the Martner Z-R using the raw KPIX reflectivity observations. All QPE results were compared to the independent gauge IDW QPE rainfall maps. Preliminary results from a rainfall event show the correlation coefficient is highest using KPIX only as input in the NMQ system. The least correlated QPE results when adding KPIX as input along with NEXRAD radars. Research is on-going to determine why the KPIX-only results are so much better compared to the multiple radar results. Further work will be done to improve the QPE analyses.