Evaluation of High-Resolution Tropical Weather Forecasts Using Satellite Passive Millimeter-Wave Observations

This paper presents the following: 1) a method for evaluating weather forecasts using satellite passive millimeter-wave observations and 2) tropical weather forecast evaluation for a weather forecasting system composed of the fifth-generation NCAR/Penn State Mesoscale Model employing its best physics options and gridded analyses and forecasts from the NCEP Global Forecast System used for initial and boundary conditions. Forecasts at 5-km resolution between 8 and 10 h after the initial time for 79 storm systems that are distributed over Thailand and nearby regions and span a year are evaluated using brightness temperatures (TBs) and precipitation parameters coincidentally observed by the passive millimeter-wave spectrometer Advanced Microwave Sounding Unit (AMSU) aboard the NOAA-16 satellite. The precipitation parameters including surface precipitation rate, peak vertical wind, and water paths for rainwater, snow, and graupel; the sum of rainwater, snow, and graupel; cloud liquid water; and cloud ice are from the AMSU MIT Precipitation (AMP) retrieval products. Despite the difficulty of weather forecasting in the tropics, results are encouraging. Forecasted TBs and precipitation parameters statistically agree with AMSU observations. The forecasted storm location, morphology, intensity, and area are generally similar to AMSU observations. The main forecast errors include storm location offsets and overforecasting of large ice particles. The weather forecasting system can provide useful high-resolution forecasts for the tropics 8-10 h in advance. Weather forecasts in the tropics will be substantially improved if storm locations can be accurately corrected by employing satellite passive millimeter-wave observations.