A Study of the Characteristics and Assimilation of Retrieved MODIS Total Precipitable Water Data in Severe Weather Simulations

The Moderate Resolution Imaging Spectrometer (MODIS) provides total precipitable water (TPW) data over land and ocean using near-infrared reflectance at a resolution of 1km, which gives it the potential to significantly improve weather forecasts. This study determined the accuracy and biases associated with MODIS TPW data and investigated the impact of its assimilation into the Weather Research and Forecast Model (WRF). A comparison between MODIS TPW and Global Positioning System (GPS) TPW over the United States showed that the root mean square (RMS) differences between these two data sets was about 5.2 and 3.3 mm for infrared (IR) and near-infrared (nIR) TPW, respectively. The comparison also showed that there were biases for both retrieved IR and nIR TPW data. The MODIS IR TPW data were overestimated in a dry atmosphere but underestimated in a moist atmosphere. In contrast, the nIR values were slightly underestimated in a dry atmosphere but overestimated in a moist atmosphere. Linear relations were developed to correct the biases associated with these data. The bias correction of MODIS nIR TPW reduced the RMS difference to 2 mm. Comparisons with radiosondes in the US and Australia also showed a similar trend in the differences with MODIS TPW, but with a larger offset, which may be associated with the previously reported dry bias in radiosonde measurements. Two severe weather simulations, a severe thunderstorm system (2004) over land and Hurricane Isidore (2002) over the ocean, were used to assess the impact of assimilating MODIS nIR TPW data, as well as conventional observations and Special Sensor Microwave Imager (SSM/I) retrievals. One set of experiments was conducted for the thunderstorm case and two sets for the Isidore case. The use of conventional observations alone had almost no impact on the thunderstorm system over land, but conventional observations were able to improve the Isidore simulations, in particular in the direction of storm motion. The assimilation of original or biascorrected nIR MODIS TPW showed a slightly positive impact on simulated rainfall over Oklahoma, the region of interest, for the thunderstorm case. For Isidore, the simulated storm intensities were too weak or too strong depending on the start time of integration of the simulation and the error in the reproduced storm track. MODIS nIR TPW was able to enhance Isidore’s intensity when the storm track was reasonably simulated. However, compared with original data, the bias correction of MODIS nIR TPW did not show any improvement and even slightly degraded one of the Isidore simulations. The assimilation of SSM/I data had a positive impact on both severe weather simulations and the impact was comparable to or slightly better than that of MODIS data. In addition, the influence of SSM/I data on the Isidore simulation occurred earlier than that of MODIS data due to clouds over the storm region that reduced the quantity of MODIS data available.