Retrieval of precipitable water vapor over land using TRMM/VIRS thermal split window measurements

Abstract

Water vapor plays a crucial role in atmospheric processes that act over a wide range of temporal and spatial scales, from global climate to micrometeorology. Determining water vapor distribution in the atmosphere and its changing pattern is very important. The algorithm based on some satellite remote sensors is mature (i.e., moderate resolution imaging spectra-radiometer (MODIS)). Water vapor inversion algorithms based on other sensors remain in the scientific research stage or has no corresponding water vapor algorithm (i.e., Tropical Rainfall Measuring Mission (TRMM) and Visible and Infrared Radiometer System (VIRS)). TRMM/VIRS data were widely used to study precipitation. This paper uses thermal infrared split window channels at 10.8 μm and 12.0 μm of VIRS to retrieve Precipitable Water Vapor (PWV). An improved physically based algorithm for the retrieval of PWV over cloud-free land surfaces was applied in this paper. First, the Split-Window Covariance-Variance Ratio (SWCVR) method was reviewed. The surface emissivities of the two split window channels were assumed equal. Moving window method was adopted to keep the spatial resolution of the original data. Then, an operational use of this method was developed and applied to VIRS datasets. Cloud liquid water information obtained from TRMM Microwave Imager (TMI) was used to identify the clear sky area. Given that TMI and VIRS are both mounted on the TRMM satellite, the data obtained by the two instruments are consistent in time and space and avoid data match problems. A total of 2000 radiosonde profiles were input into MODTRAN to simulate the brightness temperatures under the configuration of VIRS and find the relationship between the transmittance of the two split window channels and PWV. The profiles used in this research were randomly selected from the land sounding sites around the world to represent all kinds of water vapor condition and types of surface. The profiles were evenly distributed in the four seasons, although winter has slightly fewer profiles. Compared with the GPS results, the root mean square error of the results is 5.76 mm, and the bias is -1.2 mm for the research area. Regional consistency was found between the results obtained by MODIS and the proposed algorithm. The proposed algorithm can yield reasonable results that are accurate in most cases with a split-window technique using VIRS data. Validation results indicate that the PWV retrieved by VIRS has high precision, and has a reference meaning to China FY data for retrieving PWV based on infrared split window channels. However, the precision of the algorithm in this paper was lower than the PWV results retrieved by MODIS near infrared data. The infrared channels were sensitive to the PWV in the upper atmosphere, whereas most of the PWV existed in low atmosphere. Accuracy was almost the same between the PWV retrieved by the infrared data of MODIS and the results in this paper. Few preliminary results were obtained by the present study, but the existing algorithm can be further developed and improved to reach the degree of business.