Remote sensing of water vapor over land using the advanced microwave sounding unit
Water vapor is a fundamentally important variable in the atmosphere for making accurate forecasts. Its global distribution is a challenge to determine and can change rapidly in both space and time. Several ground and space based methods are currently employed to determine its spatial and temporal variability. The microwave spectrum is very useful for remote sensing due to its ability to penetrate through clouds at most frequencies. Microwave satellite sensors have been used to retrieve atmospheric state parameters for several decades, however the retrievals of certain parameters have not been performed satisfactorily over land thus far. Retrievals rely on the ability to extract the atmospheric state from the upwelling radiation, most of which comes from emission from the surface. Knowing the surface emissivity to a high degree of accuracy is essential for calculating the land surface temperature, however it is also important because this emission must be removed in order to retrieve the atmospheric parameters desired. Land type, vegetation, snow, ice, rain, urbanization effects, and many other factors have an effect on the aggregate emission within each viewing scene and results in a strong sensitivity and variability of microwave emissivity on small scales. A physically based iterative optimal estimation retrieval has been implemented to retrieve atmospheric parameters from the Advanced Microwave Sounding Unit (AMSU). This retrieval is based on the method of Engelen and Stephens (1999). The retrieval uses a first guess of water vapor and temperature profiles (currently from radiosondes, but will soon be from GDAS), and uses a first guess of emissivity at each of five frequencies (from the MEM). The retrieval was run with a highly accurate first guess in order to detect bias, and the total precipitable water amounts were validated against a radiosonde match-up dataset. The match-up showed fair agreement between the radiosondes and the retrieval (within 20%), however a systematic bias was detected due mostly to coastline contamination. Data from the Global Positioning System (GPS) was also used to validate the total precipitable water, however the results showed less agreement than the radiosonde results (variations of ~20-35%). Most of this disagreement stemmed from geographical co-location differences. The analytical Jacobian was also examined to determine the sensitivities of all channels to the state vector parameters. This enables any retrieval user to pick a channel configuration that gives the desired sensitivities. Vertical profiles of water vapor sensitivities at four varying emissivities were investigated. Sensitivities of water vapor to emissivity were also examined at three distinct atmospheric pressure levels. The Jacobian determined that water vapor is able to be detected throughout a vertical column with adequate skill, although problematic areas occurred between 600 and 800 mb as the emissivity approached unity (e>0.99) for a wet atmospheric case. These results give confidence that AMSU can detect TPW over land for both weather forecasting and for climate studies. The current capabilities may be improved further once bias sources are dealt with satisfactorily.
Water vapor, Atmospheric -- Remote sensing
aMicrowave remote sensing