Atmosphere Water Vapor Retrieval Over Clear Sky On Land Based On MODIS Thermal Infrared Splitting-window Band

Water vapor play an important role on the earth's atmosphere,which is of great value in the study of weather systems and the energy cycle of the earth.And its temporal and spatial distribution are influenced by many parameters such as time,terrain and latitude and longitude.Based on the analysis of the absorption of atmospheric gases by thermal infrared band,we studied the band characteristics of MODIS.MODIS has multiple water vapor absorption bands in the thermal infrared region of 3?14?m.It is necessary to take into account the thermal radiation received by the sensor and the attenuation effect of radiation through the atmosphere.So we select the 10?12?m that the strongest radiation band of surface and the weak absorption band of water vapor.The process of retrieval atmospheric water vapor relied on the original data that include TIGR(Thermodynamic Initial Guess Retrieval)atmospheric profile set and the MODIS31 and 32 band spectral response functions.The atmospheric water vapor and the atmospheric transmittance calculated by the atmospheric radiation transmission model MODTRAN(MODerate resolution atmospheric TRANsmission)were used to get the two order relation of the atmospheric water vapor content and the two band atmospheric transmittance ratio.According to the traditional split window method,the surface emissivity ratio was always set 1 when the ratio of atmospheric transmittance is calculated by brightness temperature of two band.In this paper,NDVI is used to improve the estimation of surface emissivity.The composition of mixed pixels is calculated by NDVI,and surface emissivity is indirectly obtained by mixed pixels.The MODIS image from May to September of 2015 in North America is selected as the main data source.The atmospheric transmittance obtained by the traditional method and the NDVI improved method is used in the two order relation to get water vapor content.The water vapor retrieval results from the split window method are compared with the GPS water vapor data and MODIS near infrared water vapor products with the root mean square error 0.766g/cm2 and 0.792g/cm2 respectively.The water vapor retrieval results from the split window method are compared the same validation with the root mean square error 0.641 g/cm2 and 0.551 g/cm2.It is obvious that new method is better than the traditional split window method.In addition,from the water vapor scatter plot,it is found that the traditional split window method has high water vapor content,and the NDVI improved method effectively reduce this phenomenon,especially when the water vapor is low.In this paper,both two methods are not considered many things in study,such as the surface temperature,the atmospheric structure,the TIGR profile of the original data and the sensor noise and so on.The result by MODTRAN simulation shows that the surface temperature 250K and 330K can cause 0.76g/cm2 water vapor error with the water vapor content set 4g/cm2.The atmospheric structure can cause 0.90g/cm2 water vapor error with the water vapor content 5g/cm2.And equivalent noise temperature difference about 0.1 K can also cause 0.1?0.2g/cm2 water vapor outcome bias.And it found that different evaluation has little impact on calculation with selected RAOB profile by different ground level.