Analysis The Data Of Land Surface Emissivity And Optimization Study Of Land Surface Emissivity Remote Sensing Retrieval Method

As one of an important geophysical parameter, Land surface temperature (LST) plays an important role in the process of material and energy exchange between ground and air. An effective way to acquire surface temperature of large area quickly is through the remote sensing, surface emissivity is the essential parameter of remote sensing inversion of temperature, and also the most important one. However, emissivity is a variable related to multiple factors, its measurement and inversion has certain difficulty. Research shows that, in the8-12μm band, every0.01change of emissivity (e) lead to0.5K difference in land surface temperature inversion (Ts), fast and efficient acquisition of land surface emissivity of large area with high precision have been the pursuit of many scholars.In this paper, we analyzed the source of existing emissivity data and the spatial and temporal characteristics of MODIS emissivity products using the data of MODIS images, National Weather Station meteorological observation data, MODIS emissivity and surface type products and existing published emissivity data. Compare the MODIS emissivity products with the emissivity which was obtained by using the method of NDVI threshold. Improve the original NDVI threshold method, and put the emissivity results into the split window algorithm. The research work main conclusions are as follows:(1)The land surface emissivity is one of the important parameters in temperature inversion from thermal infrared remote sensing. Nowadays, increasingly mature emissivity inversion and measurement skills have provided us a majority of emissivity data, such as MODIS emissivity products, emissivity spectra database and so on. In this paper, we collected different features or land types emissivity data from various of materials and analyze the main sources of the emissivity data,with emphasis on the emissivity data set based on IGBP and USGS land cover classification system, the results show that:The records of the emissivity spectrum database about the environment factors and measured status are inadequate; The quantitative relationship between emissivity and the factors need to be studied; Different features or land types emissivity need defined; Lack of ground validation in the area of China about remote sensing emissivity data.(2)The interannual variations of emissivity was analyzed based on China regional MODIS emissivity product of ten years from2001to2010. The moderate-high emissivity regions makes up of40%-50%of Chinese surface land area. The inter-annual variation of moderate-high emissivity region is also distinct, with two peaks (in spring and autumn) and two troughs (in summer and winter). The inter-annual variation of high emissivity regions is very significant, with a peak in winter (10%), while onlyl%and2%in other seasons.Combined with the type of land use and NDVI data in Hubei Province, we analyzed the emissivity variation characteristics of different surface types within the year, and the relation between emissivity with temperature and precipitation environmental variables. It shows that, there is a clear line of emissivity between four land types, agricultural land＞grassland＞town land＞dense thickets, the fluctuation range of emissivity is less than0.001during the ten years. Related to surface vegetation growth status, the agricultural land emissivity shows the same trend of the change curve of NDVI in January to May. The emissivity of MODIS products in summer is lower probably for the reason of much rain in summer, cloud cover, high water vapor content, strong crop transpiration and so on. In addition, as a variable which related to many factors, there is no linear relationship between the precipitation and temperature variables with emissivity.(3) MODIS image of Hubei Province in May23,2010was obtained during the pre-process. From the comparative analysis of NDVI threshold method and MODIS emissivity product in classification scale, we found that the result calculated by MODIS day/night algorithm has low spatial resolution, low insensitive to land type, small spatial difference of agricultural land area emissivity, obvious difference of forest; while there is a distinct spatial difference in the agricultural area based on NDVI threshold method. Affected by the NDVI saturation, there is no difference in forest area. In this paper, combined with the deficiency of NDVI threshold inversion method, we improved the method from two aspects:the vegetation type and soil moisture. Finally the NDVI threshold improvement method was applied to split window temperature inversion algorithm proposed by Qan Zhihao to get temperature distribution map of Hubei Province. Compared the, MODIS products with the meteorological temperature data, we found that the temperature inversion from the improved NDVI threshold is more comparable.