Font Size: a A A

Study On Estimation Of LST Under Cloudy Region In MODIS Images

Posted on:2013-06-29Degree:MasterType:Thesis
Country:ChinaCandidate:M LiuFull Text:PDF
GTID:2248330371488408Subject:Photogrammetry and Remote Sensing
Abstract/Summary:PDF Full Text Request
Land surface temperature (LST) is a very important surface process parameter. Acquiring the spatial and temporal distribution of regional LST in time is the need of the surface process and thermal infrared remote sensing applications (such as drought monitoring). However, in the case of cloud cover, land surface temperature cannot be directly retrieved from the remote sensing images. Many land surface temperature products of the existing satellite remote sensing data (such as MODIS) only can detect the cloud pixel and mark it as cloud pixel, didn’t estimated the surface temperature of the cloud pixel, result in the lack of LST in cloudy area. However, many application research (such as drought monitoring, vegetation growth and crop yield estimation) all need to completely acquire the space distribution of LST over a region. So, how to estimate LST of thermal infrared remote sensing images under cloudy pixel is a cutting-edge research problem.Today, only the microwave remote sensing can overcome the obstacles of cloud cover to acquire surface information, visible spectral remote sensing and thermal infrared remote sensing all have this problem. However, microwave remote sensing is very sensitive to surface roughness and surface moisture. Many scholars put forward lots of cloud removing method for visible spectral remote sensing, but these methods only can display image clearly. It didn’t analysis the problem from the physical perspective. In order to solve the problem of estimating LST under cloudy area and find a suitable estimation method, We choose to use MODIS land surface temperature data products which are used widely and select parts region of Anhui Province as a specific study area. We use two methods to estimate LST, including vegetation relationship method and space interpolation method and compare these methods in three aspects, such as land cover type, estimated range and the estimation accuracy.The vegetation method based on the theory that vegetation transpiration has an important effect on LST. At first, We analysis the relationship between NDVI and LST in cloudless area which is near to cloudy area and establish the equation; Base on the fact that NDVI characteristics relatively keep stable in a short, we use the NDVI value of another image to obtain the NDVI of cloudy area; Finally, We use the relationship to estimate the vegetation surface temperature of cloudy pixels. Spatial interpolation method is mainly based on spatial continuity characteristics of land surface temperature. In the study, We assume that a part of the cloudless area is covered by cloud and obtain the accuracy information by comparing the estimation LST with original temperature. At the same time, We estimated LST under cloudy area in different surface types and different sizes of cloud to get estimating surface types and cloud coverage of these two methods.In this paper, We use the MODIS LST products (June3,2009) in part region of Anhui province to obtain estimating precision of these methods, The results show that: The vegetation method can achieve better estimation accuracy of land surface temperature in small areas (less than100km2) for forest types, RSME=0.4Celsius degree; For larger areas, estimation precision of completely regularized spline is higher, can estimate LST when the cloudy area is about to2500km2and the root mean square error is about0.5Celsius degree; And the estimation precision in water and vegetation land cover types is high, the estimation mean square error separately reaches0.2Celsius degree and0.3Celsius degree when the cloudy area is about3x3 size (9km2). Therefore, when the MODIS LST cloud coverage is greater than100km, the completely regularized spline method is more suitable for estimating LST of cloudy area. It is should be noted that, in the actual application, the estimation results must be corrected for the influence of cloud cover. In this paper, we estimated the LST of MODIS LST products (June1,2011) in the actual cloud coverage area, the result image presents obvious depression effect, this phenomenon is consistent with the actual. In general, We can get high estimation accuracy (less than2500km2,the RSME≤1Celsius degree) of LST under cloudy area in MODIS LST products using the vegetation method and spatial interpolation method, It can make the MODIS LST products meet the application demand of people better, and make it widely applied.The main innovation of this paper is applying the relationship of NDVI-LST and the spatial interpolation method to estimating LST under cloud region in MODIS LST products, and discussing the practicability of these method in actual cloudy condition, At a result, We can fix up the problem that MODIS LST products lack of information under cloud region, and also make the products widely used.
Keywords/Search Tags:MODIS, Cloud Cover, Surface Temperature Estimation, VegetationRelation Method, Spatial Interpolation Method
PDF Full Text Request
Related items