Abstract:

Landsat thematic mapper(TM) has a thermal band(TM6)，and this band can be used for retrieving land surface temperature(LST).In this paper，typical surface features parameter values were retrieved base on Landsat 5 TM remote sensing data of May，2007 in the Maotiao River watershed.The NDVI(normalized difference vegetation index) can be calculated using TM3 and TM4 band data,the land surface emissivity(LSE) of TM6 can be obtained by using NDVITEM (NDVI Thresholds Method),and the LST(land surface temperature) can be retrieved by single channel algorithm.The classification of land use types uses the China land use/cover of remote sensing monitoring classification system(geographical resources of Chinese academy of sciences,1999).According to this system，the researched area can be divided into 5 class including the forest land，grassland，cultivated land，water area and urban and rural construction land,and 13 subclass by which NDVI and LST can be respectively statisticed.The results indicated that in the three main types of the land use，the forest land had the highest NDVI and the lowest LST，farmland had the lowest NDVI and the highest LST，and the features parameter of grassland was between forest land and farmland.Heat island effect was clear in urban construction land,and both LST and NDVI were low in waterbody.The analysis of land use types shows that the surface characteristic parameters in the subclass land use/coverage scale represented regularly change even in the same kind of land use type for the different structure and function.In the subclass of forest land,forested land had the highest NDVI and the lowest LST.The NDVI of shrub land，sparse woodlot and other woodlands diminished gradually while the temperature increasesd gradually.In the subclass of grassland，high cover grass had the highest NDVI and lowest LST and the NDVI of low cover grass was the lowest,while the temperature was the highest.The features parameter of moderate cover grass was between high and low cover grass.In the subclass of farmland，the NDVI of paddy fields was greater than the dry land,while the temperature of the former was less than the latter.In the subclass of urban and rural construction land，NDVI and temperature had no apparent reverse relation.In order to deeply understand the significant degree of mean differerce of LST and NDVI in any two land use types of subclass,a series of posthoc multiple comparison tests using GamesHowell method were conducted associated with each possible pairing of land use types of subclass in SPSS.Most pairings exhibited significantly different mean values of NDVI and LST,while 3 pairings including forested land and high cover grass，sparse woodlot and moderate cover grass，construction land and urban use land were not significantly different in both NDVI and LST.Overall，a significant inverse relationship between LST and NDVI was observed for the entire watershed.Comparisons of mean NDVI and LST associated with individual pairings suggest that the majority of land use types of subclass can be differentiated on the basis of both LST and NDVI,and it is meaningful to analysis with the scale of subclass.This research helps to deepen the understanding of physical characteristics of different land use types and enrich the experienced knowledge of image classification of the main land use types