Acquiring Regional Thermal Environment Information And Its Relationship With Land Use Type

Temperature is not only a key indicator of regional thermal environment, but also an important parameter in the global climate change research. And the accurate obtaining of the temporal-spatial quantitative temperature information has important scientific value. Common used temperature types are near surface air temperature and land surface temperature. At present, the air temperature data was mainly obtained from the point observation at meteorological stations which were interpolated to surface for spatial air temperature. But the number of stations, station distribution and interpolation method caused uncertainty in the result, particularly for stations in scarce areas, the accuracy of temperature interpolation is difficult to guarantee. Thermal infrared remote sensing can obtained the spatial land surface temperature at the regional scale through quantitative inversion. Although there is a relationship between the land surface temperature and near surface air temperature and we can further obtain the surface air temperature from the land surface temperature, many problems such as the calculation method and the stability still exist, and it need further study. In addition, the air temperature significant affected by the land surface characteristics and the difference land use type is one important factor for the air temperature spatial heterogeneity. Therefore, in order to better analyze the regional thermal environment, it is necessary to compare and analyze the relationship between the land use types and air temperature.This paper has taken Anhui province, which has complex topography, as the study area. At the methods to obtain the spatial air temperature, this paper assesses the ANUSPLIN interpolation algorithm based on meteorological stations and the empirical statistical model based on remote sensing inversion. Meanwhile, for the relationship between land use and air temperature, combined with multi-source spatial data for example land use data, population grid data, digital elevation data, the division of urban and rural meteorological stations was achieved. Then the heat island effect of the pixels at different land use types was compared and analyzed.The main results of this study are as following:(1) We established ANUSPLIN optimal interpolation model in different types of temperatures and different time scales from 2001 to 2005 in Anhui Province. Analysis showed that from the view of temporal scale, with the temporal scale increases, the model accuracy was improved at first and then decreased. The highest accuracy appeared on the monthly scale. From the view of variables, using elevation as interpolation covariate had greatly affected the interpolation result, especially in high temperature area. Results in annually and monthly scale represented obvious spatial temperature gradient change, while spatial difference was greater in daily scale.(2) Empirical statistical models were built to estimate near surface air temperature, based on different time-scaled MODIS remote sensing data. Analysis indicates that optimal parameter combination was different in various time-scales in different temperature. Month-scaled regression precision is higher than that of daily-scaled. Compared with AQUA, TERRA independent variables have higher regression precision.(3) BP artificial neural network was employed to build a nonlinear model between the land-use types 5 km around the meteorological stations from remote sensing and their rural/urban property. The simulated result fitted the stations' rural/urban property well. This model avoids the limitation from administrative regions'population census data and reflects the combined action of land use. Thus accurate partition of urban-rural meteorological stations is implemented. On this basis, background temperature field is built and Anhui's urban heat island intensity of annual mean temperature, annual mean maximum temperature as well as annual mean minimum air temperature are 0.582℃,0.139℃and 0.794℃respectively in 2005.(4) This paper statistics the mean temperature of pixels of various land use types, and comparatively analyze their urban heat island. It shows that the change tendencies of urban heat island intensity, which calculated by interpolation algorithm of meteorological station based on ANUSPLIN,empirical statistic model of remote sensing inversion, and the difference between urban meteorological stations and rural meteorological stations, are mainly similar.The creations of this paper mainly include three aspects as follows:(1) BP artificial neural network was employed to build a nonlinear model between the land-use types 5 km around the meteorological stations from remote sensing and their rural/urban property, which is a helpful attempt to distinguish the urban stations from the rural ones correctly.(2) The analysis and comparison among three sets of spatial air temperature data of regional thermal environment respectively based on the estimation of LST, the interpolation of AUNSPLIN and the difference between urban meteorological stations and rural meteorological stations are achieved. The change tendencies of urban heat island intensity are mainly similar based on the above three sets of spatial air temperature.