| With the acceleration of urbanization in recent years, a large number of natural surface gradually replace manual impervious surface, land cover changes affect the urban heat environment, resulting in more significant urban heat island phenomenon, such as thermal effects.Surface temperature can be used to describe the Urban thermal environment, impervious surface cover, vegetation, water bodies, etc. are the main factors affecting urban thermal environment, and therefore the quantitative analysis of factors affecting surface temperature is important for improving the urban environment.The traditional method of experimental observation of spatial data is difficult to obtain surface temperature data of the entire city, and remote sensing model inversion method provides the only fast effective means to acquire large areas of the city ground temperature data.Although there is a certain gap between temperature estimated by BP neural network combined impervious surface, vegetation, surface water bodies to the sub-pixel inversion of surface and the actual field measurements of temperature, its accuracy is much higher than the method using only one single parameter to estimate surface temperature, meets the needs of thermal infrared remote sensing imaging simulation surface temperature images and remote sensing studies provides a theoretical basis and technical support.to alleviate the heat island effect in Guilin.For growing urban heat island phenomenon Caused by the rapid development of karst in Guilin city, for the Landsat TM satellite images of the covering study area we use support vector machine SVM to extract land use information,use tasseled cap transform to extract soil brightness index, vegetation greenness index, humidity index and surface parameters,use the model to extract the normalized difference vegetation index NDVI, ratio vegetation index RVI, amend the soil adjusted vegetation index and other indices MSAVI and Water Index MNDWI,use Artis single window algorithm to estimate the thermal infrared pixel scale surface temperature, take surface temperature factors as BP neural network input to estimates30m spatial resolution of sub-pixel surface temperature,Analysis1989-2006Guilin city land use change, tasseled cap transformation characteristic component changes, vegetation parameters, Water Index change on the mechanism of surface temperature.The results showed that, SVM can improve Karst urban land use classification accuracy of remote sensing information can effectively monitor changes in karst dynamic urban land use. SVM’s terrain classification accuracy and Kappa coefficient is the highest,the overall classification accuracy was91.7%, more than90%, Kappa coefficient was0.827, which is significantly higher than the artificial neural networks, decision trees and classification maximum likelihood method.During1989and2006Guilin urban land use types changed a lot, a substantial increase in building area, while significantly reducing the area of agricultural land, an area of small water bodies shrinked or even disappeared. Correlation of tasseled cap transformation of vegetation greenness indices GVI and the surface temperature is the highest, the highest correlation coefficient is0.8907, as the optimal parameters to analyze the impact of vegetation vegetation cover on land surface temperature; Surface temperature and vegetation index NDVI, RVI, MSAVI, GVI etc. showed a significant negative correlation. Surface temperature and water index was negatively correlated,with the increase of the water surface temperature index values decreased; With the1989-2006expansion Karst in Guilin city size, water area takes4.6%of the total area fell to4.4percent, reducing the amount of0.2%of the total area,which takes4.3%of the total water area; Guilin Two Rivers and Four Lakes plays a major role to improve the ecological quality of the environment of the central city and ease the urban heat island phenomenon. |
PDF Full Text Request