Research On Changes Of Thermal Landscape In Qingdao

The study of urban heat island effect is significant to understand how human activitiesaffect the city climate, decrease the harm caused by heat island effect, and reduce socialeconomic loss. The paper systematically studied the distribution and dynamic changes of heatisland in Qingdao, the relationship between land surface temperature, impervious surface andNDVI, as well as the trend of heat island, with multitemporal Landsat TM/ETM+data. Themain achievements are as follows:(1) After the data pretreatment (geometric correction, radiation calibration andatmosphere correction), the study area was divided into three types(water, natural surface, andtown surface) using supervised classification, whose surface emissivity was estimatedaccording to the vegetation coverage. Then land surface temperature was retrieved based onradiative transfer equation.(2) The land surface temperature was divided into different levels using mean-standarddeviation method. The spatial distribution and dynamic changes of heat island wereanalysised. Further, the land-use types image was superimposed on the temperaturedistribution image. Then the research was carried out on detailed spatial distribution andvariations of thermal landscapes by landscape transfer matrix and landscape index. The resultsshow that the scope and intensity of the urban heat islan tend to expand and enhance. Theurban and rural construction area of secondary high temperature and high temperatureexpanded to encroached on the bare area of high temperature and secondary high temperaturearound, as well as various kinds of green land temperature landscapes.Some of the moderatetemperature and secondary high temperature area within the city translated into hightemperature region, with some hot spots expanding, and gradually developed into large heatisland centers.(3) The imperious surface coverage was estimated through linear spectral mixtureanalysis model using normalized reflectivity data. In order to avoid negative values in thecoverage imag, fully constrained method was used in the linear decomposition. Otheringredients within the imperious surface coverage were eliminated using temperaturedifference. Then the research was carried out on spatial distribution pattern of the imperious surface, and the relationship between the land surface temperature, imperious surface andNDVI. The results demonstrate that land surface temperature is positively correlated with ISP,and negatively correlated with NDVI. ISP is negatively correlated with NDVI. The landsurface temperature grows exponentially with ISP, which rises faster and faster as ISPincrease. That is to say, the temperature of high ISP region rises faster than that of low ISPregion.(4) Using BP neural network, the spatial distribution of heat island of2015years in thestudy area was predicted. The results show that the heat island effect tends to expand andenhance. Finally, combined with the characteristics of Qingdao city, the paper put forwardsome suggestions to relieve urban heat island effect.