The Reconstrution Of LST And It's Applicationin The Spatial Distribution Of Air Temperature

Temperature is treated as one of the most important parameters in conducting research on global climate change,natural disaster monitoring,agricultural zoning and ecological environment.Traditionally,obtaining spatial continuous distribution of temperature from spatial interpolation or drawing contour with observed values from meteorological stations could brings about uncertain result of spatial distribution simulation and limited application considering the uneven meteorological station distribution and complicated geography conditions of the target research areas.Therefore,simulating spatial distribution of temperature based on remote sensing data has become an important research area.However,the reliability of remote sensing data and simulation result is largely reduced affected by solar elevation angle,sensor and cloud coverage and so on.If we reconstruct long-term remote sensing data and take serious consideration of the common factors on temperature including geography and terrain factors,the accuracy of spatial distribution simulation of near-surface temperature will be further improved,which will be more meaningful in the application of spatial distribution of temperature in rolling areas.It mainly contains the following four aspects of research work in this paper:(1)Data quality evaluation of land surface temperature in Chongqing.Having referred to monthly and annual cloud average,sunshine percentage and LST data quality control documents in Chongqing,it turns out that the annually average sunshine percentage is between 21%~31% and that total cloud amount of annual average is 7.7 from 2006 to 2015.Compared to the LST data from Aqua,the availability of LST data from Terra is better and the availability of LST is the best in 2013 and the worst in 2012.(2)Reconstruction of long-term LST data in Chongqing.Due to frequent cloudy and misty days in Chongqing,there are large amount of missing value about LST.In this paper,it focuses on long-term sequence reconstruction from 2006 to 2015 using the Hants method and the Savizky-Golay filtering method.The construction result indicates that Savizky-Golay filtering method superior to Hants method in terms of reconstruction of LST series data in the daytime and at night while the reconstruction accuracy of two methods are within 2?at nightand about2? in the daytime.Besides,the reconstruction result is better at night than it in the daytime.And the better S-G filter results could serve as the data source of simulating temperature.(3)Sequential spatial distribution simulation of monthly temperature in Chongqing.In this paper,it focuses on the correlation between temperature and the influencing factors like meteorological data and remote sensing surface parameters and then I build the multiple linear regression models of temperature with meteorological site geographic data as well as temperature with remote sensing surface parameters.Then,I build multiple linear regression model of temperature and the selected influencing factors.By analyzing the goodness of fit and difference of these three models and verifying the accuracy the multiple linear regression model with actual validation method,it indicates that the RMSE of the monthly temperature is between 0.67?and 1.21?.Therefore,I decide the result of multiple linear regression model as the final result of the spatial continuous distribution simulation of monthly temperature.(4)Making quantitative analysis of spatial temperature distribution based on the altitude and land cover types in January,April,July and October 2015 in Chongqing.It turns out that in January,April,July and October,2015 of Chongqing,the monthly temperature value including the minimum,maximum and the average value could decrease when the altitude increases.Meanwhile,the monthly temperature first increased and then decrease at different altitudes.Besides,the maximum of monthly temperature value is closer in January,April,July and October 2015 while the average temperature value is slightly diverse in different land-use types separately building area> water body> agricultural land > bare land> vegetation.There are large differences between the minimum monthly value and the maximum monthly value in different land-use types.Among the differences,it has the smallest difference in construction area then the water body,but the vegetation and bare land and agricultural land is close to each other.