A Study Of Meteorological Factor Interpolation Considering Geographical Semantics

In recent years,the detection of local missing values of spatial data for meteorological factors has attracted wide attention.For example,the land surface temperature(LST)has an important influence on the balance of water and heat and energy conversion between the earth's surface and atmosphere,and it is an indispensable important parameter.The estimation of LST has important applications in urban heat island effect,extreme temperature monitoring,drought monitoring,heat balance and radiation of the earth's surface,and global climate change.In this paper,a semantic interpolation algorithm for LST is proposed.By embedding hierarchical geographic semantics into spatial interpolation model and using semantic similarity to measure factor weights,Semantic Inverse Distance Weighting(S-IDW)is proposed and the spatial interpolation estimation is completed.Firstly,the remote sensing inversion of LST in the study area is carried out.The accuracy of the semantic interpolation algorithm and other algorithms is evaluated by combining the inversion results with the LST observation data.The main conclusions are as follows:(1)Based on the atmospheric correction method,the Landsat 8 TIRS data was used to perform LST inversion,and the LST data of the two study areas and their adjacent areas were obtained.Based on the measured surface temperature test method,the LST of the inversion was verified.The corresponding surface of the 0 cm LST and inversion was recorded by 53 meteorological stations in different time periods of 9 meteorological stations in Anhui Province.Temperature was analyzed by Pearson correlation and regression analysis.The study found that whether the average 0 cm LST,the highest 0 cm LST or the lowest 0 cm LST has a strong correlation with the surface inversion temperature of the corresponding site,and the regression effect is better.(2)Based on the classification of land use status(national standards)and its meanings,the related concepts of geographic entities are extracted and the ontology hierarchical network structure of land use status classification is constructed.The semantic similarity algorithm is used to calculate the semantic similarity applicable to this paper.degree.The calculation of semantic similarity provides the basis for incorporating semantics into IDW.(3)The S-IDW interpolation experiments were carried out in two sample areas under three different temperature environments.The interpolation results were evaluated by different statistical methods,Pearson correlation analysis and regression analysis.It is found that the S-IDW in the two study areas under low temperature environment tends to be the best fitting value in RMSE,MAE,MAPE and RVAR than the other four interpolation methods.In the high temperature environment,the S-IDW of Zone I is closer to the best fitting values in RMSE and MAE than the other four interpolation methods.The MAPE of S-IDW is only higher than that of Kriging,and the RAVAR of S-IDW is lower than that of IDW and Natural.The S-IDW of study area 2 is closer to the best fitting value than other four interpolation methods in RMSE,MAE,MAPE and RVAR.Under normal temperature environment,the S-IDW of study area 1 tends to be the best fitting value in RMSE,MAE and MAPE than the other four interpolation methods.The RVAR of S-IDW is smaller than that of IDW and pline,but closer to the best fitting value than that of Kriging and Natural.The S-IDW of study area 2 is closer to the best fitting value in RMSE and MAE than the other four interpolation methods.The MAPE of S-IDW is only larger than that of Natural.In RVAR,S-IDW is smaller than Natural,but larger than IDW and Kriging.In the three temperature environments of the two sample areas,the correlation between S-IDW interpolation temperature and surface inversion temperature is the strongest,and the regression effect is the best.(4)Generally speaking,S-IDW can interpolate the surface temperature more accurately under different temperature conditions in different regions,which is helpful to study the disastrous weather such as urban heat island and extreme temperature.How to use S-IDW to improve the interpolation accuracy of other meteorological factors will be the focus of future research work.