| As a major meteorological disaster,drought has become a hot issue in the study of global climate change.Shaanxi Province is located on the edge of northwest China,the drought occurs frequently,which has caused great harm to the national economy of the region,especially to agricultural production.The development of remote sensing technology provides a new way for drought monitoring,on the basis of analyzing the spectral characteristics of vegetation and soil,combining the absorption spectrum of the water and using the MODIS short-wave infrared band to construct the drought index,a series of indexes have been developed to reverse the soil water condition and obtain good results.However,due to the relatively recent application of using short-wave infrared data to monitor droughts,few studies have analyzed the suitability of using MODIS short-wave infrared signals to monitor soil moisture across large areas.Meanwhile,an index which lacked in-depth analysis of the effectiveness and sensitivity of drought monitoringIn this paper,we take Shaanxi Province as the study area,four kinds of remote sensing drought monitoring indices were established based on the MODIS short-wave infrared band:NDII6,NDII7,NMDI and SWCI.The adaptability of four kinds of short-wave infrared drought Index in the study area was evaluated and analyzed by means of correlation analysis and regression analysis using multiscale SPI and measured soil relative moisture data,meanwhile the comparative study was carried out using NDVI and EVI.Then,based on the principle of index construction,the factors affecting the accuracy of the drought indexes are analyzed,such as slope type,soil texture,vegetation cover and so on.In the end,a comprehensive and multiple-factor short-wave infrared soil moisture drought monitoring model is constructed,and the accuracy of the model is analyzed and evaluated by means of average,standard deviation and absolute error.The main conclusions are as follows:(1)On the monitoring of meteorological drought,the three short-wave infrared indexes SWCI,NDII6,NDII7 and 1-6 months-time scale SPI all showed a significant negative correlation,the correlation coefficient was obviously higher than the two vegetation green indexes.The correlation coefficients of three indices and two-month-scale SPI were the largest,and were tested by 0.01-level significance.At the same time,the four indices NDII7,NDVI,NDII6 and SWCI had better response to the wet and dry conditions between years.(2)On the monitoring of soil moisture,all remote sensing indices are correlated with 10cm and 20cm soil relative moisture,but the correlation between indices and 10cm soil relative moisture is the best.The correlation between SWCI,NMDI,NDII6,NDII7 and soil relative moisture was better than that of vegetation green degree indexes.The results of linear fitting of different remote sensing indices and soil relative moisture are optimal soil moisture fitting method.(3)Through the study of the relationship between slope and band reflectivity,we know that the reflectance of sunny slopes is greater than that of the shade slope for MODIS 2nd,6th and 7th bands.From the study of the relationship between slope and vegetation index,it can be concluded that the values of NDII7 and SWCI would increase with the increase of soil moisture.At the same time,when the soil relative moisture is same,the NDII7 and SWCI of the sunny slope are larger than the shade slopes.Slope direction is an important factor affecting the accuracy of short-wave infrared drought monitoring.(4)Through the study of the relationship between soil texture and band reflectance,it can be concluded that the reflectance of MODIS 2nd,6th and 7th bands is inconsistent with the increase of sand,silt and clay content.In the same soil relative moisture level,with the decrease of soil sand content,two kinds of remote sensing indices NDII7 and SWCI showed increasing trend;with the decrease of the content of soil silt and clay,both kinds of remote sensing indices showed a decreasing trend.(5)Through the study of the relationship between vegetation coverage and NDII7 and SWCI,it can be concluded that the two kinds of remote sensing indexes have a significant increasing trend with the increase of vegetation coverage under the same soil relative moisture condition,and the vegetation cover can cause monitoring difference to the soil moisture using the same kind of remote sensing index due to different vegetation coverage.(6)From the statistical results of multivariate linear regression equation,it can be seen that overall,the comprehensive regression model of soil relative moisture with NDII7 and SWCI combined with slope,soil texture,vegetation cover and other factors has good effect.The comprehensive model can effectively weaken the influence of information such as topography,soil and vegetation on soil spectra,and can obtain more accurate soil moisture estimation results.This study analyzes the difference of surface reflectance caused by different slope directions different soil texture types and vegetation coverage,based on which the effects of various factors on drought monitoring of short-wave infrared drought Indexes were studied.Then we used multivariate linear regression method to construct the soil relative moisture estimating model combining the classification variable with single exponent,and the new model has better effect.The results of the study can provide new ideas and methods for regional drought monitoring. |
PDF Full Text Request