Risk Assessment,Remote Sensing Monitoring,and Loss Assessment Of Cold Damage On Maize In Shaanxi,Gansu And Ningxia Provinces

Shaanxi Province,Gansu Province and Ningxia Hui Autonomous Region(abbreviated as Shaanxi,Gansu and Ningxia in this paper)are the main planting areas of maize in Northwest China over the past two decades.Cold damage is a common agro-meteorological disaster,which is one of the important environmental factors that limit the yield quantity and quality of crops.Remote sensing technology has significant advantages in space and time compared with traditional manual surveys,and many studies have applied remote sensing technology to the monitoring and assessment of agricultural disasters.At present,many researches on the cold damage of maize using remote sensing technology are mainly focused on Northeast China.While systematic research on the risk assessment,monitoring and loss assessment of cold damage on maize in Shaanxi,Gansu and Ningxia provinces are rare.Maize was selected as the object of this study.Meteorological observation data and spatio-temporal remote sensing data from satellites over the years were used for risk assessment,remote sensing monitoring and loss assessment research on the cold damage of maize with some methods and technologies.These methods and technologies include comprehensive weighting methods,time series analysis,decision tree classification,passive microwave inversion temperature,scale reconstruction,temperature estimation,crop model simulation etc.The main research achievements were as follows:1.Extraction of maize planting areas in Shaanxi,Gansu and NingxiaThe MODIS Land Cover Type product of Shaanxi,Gansu,and Ningxia was applied to obtain the mask layer of potential cropland through GIS technology.The NDVI remote sensing data combined with the main crop growth stage of Shaanxi,Gansu,and Ningxia was used in the decision tree method for classification,which can comprehensively judge the pixels based on the differences in the NDVI values of maize and other crops during the growth stage.By using this method we obtained the planting area of spring maize and summer maize of Shaanxi,Gansu and Ningxia.A comparison of results obtained in this study to reference statistical data of maize planting area shows an average absolute value of the relative error for estimated areas of 9.26%,,and this can basically meet the needs of cold damage monitoring and loss assessment on maize.2.Risk assessment of cold damage on maize in Shaanxi,Gansu and NingxiaIn line with the risk assessment theory of agro-meteorological disasters,meteorological data from 1971 to 2016 were used with GIS technology and mathematical statistics methods to establish risk assessment indexes for cold damage on maize in Shaanxi,Gansu and Ningxia.The risk assessment of cold damage on maize in Shaanxi,Gansu,Ningxia was mainly studied from three aspects:environmental susceptibility of hazard on maize,risk of disaster factors,and the vulnerability of disaster-bearing body.The integrated weight based on the AHP(Analystic Hierarchy Process)method was used to determine the weight coefficients of each component of the comprehensive risk indexes,and the index weight is used to establish a comprehensive evaluation model for cold damage risk on maize.The comprehensive risk index was used to divide Shaanxi,Gansu and Ningxia into low,medium,medium-high and high levels of cold damage risk areas.The geostatistical method was applied to obtain the risk assessment of cold damage at the county level in Shaanxi,Gansu and Ningxia.The statistical results were used to validate the risk assessment of cold damage on maize.There was a significant correlation between the comprehensive cold damage risk index and the yield reduction of 0.05,indicating that the risk assessment method can be referenced in practical application.3.Construction of integrated spatio-temporal datasets by retrieving land surface temperature with AMSR and merging images of MODIS LST products in Shaanxi,Gansu and NingxiaFirstly,MODIS daily LST datasets covering the period 2003-2016 and obtained from two platforms and two phases of four scene images to complement each other,were used to obtain MODIS daily LST fusion image.In order to fill missing data gap from daily land surface temperature of MODIS products,a passive microwave radiometer was used to perform continuous spatio-temporal temperature retrieval on the basis of the microwave radiative transfer model.The radiation model was simplified to vegetation canopy,soil and atmosphere as the underlying surface.The simplified equation was used to obtain a functional relationship containing only three parameters.The equations of the three functions were iteratively solved,and the solved parameters were input into the radiative transfer model to retrieve the land surface temperature(LST).By comparing with the surface temperature of the selected assimilation data,the approximate accuracy of the model is about 2.5-3K.Afterwards,the resolution of retrieved results was increased to 1km through scale reconstruction,and the missing pixel of the MODIS product is interpolated by the sharpened data to construct a integrated spatio-temporal land surface temperature dataset.Validation of the interpolated LST from 2003-2016 has the RMSE from 0.78-3.17K.4.Remote sensing monitoring of cold damage on maize in Shaanxi,Gansu and NingxiaDue to the spatial discontinuity of the air temperature data from limited meteorological stations,the integrated spatio-temporal LST datasets were constructed by coupling the surface temperature retrieved from 2003-2016 of AMSR with the MODIS surface temperature products were used to calculate the air temperature at different land type.These datasets were analyzed and selected to establish different linear multivariate regression equation for different surface types based on land surface classification.The estimation accuracy index RMSE of the air temperature simulated model is about 1.62℃ in the land type of cropland.Based on the temperature dataset obtained by this estimation method,three types of cold damage on maize indexs(index A,index B,and index C)were compared by correlation,and the index C has the best correlation with low-temperature disaster statistics was selected to monitor the cold damage on maize of Shaanxi,Gansu and Ningxia.The monitoring results show that the three provinces(districts)of Shaanxi,Gansu and Ningxia in 2004,2010,2014 and 2008 have suffered from most severe cold damage on maize in recent ten years.There is a good agreement of results obtained with the low temperature disaster records of the statistical yearbook of agro-meteorological disasters.5.Loss assessment of cold damage on maize in Shaanxi,Gansu and NingxiaThe loss assessment of cold damage on maize in Shaanxi,Gansu and Ningxia was carried out based on the daily coverage with remote sensing monitoring of cold damage.The yield loss of cold damage on maize was simulated by using the universal crop model WOFOST.The model parameters were first corrected for regionalization by using the model parameter optimization program based on the yield trend.The data of meteorological station,ground observation and query literature were used to calibrate the crop model parameters.The adjusted growth parameters of maize were applied as input data combined with the localization improvement results,and the results of different light and temperature conditions were compared with the output of the WOFOST model.The yield loss of cold damage on maize in 2004 and 2010 of 8 counties in Shaanxi Province was calculated respectively.The research explored an available approach by using crop models for cold damage assessment as single disaster.