Remote Sensing Monitoring Of Late Frost Damage Of Wine Grape

Late frost is one of the main meteorological disasters that affects the growth of crops in spring,which seriously restricts the yield and quality of crops.The eastern foot of Helan Mountain is one of the main planting areas of wine grape in China,and it is also the area with frequent late frost damage.The researches show that the late frost damage has become the main disaster affecting the growth of wine grape in the eastern foot of Helan Mountain.The frequent occurrence of late frost damage often leads to the decline of wine grape quality,resulting in a lot of economic losses.Therefore,it is of great significance to strengthen the remote sensing monitoring of late frost damage of wine grape for disaster prevention and mitigation.At present,remote sensing technology has been widely used in the field of agricultural meteorological disaster monitoring,but there are few researches on monitoring late frost damage of wine grape based on remote sensing technology.Therefore,this paper took the wine grape production area in the eastern foot of Helan Mountain as the research area,and used multi-source remote sensing data to carry out remote sensing monitoring of wine grape late frost damage.(1)Remote sensing recognition and extraction of wine grape planting area based on multi-source remote sensing data.This paper used GF-1,Landsat-8 and Sentinel-2 data of2019-2020 long time series to build multi-source dataset,extracted spectral features,vegetation index features and texture features,sorted the importance of features through Gini index and determined the number of features,and finally selected the top 53 features to build the optimal feature set.On this basis,the random forest algorithm was used to optimize the feature set and classify.The results showed that,compared with the classification results obtained by inputting all the features,the accuracy obtained by the optimal feature set was higher,the overall accuracy and kappa coefficient were 90.47%and 0.89 respectively,the producer accuracy and user accuracy of wine grape were 91.09%and 90.22%,and the area accuracy was more than 90%.The total area of wine grape extracted was about 39837 hectares.(2)Land surface temperature retrival based on Landsat-8 and MODIS LST interpolation.In this paper,two algorithms were used to retrieve the land surface temperature in the thermal infrared band of Landsat-8.Based on the geographic auxiliary data,MODIS vegetation index products and daily land surface temperature products,the interpolation model was constructed to interpolate the missing pixels of MODIS LST.The validation results showed that the RMSE was within 2.5?,R~2 was over 0.75,and the result had high accuracy,which can be used for LST downscaling research.(3)Data fusion based on Landsat-8 LST and MODIS LST.Using ESTARFM algorithm to fuse the daily data of Landsat-8 LST and MODIS LST with the spatial resolution of 100m and1km respectively,the daily data of 30 scenes with 100m in the study area in April 2020 were obtained.Comparing the pixel values of LST data before and after fusion,the results showed that the LST after data fusion can reflect more detailed information and temperature difference between different objects.(4)Remote sensing monitoring of wine grape late frost damage.Based on the LST daily data obtained,the daily minimum temperature estimation model was constructed.Combined with the extraction results of wine grape planting area,the daily minimum temperature in the planting area in April 2020 was obtained.According to the evaluation index of wine grape late frost damage,the remote sensing monitoring of late frost damage in the planting area was carried out.The results showed that there were four frost in the study area in April.In the first three times,slight or moderate frost damage occurred in some areas,and most areas were basically not frozen.The fourth frost caused serious frost damage to wine grapes in the planting area,and the damaged area of wine grape suffering from severe frost damage reached 16381 hectares,accounting for 41.12%of the total planting area,about 8501 hectares in Yinchuan,Qingtongxia and Hongsibao total about 7393 hectares,and Shizuishan was the least affected area.Based on multi-source remote sensing satellite and ground meteorological data,a remote sensing monitoring model of late frost damage of wine grape in the eastern foot of Helan Mountain was constructed,which was of great significance for fast and efficient monitoring of late frost damage,estimation of affected area and research on disaster prevention and reduction.