| China is a large country with water resources,and the total amount of water resources ranks the sixth in the world.However,due to the large population,the per capita water resources in 2020 are only 2239.8 m3,and the utilization efficiency of agricultural water resources in China is relatively low compared with that in developed countries.Therefore,monitoring the irrigation area and planting structure has been paid more and more attention to the rational allocation of water resources.In this study,Hetao Irrigation District in Inner Mongolia was taken as an example.Based on multi-source remote sensing data and ground acquisition information,the planting structure and comprehensive irrigation information of the irrigation district were monitored by remote sensing through decision tree supervision and classification,peak-peak-valley detection program and difference model,and the accuracy of the obtained results was analyzed and verified.The main conclusions of this paper are as follows :(1)Based on long-time satellite remote sensing data and ground measured data,by analyzing the sensitivity of different remote sensing spectral information to crop classification,a crop planting structure extraction model is constructed to obtain highprecision crop planting structure information in Hetao irrigation area.In the study,three groups of different spectral information inputs(NDVI time series data,NDVI and original band combination data set,NDVI and statistical information data set)and two classification algorithms(RF random forest algorithm and C5.0 algorithm)were considered to classify and verify the accuracy respectively.The overall classification accuracy was the highest in the RF random forest algorithm and the combination of NDVI and statistical information data set,with the accuracy of 96.16 %,which proved that the statistical information data and random forest algorithm were highly sensitive to crop classification in the study area.The research results provide basic data for subsequent irrigation area extraction.(2)Combined with Sentrinel-2,3 data,the VTCI time series data set of conditional temperature vegetation index with temporal resolution of five days in the period from April1 to November 20 was constructed.Then,based on the peak detection algorithm of VTCI time series,the peak time points of urban area in 2020 and the peak and valley time points of cultivated land(i.e.,the crop range in the classification of planting structure)were obtained.Through the comparative analysis of the two time points,the comprehensive irrigation information such as the number,time and cycle of irrigation in the year was obtained.Compared with the historical irrigation data,the five irrigation cycles obtained are consistent with the local irrigation mode and the irrigation time obtained in the field.(3)Combined with the obtained irrigation time,period and other information,the difference model was used to extract the irrigation area in the irrigation area,and the applicability of the extraction results of optical data and radar data in the study of irrigation area in Hetao irrigation area was compared and analyzed by using the extraction accuracy.Firstly,the difference model was used to extract the irrigation area before and after the irrigation time point on April 30 by using Landsat-8 optical data and Sentinel-1 radar data respectively.The accuracy verification of the extracted results by the measured data shows that the accuracy of the irrigation area obtained by Sentnel-1 data is better than that of Landsat-8 data,which is due to the fact that Sentinel-1 radar data is not affected by cloud and weather factors,and the temporal resolution is also due to Landsat-8 data.Therefore,the results of Sentinel-1 data extraction are used as the extraction results of irrigation area in Hetao irrigation area under the influence of both temporal resolution and cloud cover,and the area under five irrigation time points is extracted step by step. |
PDF Full Text Request