| Crop planting distribution and soil moisture status in irrigation area can help to understanding of soil moisture and provide reference for water transmission and distribution plan in irrigation area.Remote sensing technology provides a powerful tool for rapid acquisition of crop planting distribution and soil moisture status at a large scale.In the past,most researchers have separated the study of crop planting distribution and soil moisture status,but seldom considered them together.Using remote sensing technology to retrieve soil moisture requires a lot of processing of basic remote sensing data.Most of the existing researches focused on batch programming for a certain process,and seldom formed a systematic integrated processing program.In this paper,the Huilong irrigation area of Shanxi Province is taken as the research area,and the Landsat 8 remote sensing satellite data are used to study the main crop planting distribution and soil moisture estimation in the irrigation area.An application of automatic estimation of soil moisture based on ENVI 5.3 was developed.The main research contents and conclusions are as follows:(1)Using the Landsat 8 remote sensing image data from 2018 to 2019,the time series curves of NDVI and EVI of cultivated land,orchard,village and town were drawn.Taking "cultivated land-orchard-city-town" as the classification standard,the random forest algorithm was used to classify the land use patterns in the study area.For the cultivated area of the study area,the remote sensing images of winter wheat and summer maize in their growth periods were used to obtain the of the distribution of the main crops in the study area.The results show that NDVI and EVI have a good distinction among cultivated,land,orchard,and village area.The classification quality based on NDVI is slightly better than that based on EVI.The overall accuracy and kappa coefficient of classification method based on NDVI are 96.69% and 0.947,respectively.Based on the remote sensing images of winter wheat and summer maize during their growth period,the planting distribution of winter wheat-orchard,summer maize-orchard in the study area were divided.The classification results were compared with those of cultivated land and orchard.The results show that the consistency of orchard area division are 94.66% and 98.96%,respectively.It indicates that it is feasible to use the remote sensing images during each growth period to divide the distribution of crop planting.The proposed method can provide help for quickly obtaining the distribution of main crops in irrigation area.(2)The measured soil moisture data from March to July 2019 were fitted with the vertical drought index(PDI)and Temperature Vegetation Drought Index(TVDI)obtained from Landsat 8 remote sensing image processing.The trend of the fitting accuracy of PDI and TVDI and the factors causing the trend are analyzed.The results show that vegetation coverage degree is the main cause of precision change.NDVI,an index of vegetation coverage,can be used as the basis for the division of the applicable scope of PDI and TVDI models.And two combined models based on PDI and TVDI are constructed.Through the accuracy analysis of a combined model,it is found that the combined model is better suitable for soil moisture inversion in a long period of time.And it has a good accuracy in the depth range of 30 cm soil layer,and the effect is better at 0-20cm(the mean absolute error of 0-20 cm is no more than 1.12%,0-30 cm is no more than 1.41%).(3)Using the secondary development interface provided by envi5.3 software,the program of pretreatment of Landsat 8 satellite remote sensing image and automatic processing of soil moisture estimation is compiled.The application with interactive GUI interface is formed and applied.The results show that the processing results of the application are consistent with those of ENVI 5.3,which indicates that the software can basically realize the efficient and convenient processing of soil moisture estimation.The combined model of soil moisture estimation has good performance on shallow soil(over 30cm),but not on deep soil layer.And because the model is built on the basis of regional soil moisture data,it may have regional limitations. |
PDF Full Text Request