| Evapotranspiration(ET),which includes soil evaporation and vegetation transpiration,is an important part of water balance and energy balance in agro-ecosystems.In the agricultural hydrology research of irrigation districts,continuous ET variation data is usually used to study the continuous water consumption process of different crops during the growth period.Remote sensing inversion of ET is a common method for obtaining regional evapotranspiration,the resolution of the inversion results depends on the resolution of the original remote sensing image.However,under the current conditions of remote sensing technology,it is difficult for a single satellite data to have both high temporal resolution and high spatial resolution.For high temporal resolution satellite data such as MODIS and NOAA/AVHRR data,the spatial resolution cannot meet the needs of farmland scale research.High spatial resolution satellite data such as the Landsat series of satellites,Sentinel-2A/B,EO-1,etc.,cannot guarantee continuous data in time.The "time" and "space" contradiction of remote sensing data is the main factor limiting the application of remote sensing evapotranspiration model in agricultural hydrology in irrigation districts,how to obtain high spatial and temporal resolution evapotranspiration data for complex agricultural areas is the key scientific issue to be discussed in this paper.Based on the planting structure information and the assumed hydrological unit,this paper proposes a space-time fusion model CSAFM for regional evapotranspiration.The model performs time-space fusion on the ET inversion results of Landsat and MODIS,and constructs a high-temporal resolution evapotranspiration data set The fusion results of the model are verified by the measured data of the Bowen ratio and vorticity correlation system,and the fusion accuracy of the CSAFM model and the traditional ESTARFM model is evaluated.The main conclusions obtained in the study are as follows:(1)Based on the Landsat and MODIS series data,the evapotranspiration of the study area was obtained by inversion of the SEBAL model.The results of Landsat and MODIS inversion were verified by the measured data of the Bowen ratio and vorticity correlation system.The instantaneous evapotranspiration and measured instantaneous evapotranspiration coefficient of MODIS inversion were 0.68,and the root mean square error was 0.78 mm/d,the coefficient of determination of daily evapotranspiration and measured data is 0.74,the root mean square error is 0.84 mm/d;the coefficient of daily evapotranspiration and the measured data of Landsat inversion is 0.98,and the root mean square error is 0.46 mm/d.The verification of the two inversion results shows that the SEBAL model is suitable for the calculation of evapotranspiration in the Hetao Irrigation District.(2)This paper proposes a semi-physical process spatiotemporal fusion model CSAFM(Classification-based Spatiotemporal Adaptive Fusion Model)for ET data This model is based on a hypothetical hydrological unit,and estimates the planting structure of the crop within the unit through the sequence of the NDVI,and uses the planting structure to extract the abundance of each category in the "coarse" pixel,and to perform mixed pixel decomposition on the MODIS data.Unbiased correction,by filtering appropriate similar pixels,extracting the contribution of similar pixels to the central pixel,used to predict the central pixel value at the next moment Compared with the traditional fusion model using mathematical processes,the CSAFM model is based on continuous hydrological unit partitioning and actual crop planting structure,so the physical process of fusion of this model is more explicit.(3)The verification of the fusion results by point scale and regional scale shows that the CSAFM model has higher fusion precision and is superior to the traditional ESTARFM model.The coefficient of determination for the results of the 2015 fusion and water balance calculations was 0.61 and the root mean square error was 1.01 mm/d.The coefficient of determination of the results of the growth period in 2016 and the measured results of the Bowen ratio is 0.92,and the root mean square error is 0.46 mm/d.The coefficient of determination of the fusion results and vorticity correlation results during the growth period of 2017 was 0.89,and the root mean square error was 0.65 mm/d.Compared with the traditional fusion model,the fusion coefficient of the 2017 ESTARFM model and the eddy correlation test result coefficient is 0.83,and the root mean square error is 0.77 mm/d.Compared with the ESTARFM model the CSAFM model fusion results reduced the root mean square error by 0.12 mm/d and the relative error by 16%.(4)In this paper,the CSAFM model is used to fuse the evapotranspiration data during the growth period of Linhe District of Inner Mongolia from 2009 to 2018,and the evapotranspiration data set with high temporal resolution and high spatial resolution is obtained.The planting structure classification was used to analyze the interannual variation of evapotranspiration in different land types.The three main crops(com,wheat,sunflower)are,affected by irrigation,and their evapotranspiration varies little from year to year.The variation range is mainly 500-800 mm.Water bodies,sandy land and bare land are obviously affected by weather conditions,and their evapotranspiration varies greatly from year to year. |
PDF Full Text Request