| Spatially continuous,high spatial and temporal resolution surface evapotranspiration(ET)products are of great significance for regional fine water resources management and planning.The current remote sensing ET products are limited in practical applications due to the limitations of field station observation,spatial and temporal resolution of remote sensing data and the lack of remote sensing observation data caused by cloudy and rainy weather.This paper focuses on this problem.In the Haihe River Basin,where the seasonal distribution of water resources is uneven,firstly,the MODIS surface temperature data are downscaled by the random forest algorithm to obtain the 250m resolution surface temperature of the Haihe River Basin.Secondly,we used the downscaled surface temperature and vegetation index to establish a triangular feature space based on the P-T equation for remote sensing estimation of instantaneous evapotranspiration in the Haihe River basin.The evaporation ratio invariance method is also used to extend the instantaneous evapotranspiration to daily,monthly and annual time scales,and the results are validated using station data.Finally,the water use efficiency of the Shijin irrigation area was calculated using the calculated 250 m spatial resolution evapotranspiration data and MODIS GPP primary productivity product data.The main conclusions are as follows:(1)Surface temperature downscaling.The random forest algorithm was applied to obtain the key input variable of the remotely sensed evapotranspiration model–land surface temperature(LST).We used the MODIS remote sensing product data at multiple resolutions,combined with the constructed random forest downscaling model,to obtain 250 m spatial resolution surface temperature data without missing pixels in the Haihe River basin.The validation results show that the LST errors of Daxing,Miyun and Tantao stations are 3.86K,2.37K and 1.96K respectively.(2)Estimation of regional surface evapotranspiration by remote sensing and its validation.The daily,250 m resolution surface evapotranspiration in the Haihe River basin from 2008 to2010,calculated by the spatial evapotranspiration model with surface temperature-vegetation index characteristics,showed reasonable results in terms of spatial and temporal distribution characteristics.In terms of time,the inter-annual variation of evapotranspiration in the Haihe River Basin from 2008to 2010 is small.The seasonal variation in evapotranspiration in the Haihe basin is large,with a steady increase from January to August,followed by a decline after the summer maize filling period.Spatially,the high values of evapotranspiration in the basin are mainly in the middle reaches of the Luan River basin,where precipitation is abundant,the southern Zhangwei River Plain irrigation area,and the North China Plain irrigation area,where the average value is mostly above 580mm.In the southern part of the Haihe basin,the Maqi River,Ziya River,Shijin Irrigation Area and Zhangfu River Irrigation Area are also the main distribution areas with high values of evapotranspiration,which are above 520 mm.In Beijing,Tianjin and the Yongding River in the north-west,the main areas of the Haihe basin with low values of evapotranspiration are basically below 360 mm.In terms of different Surface cover types,the mean value of evapotranspiration from cultivated land in the Haihe basin is the highest,at about 579 mm;from scrubland,at about 508 mm;from grassland,at about 450 mm;and from construction land,at about 412 mm.In comparison with the station observations,the annual scale values are slightly lower than the station monitoring data.The average evapotranspiration from 2008 to 2010 at the Tatau station was about 608 mm,and the remote sensing estimate of evapotranspiration was about 547 mm,which was about 11%smaller than the station's eddy-related system monitoring.The average evapotranspiration from 2008-2010 at Daxing station was 535mm,and the estimated evapotranspiration from remote sensing was about 501mm,which was about 6.7%smaller than that from the station's eddy-related system.The average evapotranspiration from 2008-2010 at Miyun station was 572mm,and the evapotranspiration estimated by remote sensing was about 490mm,which was about 14%smaller than the monitoring by the station's eddy-related system.(3)Water consumption characteristics and water use efficiency of agricultural fields in the Shijin Irrigation District.A typical irrigation area in the Haihe River Basin,the Shijin Irrigation District,was selected to carry out the application of surface evapotranspiration in agricultural water consumption monitoring.The water consumption patterns of crops in the Shijin irrigation area during different phenological periods and different land use types were analysed.Specifically,27.27 mm of evapotranspiration was recorded during the greening period when there was no or little precipitation,and a small amount of irrigation was seen to be sufficient at this time.The evapotranspiration is higher(64mm-116mm)at the nodulation,gestation,tassel and flowering stages,so it is appropriate to increase the amount of irrigation.The evapotranspiration of maize at sowing and emergence is 115mm,which should be irrigated appropriately,and at the nodulation,tassel and flowering stages the evapotranspiration is85mm-115mm,which should be irrigated more.In addition,the water use efficiency(WUE)of the Shijin irrigation area was calculated in conjunction with the MODIS GPP product.The results show that the highest WUE in the Shijin irrigation area is 1.58g C/kg H~2O and that the highest area is in the southern part of the Shijin irrigation area,where the land use type is arable land.Water use efficiency is low. |
PDF Full Text Request