| Evapotranspiration(ET)is an important factor for energy balance conversion,and it is also a necessary part of the water cycle on the earth's surface.The Hetao Irrigation Area is located in a typical arid area.The study of the temporal and spatial variation of evapotranspiration has a long-term impact on the reasonable irrigation of farmland,the effective use of water resources and the sustainable development of the regional economy.To reveal the different characteristics and influencing factors of evapotranspiration in different land types,the SEBS model based on the surface energy balance system was used to obtain meteorological data and Landsat remote sensing images as the source data to obtain the daily evapotranspiration in the 6-year growth season of Hetao Irrigation District from 2001 to 2018.The lysimeter,Vorticity-Correlation,Bowen ratio measured data and P-M model calculated values are used to analyze the results of the SEBS remote sensing model integrated with the Kriging interpolation algorithm,and analysis of change trend and accuracy test of inversion results.Combined with the land classification of the irrigation area based on the Softmax method,the spatio-temporal changes and differences of evapotranspiration of 6 land types in the irrigation area were analyzed.1.The average error of the relative error between the SEBS model irrigated inversion results of the improved Kriging algorithm and the calculated value of the PM model ranges from 9.31%to 14.03%,which is basically consistent with the change trend of the three measured ET values.It is closest to the observed value of Bowen ratio,and the average relative error is 14.93%.The root mean square error(RMSE)of the model estimation result and the measured value of the Bowen ratio is 0.774 mm·d-1,and the determination coefficient(R2)is 0.946;the root mean square error(RMSE)of the model estimation result and the calculated value from the PM model is 0.552 mm·d-1,the determination coefficient(R2)is 0.931.The inversion results of the improved SEBS model are consistent with the time-varying trend of evapotranspiration calculated by the PM model.2.The 2017 Landsat image classification results based on the Softmax method are basically consistent with the spatial distribution of Sentinel 2.The overall classification accuracy of the classification results for many years is higher than 83%,and the Kappa coefficient is higher than 81%,indicating that the classification credibility degree of the Softmax classification method is higher.3.Daily evapotranspiration values in different months of the year showed a unimodal distribution.The daily evapotranspiration value in April was 4.67 mm·d-1,the daily evapotranspiration value in May was 4.78 mm·d-1,and the maximum daily evapotranspiration value in June or July was 5.41 mm·d-1,the daily evapotranspiration value was 2.39 mm·d-1 in September,and the daily evapotranspiration value in October was 1.13 mm·d-1.The difference between the daily average evapotranspiration values of the same month in different years was within 1 mm·d-1.The evapotranspiration area is consistent with the boundary of different land types.The evapotranspiration values of different land types are significantly different.The evapotranspiration levels of different land types in the same month are:water?forest?cropland?grassland?urban land?bare land,and the same land type has different evapotranspiration in different months.The values were:June?April?September?October.4.The Pearson correlation coefficient was used to quantitatively analyze the influence of surface parameters and meteorological parameters on the improved SEBS model for calculating evapotranspiration.Among the surface parameters,NDVI and specific humidity are highly correlated with evapotranspiration,with Pearson correlation coefficients of 0.884 and 0.843,respectively.In the meteorological parameters,the temperature is strongly correlated with evapotranspiration,and the Pearson correlation coefficient is 0.677. |
PDF Full Text Request