The Research Of Spario-temporal Changes And Simulation Of Evapotranspiration Based On Remote Sensing Data And SWAT Model

With the in-depth study of the hydrological cycle,water quality/quantity and global climate change,the important role of evapotranspiration has gradually come into being,and more and more attention has been paid to it.From the study of single plant to field scale,with the support of existing science and technology,it has been improved.But single-point measurement or small area estimation can only be used as a reference to a certain extent in a region scale,and they are mostly specific for the evapotranspiration study of underlying surface or vegetation is not suitable for extension to regional or watershed scales.Based on this background,the rise and rapid development of RS(Remote Sensing)provides a better research direction.Remote sensing has the characteristics of real-time,rasterization of large-area and high accuracy.It can solve the problem for some areas in which lack of data or even no data.The combination of models and remote sensing data systematically improves the simulation accuracy of hydrological variables.Remote sensing evapotranspiration(ET)data was employed to investigate its ability of compensate for hydrological data in the study area,and its impact on the simulation performance of runoff and ET in the SWAT model.Xixian basin upstream of Huaihe River basin was chosen as study area.Observed runoff and remote sensing ET(MOD16A2)were used into setting up 3 scenarios for calibrating parameters and analysing their simulation performance.S1-only use the observed runoff date to calibrate parameters;S2-only use the remote sensing ET to calibrate parameters;S3-observed runoff and remote sensing ET to calibrate parameters meanwhile.Add a scenario based on the 3 scenarios,that is,the parameters of runoff and evapotranspiration are simultaneously determined after subdivision of the agricultural land(S4).It was based on S3,the type of agricultural land use was reclassified into paddy and dry land to research the Impact of different agricultural land types to simulation of evapotranspiration on the spatial scale.The results indicated that:(1)The spatial distribution of average ET during the 10-year period from 2001 to 2010 was generally low in the southwestern middle and northeast,and the spatial distribution of seasonal ET was generally consistent with the spatial distribution of annual ET.Obviously,summer(253.48mm)> spring(133.06mm)> autumn(111.32mm)> winter(60.28mm);(2)the average ET over the past 10 years is 556.61 mm,showing an insignificant upward trend year by year,77.55% of the area ET levels are mainly in the 400-600 mm interval.Evapotranspiration levels in paddy and dry land in the past 10 years have not shown a significant downward trend.The average ET of paddy fields exceeding 60% area is in the 500-600 mm interval,which is higher than in 2001-2005 In terms of the area ratio in the five-year period,from 2006 to 2010,decreased from 61.61% to 57.26%;the average ET 50% of dry land was in the 400-500 mm interval,which decreased from 49.75% to 49.44%;(3)Compared with the first period(2010-2005),the ET changes of the second period(2006-2010)in paddy and dry land were larger.The ET spatial distribution of paddy was mainly increased in the east and decreased in the middle,and that was significantly decreased in the northwest the area is rising in the northeast in the dry land.(4)when compared with S1,the NS index was decreased samely in S2/S3.But runoff simulation performance was improved obviously in S3 based on the S2.(2)As for the ET simulation on a subbasin scale,the NS index showed a increasing tendency from S1 to S3.When remote sensing ET and observed runoff was put to use,the number of subbasin whose the NS index increased occupied the whole number of subbasin reached 46% in S3 scenario when compared with S2.(5)By subdividing the type of agricultural land,the accuracy of evapotranspiration simulations in most subbasins has improved,especially in which occupied a large amount of agricultural land,such as the central,eastern and northwestern regions.It accounted for 60% of the total number of sub-watersheds,and its area reached 6625km2,accounting for about 65% of the total area.The largest increase in the simulation accuracy of subbasins was the northwest sub-watershed,and NS increased from 0.35 to 0.61.