- Aller au contenu
- Aller à la navigation principale
- Aller à la recherche
- Aller à la navigation de haut de page
- Aller à la navigation de bas de page

Accueil du site Doctorat Chine 2018 Quantitative Remote Sensing Estimation of Urban Surface Evapotranspiration Based on a Modified Penman-monteith Model

** Titre : **Quantitative Remote Sensing Estimation of Urban Surface Evapotranspiration Based on a Modified Penman-monteith Model

** Auteur : ** 张宇；

** Grade : ** Doctoral Dissertation 2018

** Université : ** China University of Mining and Technology

**Résumé **

As an important process of land surface energy balance and hydrologic cycle,evapotranspiration（ET）plays a vital role in the distributions of global energy and moisture content.A higher estimation accuracy of ET not only has a great significance on the management of water resource,especially for the water resource allocation of agricultural and forest,but also produces positive effect on the prediction and regulation of regional climate,particularly for regional thermal environment.However,all the researches on the mathematical models of ET estimation are mostly aiming at natural and agricultural surfaces,seldom models have a special purpose on the estimation of the ET in urban areas where the land surface are most covered by impervious surface and the proportion of vegetation and soil is relatively small.Therefore,the research on urban ET estimation model will not only has an important theoretical significance for expanding the research scope of land surface ET and improving land surface energy balance system,but also has a great practical significance for further exploring the adjustment mechanism of urban natural or artificial landscapes to the urban thermal environment which will have a positive effect on improving the living environment of urban inhabitants.Based on the traditional RS-PM model,this paper proposes the urban RS-PM model,which is mainly focus on two improvements in the estimation of the surface evapotranspiration in urban areas :（1）the urban RS-PM model uses vegetation fraction and soil fraction extracted through linear spectral mixture analysis to replace fractional vegetation coverage index for quantifying the proportions of vegetation and soil,since the proportion of impervious surface can’t be expressed by fractional vegetation coverage index.（2）The urban RS-PM model uses multisource parallel model to replace dual-source model for calculating the effective energy of vegetation and soil respectively,since dual-source model ignores the distribution of net radiation on the impervious surface component.Based on the urban RS-PM model the following research work have been done :（1）the urban built-up area of Xuzhou city,Jiangsu province was selected as the study area,8 periods of Landsat 8 OLI/TIRS satellite images,meteorological data,and turbulence flux observation data were chosen as research data.（2）8 periods of urban ET results of study area were estimated,and the weighted mean ET values of the source areas which has the same spatial representation as the ground EC observation data have been calculated so that the ET estimation results can be directly validated by the flux data observed by ground eddy covariance system（EC）.（3）The importance of the 3 newly added parameters in the urban RS-PM model include vegetation fraction,soil fraction and land surface temperature were tested through sensitivity analysis.（4）The regulation effects of surface evapotranspiration on urban thermal environment were determined and quantified by using mathematical analysis and buffer zone analysis.The main results and conclusions are as follows.（1）The urban RS-PM model has improved the algorithms of the land surface component fraction,the available energy and the component latent heat flux for complex underlying surfaces,which has made the simulation of the ET more reasonable in urban areas In addition,the urban RS-PM model has also optimized the calculations of the net surface radiation parameter and the component temperature parameter which include:refining the calculation of the component net radiation based on the characteristics of land surface components to simulate the distribution of net radiation on different land surface components ;simplifying the calculation of the component temperature according to the sensitivity of the parameter to make the urban RS-PM model applicable to more types of multi-spectral remote sensing data.The result of linear fitting between the urban RS-PM inversed ET weighted average values in the source areas and the ground EC observed ET values shows that the accuracy of urban RS-PM model is good with 0.8965 for fitting goodness（R2）,24.14 W·m-2 for root mean square error（RMSE）,18.5%for mean relative error（MRE）and 0.9546 for Correlation coefficient（r）.Therefore,it is feasible to apply urban RS-PM model in urban ET estimation.（2）The sensitivity analysis of 3 parameters in urban RS-PM model shows that there is a positive linear correlation between vegetation fraction（or soil fraction）.The RS-PM model is sensitive to the change of vegetation fracation parameter and less sensitive to the change of soil fraction parameter.There is a nonlinear negative correlation between land surface temperature and the modelled ET result,however,the urban RS-PM model is also less sensitive to the change of land surface temperature parameter.The sensitivity analysis shows that the vegetation fraction has a strong influence and the soil fraction or the land surface temperature has a weak influence on the urban RS-PM model,which indicate that the vegetation fraction is a key parameter in the urban RS-PM model.（3）There is a negative correlation between urban surface ET and land surface temperature,which will be more significant in high temperature seasons The buffer zone analysis of high ET areas shows that the mean land surface temperature have an increase trend from the core area to the outermost layer of the buffer zones with the opposite trend of mean urban surface ET.And there is a significant negative linear correlation between the variation values of mean land surface temperature and mean urban surface ET for each two adjacent buffer zones.The linear fitting result indicates that the urban surface ET increased by 10 W·m-2,and the corresponding land surface temperature decreased by about 0.5

** Mots clés : **Evapotranspiration； Penman-Monteith Model； Linear spectral mixture analysis； Model improvement； Remote sensing；

Page publiée le 15 avril 2019