Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Chine → 2016 → The Study of Soil Moisture Retrieval Method Based on Multi-source Remote Sense Data in Arid And Semi-arid Area

South China Agricultural University (2016)

The Study of Soil Moisture Retrieval Method Based on Multi-source Remote Sense Data in Arid And Semi-arid Area

黄资彧

Titre : The Study of Soil Moisture Retrieval Method Based on Multi-source Remote Sense Data in Arid And Semi-arid Area

Auteur : 黄资彧

Grade : Doctoral Dissertation 2016

Université : South China Agricultural University

Résumé
Soil moisture is an important variable of land environment, and it has an important relationship with climate, hydrology and ecology. The soil moisture largely affects the surface energy balance, regional surface water runoff and crop yield, etc. Soil moisture monitoring can get some moisture held by vegetation and potential drought conditions, which for the guidance of agricultural production, enhance crop resistance to disasters and crop yield estimation provides a good basis for decision making. There is no doubt that traditional methods for soil moisture determination, such as field measurement, can provide the most accurate measurements of soil moisture layers, but this method not only requires a lot of manpower, material and financial resources and limited to a single point measurement, it is impossible to obtain large-scale, high-efficiency measurements. At the same time, the difference of landform and soil type make the representative of single point data is bad, limiting its scope of application. Monitoring soil moisture by remote sensing can overcome the shortcomings of traditional methods, with fast, efficient, sustainable and coverage advantages. Therefore, the study of soil moisture remote sensing inversion model still has great practical significance.In this paper, by using HJ-1/Landsat 8 multi-source remote sensing data and related ground observation data, we conducted a study of the surface soil moisture remote sensing inversion method in the Heihe River Zhangye city and the Yellow River Source Region Maduo County.(I)The surface albedo is one of the key variables for estimate surface net radiation and to improve the thermal inertia method. In this study, we proposed a method for estimating the surface albedo from HJ-1/Landsat 8 data using anisotropy information from concurrent MODIS 500-m observations as priori knowledge to resolve the problem that HJ-1/Landsat 8 data only has a single view angle. And a further narrowband to broadband conversion based on 6S radiative transfer simulations was applied to produce the broadband albedo.(II)The surface temperature is a key parameter to improve the thermal inertia method and TVDI method, in this study, the improved generalized single-channel algorithm was adopted for retrieved landsat surface temperature from HJ-1data, and the split window algorithm was adopted for retrieved landsat surface temperature from Landsat 8 data. On this basis, the particle swarm optimization algorithm was adopted for retrieved component temperature and then using the inversion vegetation temperature instead of the surface temperature of traditional temperature–vegetation index method(TVX) method to estimate near-surface air temperature. The retrieved surface albedo was validated by the observed data, and the results showed that the method used in this paper is effective for retrieving surface albedo and can yield a reasonable estimation of surface albedo. The surface temperature, vegetation temperature, soil temperature, near-surface air temperature were validated by the observed data, The surface temperature retrieved with the improved algorithm were consistent with those provided by the MODIS product, and that the satellite-derived air temperature also had a consistent distribution with land surface temperature, the satellite-derived air temperatures were in good agreement with the meteorological observed values, and the accuracy was in line with previously reported results for the TVX method. Compared to the direct use of surface temperature to estimate near-surface air temperature, we present a more accurate method.(III)The surface net radiation is also a critical parameter to improve thermal inertia model method. Numerous studies have developed frameworks to estimate net radiation or its components from remote sensing data for clear sky conditions. In this paper, a simple scheme is proposed to estimate instantaneous net radiation for clear sky days using HJ-1/Landsat 8 data, our method attempts to develop an algorithm which primarily uses remote sensing information and eliminates the need for ground information as model input. The radiation four components and surface net radiation were validated by the observed data, Comparing with the traditional ground-based measurements, satellite remote sensing provides a straightforward and consistent way to obtain net radiation over regional and global scales with more spatially detailed information.(IV)For the improved thermal inertia model still requires ground data calculated maximum surface temperature. In this paper, we use MODIS LST products derived daily surface temperature and then calculate the maximum surface temperature, so that the improved thermal inertia model to get rid of dependence on ground observation data. The retrieved soil moisture was validated by the observed data. The MAE and RMSE for the improved thermal inertia method inversion results and observed data were 3.14 and 3.62, respectively.(V)Taking into consideration that NDVI is just a parameter indicate surface green degree in traditional TVDI method, and it is restricted by remote sensing sensor resolution, we use fractional vegetation cover(Fr) which is more representative of the relative proportions of soil and vegetation in pixel to build TVDI characteristic space, and a practical algorithm was adopted to determine quantitatively the dry and wet edges of this space in arid and semiarid regions, which can reduces "false" dry spots’ impact on the dry side determined. The retrieved soil moisture was validated by the observed data. The MAE and RMSE for the improved TVDI method inversion results and observed data were 3.41 and 3.87, respectively.(VI)In order to improve the soil water inversion accuracy, we integrated use the two methods to retrieved surface soil moisture, and the MAE and RMSE for the inversion results and observed data were 2.72 and 3.29, respectively. The inversion accuracy of this method has improved significantly

Mots clés : soil moisture; HJ-1; Landsat 8; surface temperature-fractional vegetation cover characteristic space; thermal inertia model;

Présentation (CNKI)

Page publiée le 31 janvier 2017, mise à jour le 11 septembre 2017