双极化SAR数据裸露地表土壤水分反演

Abstract

As we all know, soil moisture has important implications in water cycle research, ecosystem research, drought monitoring, weather forecasting, and water conservancy construction. Most of the northwest region of China is in arid and semi-arid regions. Therefore, the provision of soil moisture products of small and medium scale can provide important information support for drought monitoring and ecological monitoring in the northwest region.

The soil moisture data of traditional meteorological stations mainly rely on soil moisture sampling from ground sampling sites. Although this method has high accuracy, it is difficult to achieve large-scale, high-precision, real-time monitoring of soil moisture. With the extension of remote sensing bands and the improvement of spatial resolution and temporal resolution of remote sensing data, the use of remote sensing methods to monitor soil moisture in real time and accurately has become a trend. Visible light and thermal infrared and other optical remote sensing monitoring of soil moisture have many limitations, and passive microwave is more suitable for large-scale monitoring of soil moisture. Synthetic aperture radar has become the main method for the retrieval of soil moisture at medium and small scales.

When domestic and foreign scholars studied microwave remote sensing to invert soil moisture, it was found that soil moisture and soil roughness are two factors that mainly affect the backscatter coefficient, and the general soil roughness consists of two parameters. How to accurately describe the soil roughness is a hot issue. The current research on soil moisture mainly concentrates on the research area with measured roughness data. There is less research on soil moisture retrieval in areas with no measured roughness. In actual soil moisture inversion, roughness is a parameter difficult to measure and obtain. Therefore, how to obtain surface roughness parameters through SAR images and invert soil moisture with higher accuracy is also a hot issue.

This paper mainly compares the soil moisture inversion with the actual roughness field and the actual roughness field. For the study area with measured roughness, by using the AIEM model to input different parameters, the functional relationship between different

polarization backscattering coefficients and soil moisture and combined roughness was analyzed, and two different polarization backscatter coefficients were established. In conjunction with the nonlinear equations of soil moisture and combined roughness (ie, soil moisture inversion model), an improved particle swarm algorithm for soil moisture inversion was designed. Based on the model and inversion procedures, pre-processed ASAR data was used for the study. The soil moisture inversion was performed in the area and the inversion accuracy was evaluated. For the study area without measured roughness, based on the simulation of surface backscattering database based on the AIEM model, the root mean square height and correlation length inversion method was proposed, and then the combination was established based on the combination. Roughness soil moisture inversion model and LUT soil moisture inversion method, and the accuracy evaluation of these soil moisture inversion methods to obtain the best soil moisture inversion model, based on the use of pre-processed Radasat-2 Radar data soil moisture inversion.

The main research results are as follows:

1、For the measured roughness study area, two sets of nonlinear equations with different polarization methods: backscattering coefficient, soil moisture, and combined roughness were established. An improved particle swarm algorithm for soil moisture inversion was designed. The accuracy of soil moisture retrieval in the same study area (R2=0.71, R2=0.745) is lower than that of the soil water inversion accuracy based on the improved particle swarm algorithm in this paper(R2=0.77786).The improved particle swarm algorithm is better than the previous solution. The elimination method used by the model not only improves the accuracy but also avoids the problems of multiple solutions and no solutions. The feasibility of retrieving soil moisture based on improved particle swarm algorithm is fully demonstrated.

2、For the no-measured roughness study area, two sets of soil moisture inversion models were constructed and the four inversion schemes were compared and analyzed: The results show that the soil moisture retrieval model based on effective correlation length has good accuracy, and its R2= 0.689, MAE=31%, MRE=3.02%, RMSE=0.04, suitable for inversion of soil moisture in areas with no measured roughness.

Key words:soil moisture, inversion, AIEM, improved particle swarm algorithm, LUT, ASAR, Radarsat-2

目录

第一章绪论 (1)

1.1选题背景和意义 (1)

1.2国内外研究现状 (2)

1.3选题依据 (4)

1.4研究内容 (5)

1.5论文结构 (5)

第二章土壤水分反演理论基础 (7)

2.1 微波遥感反演土壤水分原理 (7)

2.2 雷达系统和地表参数 (7)

2.3 裸露地表散射模型选择 (10)

第三章研究区介绍和数据处理 (12)

3.1有实测粗糙度数据的研究区 (12)

3.1.1研究区概况 (12)

3.1.2卫星数据介绍 (12)

3.1.3ASAR数据处理 (13)

3.1.4实测数据处理 (14)

3.2 无实测粗糙度数据的研究区 (14)

3.2.1 研究区概况 (15)

3.2.2 Radarsat-2数据介绍 (15)

3.2.3Radarsat-2数据处理 (15)

第四章土壤水分反演模型建立与应用 (18)

4.1 基于有实测粗糙度的研究区土壤水分反演 (18)

4.1.1 构建组合粗糙度 (18)

4.1.2 组合粗糙度与后向散射系数的关系 (19)

4.1.3 土壤水分与后向散射系数的关系 (19)

4.1.4 基于组合粗糙度的土壤水分反演模型 (20)

4.1.5 实验与分析 (22)

4.1.6 反演精度评价 (27)

4.2 基于无实测粗糙度数据的研究区土壤水分反演 (27)

4.2.1 均方根高度反演 (27)

4.2.2 相关长度反演 (30)

4.2.3 组合粗糙度土壤水分反演模型 (31)

4.2.4 基于组合粗糙度的土壤水分反演 (34)

4.2.5基于粗糙度定标的土壤水分反演 (35)

4.2.6 反演模型精度评价 (37)

4.2.7 土壤水分反演 (37)

4.3 本章小结 (39)

结论与展望 (40)

参考文献 (42)

攻读学位期间取得的研究成果 (46)

致谢 (47)

相关主题
相关文档
最新文档