梁勇, 杨磊, 吴秋兰, 洪学宝, 汉牟田, 杨东凯. 地表粗糙度影响下的GNSS-R土壤湿度反演仿真分析[J]. 武汉大学学报 ( 信息科学版), 2018, 43(10): 1546-1552. DOI: 10.13203/j.whugis20160557
引用本文: 梁勇, 杨磊, 吴秋兰, 洪学宝, 汉牟田, 杨东凯. 地表粗糙度影响下的GNSS-R土壤湿度反演仿真分析[J]. 武汉大学学报 ( 信息科学版), 2018, 43(10): 1546-1552. DOI: 10.13203/j.whugis20160557
LIANG Yong, YANG Lei, WU Qiulan, HONG Xuebao, HAN Moutian, YANG Dongkai. Simulation of Soil Roughness Impact in GNSS-R Soil Moisture Retrieval[J]. Geomatics and Information Science of Wuhan University, 2018, 43(10): 1546-1552. DOI: 10.13203/j.whugis20160557
Citation: LIANG Yong, YANG Lei, WU Qiulan, HONG Xuebao, HAN Moutian, YANG Dongkai. Simulation of Soil Roughness Impact in GNSS-R Soil Moisture Retrieval[J]. Geomatics and Information Science of Wuhan University, 2018, 43(10): 1546-1552. DOI: 10.13203/j.whugis20160557

地表粗糙度影响下的GNSS-R土壤湿度反演仿真分析

Simulation of Soil Roughness Impact in GNSS-R Soil Moisture Retrieval

  • 摘要: 基于双天线全球导航卫星系统反射技术(global navigation satellite system reflectometry,GNSS-R),建立了两个修正地表粗糙度影响的土壤湿度反演模型——解析模型和人工神经网络模型,并以GPS L1 C/A码为例建立了GNSS-R土壤湿度仿真平台,仿真分析了地表粗糙度对两个模型反演精确度的影响。结果表明,当地表均方根高度大于0.010 m时,必须对解析模型进行粗糙度修正。粗糙度影响修正结果显示,小粗糙度情况下修正的解析模型取得了良好的结果,但对于大粗糙度有一定局限性。在均方根高度大于0.025 m时,进行土壤粗糙度修正前,人工神经网络模型精度比解析模型提高了36.83%~72.36%。进行修正后,人工神经网络模型的精度比解析模型提高了42.86%~54.40%。人工神经网络模型在修正前后取得了相近的精度,无修正的人工神经网络模型精度比有修正的解析模型精度仍提高了35.83%~53.48%。

     

    Abstract: This paper presents two dual antenna GNSS-R(global navigation satellite system reflectometry) soil moisture retrieval models with soil roughness compensation-an analytic model and an artificial neural network (ANN) model. Then a simulator for GNSS-R soil moisture retrieval is built in consideration of GPS L1 C/A code modulation. After that the impact of soil roughness is elaborated. The simulation results show that the roughness compensation is necessary for the analytic model when the RMSH(root mean square high) is larger than 0.010 m. The roughness compensation works well for small roughness, but there are some limitations for large roughness. Under the situation where RMSH is greater than 0.025 m, the accuracy of ANN model is 36.83%-72.36% higher than the analytic model without roughness compensation, and the accuracy of ANN model is 42.86%-54.40% higher than the analytic model with roughness compensation. The ANN model achieves similar accuracy regardless of roughness compensation, and the accuracy of ANN model without compensation is still 35.83%-53.48% higher than the analytic model with compensation.

     

/

返回文章
返回