【作者】 王丽霞； 王涛； 张双成； 张京江； 赵桂生； 彭继伦；

【Author】 Lixia WANG;Tao WANG;Shuangcheng ZHANG;Jingjiang ZHANG;Guisheng ZHAO;Jilun Peng;College of Geology Engineering and Geomatics, Chang’an University;Institute of Desert Meteorology in China Meteorological Administration;Institute of Urban Meteorology,CMA;

【机构】 长安大学地质工程与测绘学院； 中国气象局乌鲁木齐沙漠气象研究所； 北京城市气象研究院；

【摘要】 土壤湿度作为能够模拟天气和气候等水文变化过程的重要参数,实时、高效的监测有着重要的科学意义。普通大地测量型接收机收集的SNR数据可用于推断测站周围土壤湿度波动,这被称为全球定位系统干涉反射(GNSS Interferometric Reflectometry,GNSS-IR)技术。目前大多数研究侧重于较短时间尺度上的土壤湿度反演问题,且较少建立不同反演模型进行对比验证。本文通过分析GNSS-IR技术反演土壤湿度参数的基本原理与过程,提取了长时间序列的反射信号相位偏移,通过与土壤体积含水量进行时序特征及相关性分析,建立了基于GPS反射信号的线性反演模型和BP神经网络反演模型。分析发现,GPS反射信号相位偏移能够反映土壤湿度的波动,且很好的表现出了土壤湿度的季候特性和年周期变化,年内相关性可达到0.8以上,多年际相关性也可达0.72,且利用BP神经网络反演模型估算的土壤湿度明显优于线性反演模型。研究表明,通过利用长时间序列SNR数据反演土壤湿度变化趋势是完全可行的,这将为低成本、大范围的监测地表土壤湿度提供一种新手段。更多还原

【Abstract】 Soil moisture is an important parameter, which can simulate weather and climate change hydrological processes, and real-time and efficient monitoring has important scientific significance. The signal-to-noise ratio data collected by ordinary geodetic receivers can be used to infer soil moisture fluctuations around the station. This is known as GNSS Interferometric Reflectometry(GNSS-IR). At present, most researches focus on soil moisture inversion on a short time scale, and few inversion models are established for comparison and verification. In this paper, By analyzing the principle and process of GNSS-IR technology to invert soil moisture, extracts the reflected signal phase shift of long-term sequence, then the characteristics of time series and correlation between the phase shift and measured soil moisture are analyzed, at last, inversion models with linear regression and BP neural network based on GPS reflection signal is presented. The analysis found that the phase shift of GPS reflected signal can reflect the fluctuation of soil moisture, and it shows the seasonal characteristics and annual cycle of soil moisture. The correlation within the year can reach above 0.8, and the inter-annual correlation can also be Up to 0.72, and the soil moisture estimated by BP neural network inversion model is obviously better than that by linear inversion model. The research indicates that it is completely feasible to invert the trend of soil moisture through signal-to-noise ratio data of long-term sequence, which will provide a new means for monitoring surface soil moisture at low cost and in a wide range.更多还原