A New GPS SNR-based Combination Approach for Land Surface Snow Depth Monitoring
A New GPS SNR-based Combination Approach for Land Surface Snow Depth Monitoring
复制标题
DOI：
10.1038/s41598-019-40456-2
复制
发表时间：
2019-03
影响因子：
4.6
通讯作者：
W. Zhou;Lilong Liu;Liangke Huang;Yibin Yao;Jun Chen;Songqing Li
中科院分区：
文献类型：
--
作者：
W. Zhou;Lilong Liu;Liangke Huang;Yibin Yao;Jun Chen;Songqing Li
雪不仅是水文循环中的重要储存成分，也是气候研究的重要数据；然而，降雪观测资料很少。最近，信噪比(SNR)被应用于积雪深度的传感。大多数研究只考虑全球定位系统(GPS)L1或L2信噪比数据。在本研究中，提出了一种利用GPS三频(即L1、L2和L5)信号的多路径反射率和信噪比组合来估计积雪深度的新方法。信噪比组合方法描述了天线高度变化与频谱峰值频率之间的关系。积雪深度是从YEL2和KIRU站点的SNR组合数据中提取的，并通过与情景观测进行比较来验证。仰角从5°到25°不等。两个测点的相关系数分别为0.99和0.97。通过与现有模型的比较，对新方法的性能进行了评估。该方法的相关性高达0.95，精度(以均方根误差计算)提高30%以上。研究结果表明，该方法可用于积雪深度监测，并可为建立多系统、多频率的全球导航卫星系统反射率模型提供参考。
Snow is not only a critical storage component in the hydrologic cycle but also an important data for climate research; however, snowfall observations are only sparsely available. Signal-to-noise ratio (SNR) has recently been applied for sensing snow depths. Most studies only consider either global positioning system (GPS) L1 or L2 SNR data. In the current study, a new snow depth estimation approach is proposed using multipath reflectometry and SNR combination of GPS triple frequency (i.e. L1, L2 and L5) signals. The SNR combination method describes the relationship between antenna height variation and spectral peak frequency. Snow depths are retrieved from the SNR combination data at YEL2 and KIRU sites and validated by comparing it within situobservations. The elevation angle ranges from 5° to 25°. The correlations for the two sites are 0.99 and 0.97. The performance of the new approach is assessed by comparing it with existing models. The proposed approach presents a high correlation of 0.95 and an accuracy (in terms of Root Mean Square Error) improvement of over 30%. Findings indicate that the new approach could potentially be applied to monitor snow depths and may serve as a reference for building multi-system and multi-frequency global navigation satellite system reflectometry models.