Abstract Satellite remote sensing provides unprecedented information on near-surface soil moisture at a global scale, enabling a wide range of studies such as drought monitoring and forecasting. Data Assimilation (DA) has been recognized as an effective means to incorporate such observations into hydrologic models to better predict and forecast hydroclimatic variables. In this study, we use a recently developed Evolutionary Particle Filter with Markov Chain Monte Carlo (EPFM) approach to assimilate Soil Moisture Active Passive (SMAP) soil moisture data into Variable Infiltration Capacity (VIC) hydrologic model to provide more reliable topsoil layer moisture (0~–5 cm) over the entire Continental United States (CONUS). The EPFM outperformed an Ensemble Kalman filter (EnKF) in terms of correlations and the unbiased root mean square error (ubRMSE) with in situ measurements from the Soil Climate Analysis Network (SCAN) and the United States Climate Reference Network (USCRN). Also, we used a multivariate probability distribution based on a Copula function to integrate the posterior soil moisture, precipitation (from the North American Land Data Assimilation System (NLDAS)) and evapotranspiration (from the Moderate Resolution Imaging Spectroradiometer (MODIS)) information to develop a new integrated drought index, i.e. SPESMI. To validate the usefulness of the developed integrated drought index, we compared the drought events detected by this index with those reported by the United States Drought Monitor (USDM). The results indicated a strong temporal consistency of the drought areas detected by our approach and the USDM over the entire period of study (April 2015 to June 2018). In addition to such promising results, we noticed that our approach could capture the flash drought in 2017 in the U.S. Northern Plains earlier than the USDM, and could identify some severe to extreme drought events that had been underestimated by the USDM. Moreover, the SPESMI has a high correlation with the yield loss of spring and winter wheat in the United States. This novel drought monitoring framework can serve as an independent and potentially complementary drought monitoring system.

中文翻译：

---

使用卫星土壤水分、数据同化和综合干旱指数进行大陆干旱监测

摘要 卫星遥感在全球范围内提供了前所未有的近地表土壤水分信息，使干旱监测和预测等广泛研究成为可能。数据同化 (DA) 已被公认为将此类观测结果纳入水文模型以更好地预测和预报水文气候变量的有效手段。在这项研究中，我们使用最近开发的带有马尔可夫链蒙特卡罗 (EPFM) 方法的进化粒子滤波器将土壤水分主动被动 (SMAP) 土壤水分数据同化为可变入渗能力 (VIC) 水文模型，以提供更可靠的表土层水分。 0~–5 cm) 覆盖整个美国大陆 (CONUS)。在土壤气候分析网络 (SCAN) 和美国气候参考网络 (USCRN) 的原位测量的相关性和无偏均方根误差 (ubRMSE) 方面，EPFM 优于集成卡尔曼滤波器 (EnKF)。此外，我们使用基于 Copula 函数的多元概率分布来整合后验土壤水分、降水（来自北美土地数据同化系统 (NLDAS)）和蒸散量（来自中分辨率成像光谱仪 (MODIS)）信息以开发一个新的综合干旱指数，即 SPESMI。为了验证开发的综合干旱指数的有用性，我们将该指数检测到的干旱事件与美国干旱监测机构 (USDM) 报告的干旱事件进行了比较。结果表明，在整个研究期间（2015 年 4 月至 2018 年 6 月），我们的方法和 USDM 检测到的干旱地区具有很强的时间一致性。除了这些有希望的结果，我们注意到我们的方法可以比 USDM 更早地捕捉到 2017 年美国北部平原的突发干旱，并且可以识别出一些被 USDM 低估的严重到极端干旱事件。此外，SPESMI 与美国春、冬小麦的产量损失具有高度相关性。这种新颖的干旱监测框架可以作为一个独立的、潜在的补充干旱监测系统。我们注意到，我们的方法可以比 USDM 更早地捕捉到 2017 年美国北部平原的突发干旱，并且可以识别出一些被 USDM 低估的严重到极端干旱事件。此外，SPESMI 与美国春、冬小麦的产量损失具有高度相关性。这种新颖的干旱监测框架可以作为一个独立的、潜在的补充干旱监测系统。我们注意到，我们的方法可以比 USDM 更早地捕捉到 2017 年美国北部平原的突发干旱，并且可以识别出一些被 USDM 低估的严重到极端干旱事件。此外，SPESMI 与美国春、冬小麦的产量损失具有高度相关性。这种新颖的干旱监测框架可以作为一个独立的、潜在的补充干旱监测系统。