Uday Gowri, Purse Bethan V, Kelley Douglas I, Vanak Abi, Samrat Abhishek, Chaudhary Anusha, Rahman Mujeeb, Gerard France F
Ashoka Trust for Research in Ecology and the Environment, Bengalore, India.
UK Centre for Ecology and Hydrology, Crowmarsh Gifford, Wallingford, United Kingdom.
PLoS One. 2025 Jan 24;20(1):e0314033. doi: 10.1371/journal.pone.0314033. eCollection 2025.
Surface water plays a vital role in the spread of infectious diseases. Information on the spatial and temporal dynamics of surface water availability is thus critical to understanding, monitoring and forecasting disease outbreaks. Before the launch of Sentinel-1 Synthetic Aperture Radar (SAR) missions, surface water availability has been captured at various spatial scales through approaches based on optical remote sensing data. A critical drawback of the latter is data loss due to cloud cover, however few studies have quantified this. This study evaluated data loss due to clouds in three Western Ghats (India) districts. These forest-agricultural mosaic landscapes, where water-related diseases are prevalent, experience the Indian monsoon. We compared surface water areas mapped by thresholding 10m Sentinel-1A SAR data with the optical 30m Landsat-derived Joint Research Centre (JRC) Global Surface Water product, currently the only globally available long-term monthly surface water data product. Backscatter thresholds were identified manually, and our Bayesian algorithm found these thresholds were very likely (>97%). While the Sentinel-1 SAR-based and JRC's optical-based approach mapped surface water extent with high overall accuracy (> 98%) when the cloud cover was low, the unmapped surface water area was substantial in the JRC product during the monsoon months. Across the districts, the average cloud cover in the July-August period was 92% or 90% for 2017 and 2018 respectively, resulting in 25% or 23% of the surface water area being unmapped. Also, the more detailed 10m resolution of Sentinel-1A SAR helped detect the many small water features missed by 30m JRC. Thus, for predicting water-related disease risks linked to small water features or monsoon rainfall, Sentinel-1A SAR is more effective. Finally, automatic backscatter thresholding for unvegetated surface water mapping can be effective if threshold values are adapted to regional-specific backscatter spatial and temporal variations.
地表水在传染病传播中起着至关重要的作用。因此，有关地表水可利用性的时空动态信息对于理解、监测和预测疾病爆发至关重要。在哨兵 -1合成孔径雷达（SAR）任务发射之前，地表水可利用性已通过基于光学遥感数据的方法在不同空间尺度上进行了获取。然而，后者的一个关键缺点是由于云层覆盖导致的数据丢失，不过很少有研究对此进行量化。本研究评估了印度西高止山脉三个地区因云层导致的数据丢失情况。这些森林 - 农业镶嵌景观地区与水相关的疾病盛行，且受印度季风影响。我们将通过对10米哨兵 -1A SAR数据进行阈值处理绘制的地表水区域与光学30米陆地卫星衍生的联合研究中心（JRC）全球地表水产品进行了比较，该产品是目前唯一全球可用的长期月度地表水数据产品。后向散射阈值是手动确定的，我们的贝叶斯算法发现这些阈值非常可能（>97%）。当云层覆盖较低时，基于哨兵 -1 SAR的方法和JRC基于光学的方法绘制地表水范围的总体精度都很高（>98%），但在季风月份，JRC产品中未绘制的地表水面积相当大。在各个地区，2017年和2018年7 - 8月期间的平均云层覆盖率分别为92%和90%，导致25%或23%的地表水面积未被绘制。此外，哨兵 -1A SAR更详细的10米分辨率有助于检测到30米JRC遗漏的许多小水体特征。因此，对于预测与小水体特征或季风降雨相关的水传播疾病风险，哨兵 -1A SAR更有效。最后，如果阈值能够适应区域特定的后向散射时空变化，那么用于无植被地表水绘图的自动后向散射阈值处理可能会很有效。
