An approach for global monitoring of surface water extent variations in reservoirs using MODIS data
An approach for global monitoring of surface water extent variations in reservoirs using MODIS data
复制标题
DOI：
10.1016/j.rse.2017.05.039
复制
发表时间：
2017-12-01
影响因子：
13.5
通讯作者：
Kumar, Vipin
中科院分区：
文献类型：
--
作者：
Khandelwal, Ankush;Karpatne, Anuj;Kumar, Vipin
淡水资源是人类社会最基本的需求之一。然而，关于淡水水体面积的时空变化以及其中储存的水的全球信息令人惊讶地有限。我们介绍了一种新的方法，使用MODIS多光谱数据，以500米的空间分辨率，在名义上8天的时间间隔，从2000年到2015年映射的全球面积范围和已知水库的时间变化。我们评估的方法对94个水库的性能比较表面积范围的变化与卫星雷达测高测量。此外，我们提出了详细的案例研究，在四大洲的五个水库，以证明不同的挑战对该方法的性能的影响。对于这些案例研究中的三个，我们还使用基于Landsat的表面范围参考地图评估表面积估计。基于高度测量的94个水库显示更高的相关性与使用我们的方法计算的表面积相比，使用以前的方法计算的表面积。这些改进的主要原因之一是一种新型的后处理技术，该技术利用监督分类方法在多个日期产生的不完美标签来估计位置的高程结构，并使用它来增强不完美标签的质量和完整性。然而，这种海拔结构的有效估计需要水体显示足够的面积变化。因此，后处理方法对于大部分没有变化或太小而没有足够变化的水体将无效。在云层频繁、冰雪覆盖或几何形状复杂的地区，这种方法仍然面临挑战，因为这些地区需要更高空间分辨率的遥感数据。我们在这里描述的表面积估计值是可通过计算机获得的。(C)2017由Elsevier Inc.出版
Freshwater resources are among the most basic requirements of human society. Nonetheless, global information about the space-time variations of the area of freshwater bodies, and the water stored in them, is surprisingly limited. We introduce a new approach that uses MODIS multispectral data to map the global areal extent and temporal variations of known reservoirs at 500 m spatial resolution at nominal eight-day intervals from 2000 to 2015. We evaluate the performance of the approach on 94 reservoirs by comparing the variations in surface area extents with satellite radar altimetry measurements. Furthermore, we present detailed case studies for five reservoirs on four continents to demonstrate the impact of different challenges on the performance of the approach. For three of these case studies, we also evaluate surface area estimates using Landsat-based surface extent reference maps. Altimetry based height measurements for the 94 reservoirs show higher correlation with surface area computed using our approach, compared to surface area computed using previous approaches. One of the main reasons for these improvements is a novel post-processing technique that makes use of imperfect labels produced by supervised classification approaches on multiple dates to estimate the elevation structure of locations and uses it to enhance the quality and completeness of imperfect labels. However, effective estimation of this elevation structure requires that the water body shows sufficient area variations. Hence, the post-processing approach will not be effective for water bodies that are mostly unchanged or are too small to have sufficient variation. The approach is still challenged in regions with frequent cloud cover, snow and ice coverage, or complicated geometries that will require remote sensing data at finer spatial resolution. The surface area estimates we describe here are publically available. (C) 2017 Published by Elsevier Inc.