Spatial Feature Reconstruction of Cloud-Covered Areas in Daily MODIS Composites
Spatial Feature Reconstruction of Cloud-Covered Areas in Daily MODIS Composites
复制标题
DOI：
10.3390/rs70505042
复制
发表时间：
2015-04
期刊：
影响因子：
--
通讯作者：
S. Paul;S. Willmes;O. Gutjahr;A. Preußer;G. Heinemann
中科院分区：
文献类型：
--
作者：
S. Paul;S. Willmes;O. Gutjahr;A. Preußer;G. Heinemann
云的不透明度是中分辨率成像光谱仪（MODIS）等光学和热星载传感器的主要问题。特别是在极地夜间，云和底层雪/冰之间的低热对比度导致MODIS云掩模和受影响产品的缺陷。有不同的方法来检索有关频繁被云覆盖的地区的信息，这些方法通常在很长一段时间内将大量的天数聚合到单个复合数据中。这些方法非常适合于静态、缓慢变化的表面特征（例如，坚冰范围）。然而，这并不适用于快速变化的特征，如海冰冰穴。因此，我们开发了一个空间特征重建，以获得信息的云覆盖的海冰面积的基础上，周围的日子加权成正比，其时间接近最初的一天的利益。它的性能进行测试的基础上人工筛选和人工云覆盖的案例研究的MODIS派生的冰间湖面积数据的冰间湖在南极洲布伦特冰架地区。平均而言，我们能够完全恢复人工云覆盖的测试区域的空间相关性为0.83，平均绝对空间偏差为21%。
The opacity of clouds is the main problem for optical and thermal space-borne sensors, like the Moderate-Resolution Imaging Spectroradiometer (MODIS). Especially during polar nighttime, the low thermal contrast between clouds and the underlying snow/ice results in deficiencies of the MODIS cloud mask and affected products. There are different approaches to retrieve information about frequently cloud-covered areas, which often operate with large amounts of days aggregated into single composites for a long period of time. These approaches are well suited for static-nature, slow changing surface features (e.g., fast-ice extent). However, this is not applicable to fast-changing features, like sea-ice polynyas. Therefore, we developed a spatial feature reconstruction to derive information for cloud-covered sea-ice areas based on the surrounding days weighted directly proportional with their temporal proximity to the initial day of interest. Its performance is tested based on manually-screened and artificially cloud-covered case studies of MODIS-derived polynya area data for the polynya in the Brunt Ice Shelf region of Antarctica. On average, we are able to completely restore the artificially cloud-covered test areas with a spatial correlation of 0.83 and a mean absolute spatial deviation of 21%.