Global lake area time series from Landsat images (GLATS) V1.0

The GLATS V1.0 dataset contains time-varying lake areas from Landsat satellites for 576 global lakes (1 – 81,844 km2) spanning 1992 to 2018 at a near monthly frequency. The method is described in detail in Yao et al. (2019), but a quick overview is provided here.

1. We carefully delineate the Region of Interest (ROI) for each water body based on the UCLA global lake inventory (Sheng et al., 2016) and ensure that other major water bodies except the studied one were excluded from the ROI.
2. We consider a total of six commonly used water indices and apply the optimum water index for each water body.
3. We incorporate a refinement process for reducing the errors due to disturbance, such as clouds, shadows, and aquatic vegetation, based on the topographical constraint of lake shrinkage and growth.
4. We develop a method for recovering full water areas from partially contaminated images using lake isobaths mapped from contamination-free images and vector-based interpolation.
5. We validate our method on 428 water bodies worldwide. The recovered water areas show a relative root-mean-squared error of 2.2%, and the errors for over 95% of the water bodies below 6.0%.

This dataset includes:

1. GLATS_lakes_location.shp: spatial locations of studied lakes in a vector format
2. Coordinates_GLATS.csv: coordinates of studied lakes in a cvs format
3. Timevarying_areas folder: monthly water area time series for each lake