A simple cloud-filling approach for remote sensing water cover assessments

Abstract

Abstract. The empirical attribution of hydrologic change presents a unique data availability challenge in terms of establishing baseline prior conditions, as one cannot go back in time to retrospectively collect the necessary data. Although global remote sensing data can alleviate this challenge, most satellite missions are too recent to capture changes that happened long ago enough to provide sufficient observations for adequate statistical inference. In that context, the 4 decades of continuous global high-resolution monitoring enabled by the Landsat missions are an unrivaled source of information. However, constructing a time series of land cover observation across Landsat missions remains a significant challenge because cloud masking and inconsistent image quality complicate the automatized interpretation of optical imagery. Focusing on the monitoring of lake water extent, we present an automatized gap-filling approach to infer the class (wet or dry) of pixels masked by clouds or sensing errors. The classification outcome of unmasked pixels is compiled across images taken on different dates to estimate the inundation frequency of each pixel, based on the assumption that different pixels are masked at different times. The inundation frequency is then used to infer the inundation status of masked pixels on individual images through supervised classification. Applied to a variety of global lakes with substantial long term or seasonal fluctuations, the approach successfully captured water extent variations obtained from in situ gauges (where applicable), or from other Landsat missions during overlapping time periods. Although sensitive to classification errors in the input imagery, the gap-filling algorithm is straightforward to implement on Google's Earth Engine platform and stands as a scalable approach to reliably monitor, and ultimately attribute, historical changes in water bodies.

High-resolution surface water dynamics in Earth’s small and medium-sized reservoirs

“…Numerous algorithms and datasets have been developed over the last decade focusing on the accurate estimation of surface water dynamics from optical satellite imagery 18 , 58 – 63 . A key requirement to use those algorithms that can classify water where it is occluded due to clouds 24 , 64 , 65 . To detect water mask in every satellite image that intersects with a given reservoir geometry, we first discriminate surface water using NDWI spectral index by applying the method of local thresholding based on the Canny edge detector and a binary version of the Otsu thresholding algorithm 61 , 66 .…”

From wetlands to wetlandscapes: Remote sensing calibration of process‐based hydrological models in heterogeneous landscapes

“…The accuracy of unsupervised classification is sensitive to the relative number of pixels assigned to each of the two clusters, so the automatic identification of a segmentation threshold can be problematic for images dominated by either land or open water. Following Mullen et al (2021), we addressed this issue by (i) automatically classifying each image using k‐means, (ii) recording the segmentation threshold (i.e., the MNDWI level above which a pixel is classified as water) for each image and (iii) using the median value of the obtained segmentation thresholds to segment all the images.…”

“…Yet cloud cover and image quality issues have limited their use for the short spatial and temporal scales that are ecologically relevant for wetlandscapes with ephemeral wetlands. More recently, new gap‐filling algorithms have been proposed with promising ability to detect open water on Landsat images in cloudy weather (Mullen et al, 2021; Schwatke et al, 2019; Zhao & Gao, 2018), but their potential to improve the monitoring of wetlandscape dynamics at the relevant spatial and temporal scales remains to be evaluated. This paper fills this gap by comparing gap‐filled water cover estimates from Landsat to two alternative data sources—namely in situ observations and metric‐resolution imagery—each time focusing on a key application and the most relevant tradeoff.…”

“…threshold can be problematic for images dominated by either land or open water. FollowingMullen et al (2021), we addressed this issue by (i) automatically classifying each image using k-means, (ii) recording the segmentation threshold (i.e., the MNDWI level above which a pixel is classified as water) for each image and (iii) using the median value of the obtained segmentation thresholds to segment all the images.We then used the approach described inMullen et al (2021) to transform the three-class images ('masked', 'dry' and 'wet') into binary water extent images('dry' and 'wet'). Note that the 'masked' pixel class includes both the pixels masked by clouds and the pixels with no data due to the scanline corrector failure on Landsat 7 images taken after September 2003.…”

Monitoring Surface Water Inundation of Poyang Lake and Dongting Lake in China Using Sentinel-1 SAR Images

“…Thirdly, the pattern of the temporal filtering from frequency maps in this study can be further extended and applied on different study areas such as vegetation extraction and cloud removal using different datasets such as Landsat series and Sentinel-2 data. The filtering method can be conversely utilized as filling method in cases where the images are missing or severely contaminated, which has been explored in former work [79,80]. Lastly, all the methods can be integrated into an automatic surface water mapping software in order to supply near-real-time and high-quality water inundation products for monitoring.…”