Evaluation of the Continuous Monitoring of Land Disturbance Algorithm for Large-Scale Mangrove Classification

Abstract

Mangrove forests are of high biological, economic, and ecological importance globally. Growing within the intertidal zone, they are particularly vulnerable to the effects of climate change in addition to being threatened on local scales by over-exploitation and aquaculture expansion. Long-term monitoring of global mangrove populations is therefore highly important to understanding the impact of these threats. However, data availability from satellites is often limited due to cloud cover. This problem can be mitigated using a season-trend modelling approach such as Continuous Monitoring of Land Disturbance (COLD). COLD operates by using every available observation on a pixel-wise basis, removing the need for whole cloud free images. The approach can be used to better classify land cover by taking into account the underlying seasonal variability, and can also be used to extrapolate between data points to obtain more accurate long term trends. To demonstrate the utility of COLD for global mangrove monitoring, we applied it to five study sites chosen to represent a range of mangrove species, forest types, and quantities of available data. The COLD classifier was trained on the Global Mangrove Watch 2010 dataset and applied to 30 years of Landsat data for each site. By increasing the period between model updates, COLD was successfully applied to all five sites (2253 scenes) in less than four days. The method achieved an overall accuracy of 92% with a User’s accuracy of 77% and a Dice score of 0.84 for the mangrove class. The lowest User’s accuracy was for North Kalimantan (49.9%) due to confusion with mangrove palms. However, the method performed extremely well for the Niger Delta from the 2000s onwards (93.6%) despite the absence of any Landsat 5 data. Observation of trends in mangrove extent over time suggests that the method was able to accurately capture changes in extent caused by the 2014/15 mangrove die-back event in the Gulf of Carpentaria and highlighted a net loss of mangroves in the Matang Forest Reserve over the last two decades, despite ongoing management. COLD is therefore a promising methodology for global, long-term monitoring of mangrove extent and trends.

Global Mangrove Watch: Updated 2010 Mangrove Forest Extent (v2.5)

“…In total, 11,262 Sentinel-2 acquisitions were downloaded and used for this analysis. To generate a standardised reflectance product for the classifications, the ARCSI software [34] was employed, as successfully demonstrated in past studies [15,26,27,35]. The ARCSI software uses the 6S model [36] through the Py6S module [37] parameterised using the image header information and an aerosol optical depth estimated from a dark object subtraction [15].…”

“…However, these studies have typically been undertaken over small spatial extents or for a few countries (e.g., [25]), single countries (e.g., [24]), or particular areas of interest at sub-national scales (e.g., [23]). Alternative approaches to mangrove mapping that have focused on mapping through time have also been proposed, such as [26,27], which have used the COntinuous monitoring of Land Disturbance (COLD) [28] method to provide individual site level time-series maps of mangrove extent. However, these time-series approaches are computationally intensive and therefore difficult to apply at a global scale.…”

Global Mangrove Watch: Monthly Alerts of Mangrove Loss for Africa

“…More complex change detection algorithms might be considered, such as the continuous monitoring of land disturbance (COLD) algorithm [30], which has already been demonstrated for mangrove applications [24,31]. These approaches make better use of the full time series with model fitting allowing regions with more variability in response (e.g., inter-tidal areas with open canopies) to be taken into account and reducing the number of false positives.…”

“…Many studies (e.g., [2,15,[22][23][24]) have considered mapping land cover change, including forest loss, through remote sensing imagery using various sensors, modalities and methods. There are a number of approaches to mapping land cover change, but these fit within four broad categories, (1) map-to-map, (2) image-to-image, (3) map-to-image and (4) time series approaches.…”