Forest Cover Classification in Panama Using Multi-Satellite Optical Images

Abstract:

The country of Panama in Central America has a hot and humid, tropical climate, with a short dry season from mid-December to mid-April, and a long rainy season for the rest of the year. Panama is a signatory to the United Nations Sustainable Development Goals (SDG) and has been working towards informing on SDG 15.2.1 – “progress towards sustainable forest management”, which consists in tracking net forest area change on an annual basis. Their efforts have been based on the use of optical imagery to generate land cover maps of the country, which run into coverage challenges as approximately 10% of each annual mosaic is blocked by clouds over areas primarily covered with forests. A way to mitigate this is a consolidated land cover product based on multiple optical datasets with greater coverage and accuracy than land cover maps based on any single source.

We employ a decision tree based supervised classification algorithm called Random Forests (RF) to generate annual and seasonal land cover classification maps for Panama at 30m resolution and less using optical data from Landsat-8 and Sentinel-2 MSI. Elevation data from SRTM is also used as a feature in both datasets. An RF model is fitted for Panama and annual forest/non-forest classification maps are generated, alongside seasonal maps for select years. Classification is carried out on median composite rasters for the duration of interest, and an ensemble modeling approach is also used wherein individual satellite images covering a part of Panama are classified separately and the inference combined post-hoc. The ensemble models show comparable and in most cases better performance than the median composite models, with forests being classified with higher accuracy than non-forests. We validate these maps through comparison with in situ data and other validated remote sensing-based products. The Landsat-8 and Sentinel-2 models are also compared for consistency and a combined land cover product is provided at 20m resolution. Results help support efforts to inform on SDG 15.2.1.

This work was carried out at the University of Maryland, Baltimore County and at the Jet Propulsion Laboratory, California Institute of Technology under contract with the National Aeronautics and Space Administration.