Consistent estimates of forest land-use and change over time are important for understanding and managing human activities on the Earth’s surface, parameterizing models used for global and regional climate change analyses and a critical component of reporting requirements faced by countries as part of the international effort to Reduce Emissions from Deforestation and Degradation (REDD). In this study, object-based image analysis methods were applied to a global sample of Landsat imagery from years 1990, 2000 and 2005 to produce a land cover classification suitable for expert human review, revision and translation into forest and non-forest land use classes. We describe and analyse here the derivation and application of an automated, multi-date image segmentation, neural network classification method and independent, automated change detection procedure to all sample sites. The automated results were compared against expert human interpretation and found to have an overall agreement of ~76% for a 5-class land cover classification and ~88% agreement for change/no-change assessment. The establishment of a 5 ha minimum mapping unit affected the ability of the segmentation methods to detect small or irregularly-shaped land cover change and, combined with aggregation rules that favour forest, added bias to the automated results. However, the OBIA methods provided an efficient means of processing over 11,000 sample sites, 33,000 Landsat 20 × 20 km sample tiles and more than 6.5 million individual polygons over three epochs and adequately facilitated human expert review, revision and conversion to a global forest land-use product.

A third of the world’s ecosystems are considered degraded, and there is an urgent need for protection and restoration to make the planet healthier. The Sustainable Development Goals (SDGs) target 15.3 aims at protecting and restoring the terrestrial ecosystem to achieve a land degradation-neutral world by 2030. Land restoration through inclusive and productive growth is indispensable to promote sustainable development by fostering climate change-resistant, poverty-alleviating, and environmentally protective economic growth. The SDG Indicator 15.3.1 is used to measure progress towards a land degradation-neutral world. Earth observation datasets are the primary data sources for deriving the three sub-indicators of indicator 15.3.1. It requires selecting, querying, and processing a substantial historical archive of data. To reduce the complexities, make the calculation user-friendly, and adapt it to in-country applications, a module on the FAO’s SEPAL platform has been developed in compliance with the UNCCD Good Practice Guidance (GPG v2) to derive the necessary statistics and maps for monitoring and reporting land degradation. The module uses satellite data from Landsat, Sentinel 2, and MODIS sensors for primary productivity assessment, along with other datasets enabling high-resolution to large-scale assessment of land degradation. The use of an in-country land cover transition matrix along with in-country land cover data enables a more accurate assessment of land cover changes over time. Four different case studies from Bangladesh, Nigeria, Uruguay, and Angola are presented to highlight the prospect and challenges of monitoring land degradation using various datasets, including LCML-based national land cover legend and land cover data.

In this paper, results related with the assessment of the capabilityto detect forest degradation by analyzing NDFI time series through the BFAST algorithm are presented. Recent studies have shown the potential of the BFAST algorithm applied to a time-series of satellite-derived spectral indices such as NDVI or EVI to detect unambiguous and subtle perturbations of the forest cover canopy both positive (e.g. regeneration) and negative (e.g. deforestation). Similarly, these results suggest the feasibility to distinguish between several types of forest degradation and their causal agents such as selective logging and forest fire. In this context, the results derived from this research show that using NDFI as a data source in the BFAST algorithm improves the detection of forest degradation, and additionally provides information to understand both temporal and spatial approaches related with the dynamics of perturbations of the forest canopy

In this work, a multi-sensor approach to extract misclassified oil palm plantations from forest cover map using a modified Pauli Decomposition technique is presented. The proposed method includes the generation of a primary forest cover map built using a Landsat-based Normalized Difference Fraction Index, and then the palm oil plantation is filtered out using scattering mechanisms through the Modified Pauli Decomposition technique based on the fusion of Sentinel 1 and Alos Palsar data. Accuracy assessment of the final product, produces accuracy values of 0.946 for forest class, while the classification accuracy for non-forest class is 0.92.

Generating satellite-based deforestation alerts with actionable latency requires frequent imaging, creating an imperative to use different sensors together. We introduce a simple and open-source framework called the Disturbance Index Alert System (DIAS), which is based upon transformation of imagery from different sources into an interoperable stream of Disturbance Index (DI) values. Whereas most alert systems target divergence of forested pixels from historical states, DIAS targets movement of a pixel’s Z-score position relative to the image-wide population of forest pixels along a forest-sensitive axis. This strategy provides the following practical benefits: (1) it reduces the need to process the historical archive; (2) it reduces dependence upon stable sensor calibration; (3) it allows Z-score-based DI values to be combined across sensors; and (4) it accommodates changes to the group of sensors providing measurements. We demonstrated in Madagascar that sensor integration through DIAS can provide more timely alerts than both conventional individual-sensor systems and additive combination of such systems. Across our study sites, using a commercial source of daily imaging (PlanetScope) in conjunction with imagery from public sources (Landsat, Sentinels-1 and -2) allowed high-confidence detection (false alert rate of approximately 20%) of two-thirds of deforestation occurring at 10 m reference pixels within one month; 40% were detected in that timeframe with public data alone. As commercial options for Earth observation proliferate, flexible and computationally lightweight approaches such as DIAS are needed to accommodate diverse and sometimes only loosely calibrated instruments in support of timely forest monitoring.