Continuous satellite-based indicators for mapping subtropical forest degradation and its environmental impacts (REFORCHA)

Start-End 01/12/2016 -  31/03/2021
Programme STEREO 3
Contract SR_00_338
Objective In the past, land use change assessments have mainly focussed on mapping conversion, dominantly from natural ecosystems to agriculture, and their associated environmental impacts (Hansen et al. 2013). This is incomplete for at least two reasons. First, land use may affect ecosystems substantially even without conversion, e.g. where logging results in forest degradation. Second, ecosystem properties may continue to change after conversion, e.g. where intensified agriculture leads to soil degradation. From a remote sensing perspective, measuring such processes is challenging because vegetation degradation results in more subtle changes than conversion (Foley et al., 2007; Reiche et al., 2016), and may thus be difficult to quantify using traditional, classification-based change detection methods, coarse-resolution imagery, or snapshots in time (Hill et al. 2008). Objective measurements of the extent and severity of degradation are lacking for most parts of the world (Vogt et al. 2011). This constitutes a major obstacle for assessing the connections between land use change and ecosystem functioning, including feedbacks between changes in vegetation composition and density, water use, and soil resources using processed-based models. Likewise, a scientifically robust baseline is needed to assess adaptive behaviour and management of diverse actor groups following ecosystem degradation.

The central objective of our research team is to integrate multi-source remotely sensed data, field observations and land surface models to enhance our insights in ecosystem degradation in dryland forests. More specifically, we aim to understand and quantify the intertwining of land-use dynamics, vegetation and soil degradation, following both conversions and subtle changes in subtropical woodlands, how these changes vary among actor groups, and how they affect land use decisions and land use policies.
Method The project will implement a multiscale assessment using moderate, high and very high resolution spectral data. This will involve a large-scale (over the entire study domain) mapping of state-of-the-art vegetation information, based on vegetation indices extracted from moderate resolution SPOT-VGT/PROBA-V/MODIS images. At this spatial resolution, gradual changes in forest cover are difficult to detect (Olsen et al. 2015). A site-scale analysis to monitor subtle changes in forest cover and structure will be based on three representative primary focus sites with different land use history. New algorithms based on time series and trajectory analyses (DeVries et al. 2015; Zhu & Woodcock 2014) of blend imagery from low-temporal/high-spatial resolution and high-temporal/moderate-spatial resolution sensors will be developed. These algorithms will be fully tested using UAV-based aerial images (Wallace et al. 2016) and ground-data on forest canopy structure and closure (Gasparri and Baldi, 2013). Besides, robust trend analyses following (e.g. Verbesselt et al. 2010) will be applied to the time series phenology derived from moderate resolution SPOT-VGT/PROBA-V/MODIS vegetation indices. Statistical methods (e.g. RETREND, Fensholt et al. 2015) will be used to distinguish long-term phenological change from temporal climate (rainfall or temperature) variability; which will enable us to provide a first large-scale assessment of the degree and intensity of degradation in the Dry Chaco. The remote sensing products will be thoroughly validated in small 1 by 1 km study sites, where an in-depth analysis of vegetation and soil degradation will be realized, using state-of-the-art field techniques (Nosetto et al. 2013; Clapuyt et al. 2015). To facilitate statistical spatial upscaling of the results, we will apply a probability sampling design involving a two-stage cluster sampling (Potapov et al. 2014) with the first stage being the stratified random sample of clusters (so-called primary focus sites, for site-scale analysis) using WRS1 Landsat footprints (~170 by 180 km) as sampling frame. Three clusters will be sampled that represent zones with high, intermediate and low change in forest cover based on categorical deforestation maps from the Humboldt team and prior studies (Vallejos et al. 2015; Hansen et al. 2010; Gasparri & Grau 2009). The second stage consists in a simple random sample of 15 blocks of 1 by 1 km within the primary focus sites.
Result Expected Scientific Results

Using the Dry Chaco in Argentina as an example, the project will develop advanced remote sensing methods and indicators to: (a) more reliably map forest degradation (e.g., via
logging or forest grazing) and to separate it from forest conversion and natural disturbances (e.g., fire); (b) assess the biophysical effects of forest dynamics on soil degradation and salinization, and (c) evaluate whether the extent and severity of vegetation and soil degradation differ among actor groups and whether conservation policies are effective in mitigating degradation in dryland forests.
    
Expected Products and Services

-    Spatio-temporal reconstruction of land degradation in the Dry Chaco
-    Evaluation of an ensemble of land surface model simulation in terms of soil moisture
-    Longitudinal analysis of impact of land use on forest cover change and land degradation
-    Temporally complete maps of moderate-scale vegetation indices derived from moderate resolution imagery

Potential Users

The area is lacking information on land degradation at a moderate spatio-temporal scale. This data will allow the scientific community to understand human-environment interactions and the impacts of global dynamics on forests and soils degradation. Each one of the four countries that form part of the larger Chaco region have a specific legislation on forest and soil protection, and low capacity of monitoring and controlling forest degradation. In Argentina, provincial institutions manage natural resources. These institutions are supplied with low technological material and financial support. The expected products will be transferred to local institutions and governmental administrations, and allow them to improve the monitoring and management of natural resources.
Website link
Team
Location
Sensors used
Applications
Related presentations
Related publications
Datasets used