Monitoring changes in the extent, distribution and quality of habitat for our wildlife targets over time is key to evaluating the impact of our conservation efforts.    For terrestrial systems, remote sensing image analysis is the most cost-effective approach for tracking habitat change because it can be undertaken relatively cheaply at an appropriate spatial and temporal scale.  

To date most approaches to habitat change have required costly and time-consuming land-cover/land-use mapping across two or more time periods.  Rather than attempting to evaluate where specific land-cover and land-use transitions are occurring within a landscape (e.g., rainforest to oil palm, wetlands to rice cultivation, logging to tree plantations, etc.), it is easier and less expensive to simply identify locations of change through hot-spot detection (i.e., places where we know something has happened to the vegetated surface, but are not sure what).  The real novelty of this “hot-spot” monitoring approach is that it integrates relatively automated satellite remote sensing analysis with targeted manned or unmanned aerial surveys, or ranger patrols to identify conservation what has changed with detected hot spots.

To detect hotspots of change we propose using free, time-series, 250m spatial resolution, cloud-free composited data from the MODIS satellite to generate annual maxNDVI images (i.e., a proxy measure for how much green vegetation is present in every pixel across the landscapes). These composite maxNDVI images represent the peak greenness of any pixel in the image in a given year with one composite image for each year in the detection period.  The time-series of annual maximum value NDVI images are then compared and hotspots detected where the difference in NDVI value of any pixel between two time periods exceeded a threshold (to be determined).   In areas prone to NDVI saturation associated with dense vegetation cover we need to consider incorporating land surface temperature into the analysis (Julien et al. 2011).  Given available tools we believe that hotspots detection using MODIS NDVI data could be largely automated, dramatically simplifying the process.  Hot-spot detection has been shown to be an effective tool for rapid identification of change and is now used by NASA to provide a global quarterly indicator of change (http://geo.arc.nasa.gov/sge/casa/latest.html).  The World Resources Institute (WRI) is using a similar approach for their pantropical near real time deforestation alert system that is part of their soon to be launched – Global Forest Watch 2.0. 

Once hotspot locations are detected using the NDVI data, land cover/land use change could be both confirmed and characterized (i.e., what change in land cover/land use has actually occurred) in three ways: 1) obtain and visually examine very high spatial resolution satellite imagery such as QuickBird, Pléiades, and GeoEye; 2) use SMART LEM software to direct upcoming eco-guard or ranger patrols to visit and evaluate each observed hotspot; or 3) conduct piloted or unmanned (auto-piloted fixed-wing or helicopter drones) aerial-imagery, over-flights.   Any or a mix of these follow-up surveys will: a) allow landscape managers to determine whether detected hotspots correspond to real changes in land cover and land use, b) allow the type of land cover/land use change to be documented, and c) help managers to decide how best to address the root causes of the detected change.

Monitoring of seagrass beds is feasible using the terrestrial hot-spots approach.   Coral reef cover is typically monitored within WCS coral seascapes during the same 50m diver transects used to monitor fish and other benthic species.