Mapping High-Conservation-Value Forests in Agricultural Zones: An Example from Ucayali, Peru

In tropical countries, economic incentives and improved infrastructure are creating forest mosaics where small-scale farming and industrial plantations are embedded within and potentially replacing native ecosystems. In this context, landscapes need to be described beyond binary forest / non-forest classes in order to strategize zoning, monitoring, and enforcement. Light detection and ranging (Lidar) sensors have been extensively employed to characterize forest structure and carbon stocks. Due to its unprecedented spatial coverage, GEDI observations will allow mapping regional forest disturbance gradients for the first time. We propose to use GEDI-derived Level-2 metrics to characterize forest structure in an Amazonian site un Ucayali, Peru. The landscape includes both natural (elevation-driven) and anthropogenic gradients in forest structure and covers a conservation unit, a buffer zone, agroforestry, and agricultural sites. Our first step was to document national definitions of forest disturbance, based on historical disturbance maps. We then asked which GEDI-derived metrics captured the disturbance classes, starting with simple canopy height estimates and subsequently adding cumulative percentile metrics. Results suggest that canopy height can be used to map disturbance locally, but additional metrics are needed to tease out disturbance classes from elevational gradients at the state level. Last, we discuss implications for mapping High-Conservation-Value forests in the context of 'sustainable landscapes' initiatives.