Mapping forest‐cover disturbances in Papua New Guinea with AVHRR data

Abstract

Abstract. Land management and land‐use planning in a forested country such as Papua New Guinea, which is subject to various anthropogenic pressures, requires an accurate mapping of forest‐cover disturbances. The central hypothesis of this study was that remote sensing indicators of forest‐cover conditions can be used to measure and map the impact of long‐term forest‐cover disturbances. This was tested with single year NOAA's Advanced Very‐High Resolution Radiometer sensor (AVHRR) data at 1.1 km resolution. First, an ordinal scale of forest‐cover disturbance was defined from field observations, a set of thematic maps and high spatial resolution satellite sensor data. Secondly, we analysed the relationship between the forest‐cover disturbance scale and several biophysical indicators derived from AVHRR data at two seasons. Thirdly, a statistical analysis identified the optimal combination of biophysical indicators and observation dates to discriminate between the forest disturbance levels defined previously. A forest‐cover disturbance map was then produced over part of Papua New Guinea. Finally, a regionalization of the study area in terms of aggregate degree of disturbance was produced and the spatial patterns of forest disturbances were interpreted for each region in terms of broad processes of forest‐cover change. The overall accuracy of the forest‐cover disturbance map was 79%. Nine regions, homogeneous in regard to the distribution and spatial pattern of disturbance categories, were identified.