A Unified Disturbance Analysis for Forests and Grasslands in New Zealand
Abstract:Land management is a key driver of land change in many parts of the world. Activities such as livestock farming and timber production can have a dramatic impact on the environment and are often guided by local policies and practices. Evaluation of these impacts is particularly important in a country like New Zealand, where since 1991 political boundaries have largely coincided with environmental boundaries (catchments). In this study we look at the entire country of New Zealand and identify disturbance trends at high spatial and temporal resolution using widely available remote sensing data, with the goal of analyzing the effect of land management practices on local ecosystems.
Existing remote sensing capabilities require a compromise between spatial and temporal resolution. Free access to the entire Landsat archive provides a valuable resource for analyzing land cover and land use change at a very useful 30m spatial resolution; however, the 16-day temporal cycle, which is often lengthened considerably by cloud cover, limits the observation of short term changes that can result from disturbance events. The revisit cycle of the MODIS sensors aboard Terra and Aqua provides a surface reflectance dataset at much higher temporal resolution, yet at 500m spatial resolution, they lack the detail necessary to accurately track small changes in the landscape. A combination of the two products offers the ideal tool for disturbance analysis.
Disturbance detection methods in forested areas are well established, but here we present an alternative method for detecting disturbance in grassland areas. Utilizing both Landsat TM/ETM surface reflectance data and MODIS Nadir BRDF-adjusted reflectance (NBAR) covering the entire country of New Zealand (26 Landsat path/rows) for the period 2000–2012, we calculate disturbance indices in both forests and grasslands for both datasets based on normalized values of the Tasseled Cap transformation and then validate and analyze the disturbance time series. We investigate the time series to assess both the subtle changes in the landscape caused by livestock grazing and the dramatic changes brought about by plantation forestry.