Historical Land Cover Assessment: Challenges and Achievements

This study aimed to develop a method for assessing changes in vegetation greenness and land cover over time at both landscape scales and for individual land parcels. It is hoped that an improved understanding of land cover history will increase understanding of soil condition, and linkages between soil health and productive capacity in Victorian agricultural landscapes. Gradual changes and differences in vegetation biomass can occur due to climatic and geomorphological factors. Sharper transitions in vegetation biomass coincident with cadastral boundaries and paddock fences also occur as a result of management history. Understanding the earlier land use history with respect to these parcels may reveal causal factors with respect to cover growth (e.g. extended fallow periods) that would explain, in part, the reasons for differences in current land productivity. This study utilised the historical Landsat archive to produce land cover and vegetation greenness information over a 38 year period in a localised area of the Wimmera region of Victoria. Data generated by this project will support efforts to examine the influences of farming systems on productivity and soil resources over time frames equivalent to soil change. The likely causes for local differentiation could include seasonal climatic variability, crop/soil management practices or local soil constraints. This paper presents some preliminary findings in this ongoing study. The work presented here focussed on creating land cover and production histories for the study area and selected paddocks from satellite imagery. The project made use of the USGS Landsat archive that yielded 54 images from July 1973 to November 2011, which were largely clear of cloud cover. For each image, Normalised Difference Vegetation Index (NDVI) was calculated and a land cover classification using spectral signatures was generated. These data were interpreted in a number of ways. A time series of NDVI values for paddocks of interest was created, which showed varied patterns between paddocks. Cumulative NDVI across the landscape was also calculated which revealed distinct paddock boundaries as a result of land management history. These data also revealed a north-south transition across the study area from mostly low NDVI to mostly high NDVI, which is likely the result of climatic and soil property variation. Analysis of land cover also allowed land management history to be interpreted in terms of crop rotation and paddock fallow duration. The results so far provide clear evidence that the influence of long term management practices on productivity (vegetation cover) can be mapped using satellite imagery.