Modelling the response in streamflow to increased forestry plantations

Abstract

Plantation forestry is an important land use in Australia which has the potential to reduce water yields compared with existing agricultural use. However, because of the biophysical complexity of landscapes, the hydrological impact of plantation expansion will vary between catchments and between locations within catchments. This paper investigates whether incorporating the history of plantation expansion in a catchment improves the prediction of streamflow by capturing the lag in catchment water balance and streamflow response in response to changes in plantation area and age. The study area for this project was the Glenelg Hopkins Region, South West Victoria, Australia; specifically the Crawford River, Eumeralla River and Darlot Creek. These catchments were selected as they were historically dominated by pasture based agricultural systems (>60%) and had seen a significant increase in plantation forestry over the last 10-15 years. The plantation history was estimated using the best available data, including an analysis of plantation data supplied by Green Triangle plantation growers as well as data from miscellaneous sources such as aerial photography and vegetation cover data. The majority of plantations in the Darlot Creek (<15% catchment) and Eumeralla River (<15% catchment) catchments were established in 2000-2005, while in the Crawford River (~30% catchment) small areas of radiata pine plantation were established between 1960-1970 and blue gum plantations were established after 1995. The potential impacts of plantation forests on streamflow were investigated using the Catchment Analysis Tool (CAT, DEPI Victoria). The CAT was selected to investigate plantation forestry, due to its ability to incorporate land use change history over time within a spatial context to predict catchment water use and streamflow. In this study two component models, CAT1D and CATNode, were used to link paddock-scale land use, soils, topography and climate data to catchment-scale groundwater systems and streamflow at the gauge. The CATNode model was calibrated using the Plantation History land use layer as the best available history of land use and land use change in the catchment. To investigate whether the inclusion of plantation history improved the prediction of streamflow compared to the Static land use, the calibrated model was rerun using the Static land use layer and the difference in cumulative flow was analysed. CAT was able to provide an adequate streamflow prediction over time (monthly Coefficient of Efficiency >0.7). In addition the incorporation of timing and pattern of plantation history demonstrated a small, consistent improvement in temporal streamflow predictions in each of the three catchments, with the results demonstrating a delayed pattern (4-5 years) in streamflow response consistent with the timing and pattern of plantation history. Due to the hydrological complexity of landscapes, the impact on streamflow of future plantation expansion will vary across the landscape; CAT presents a valuable tool for investigating the potential impacts of future plantations.