Spatially-explicit modelling for catchment-level salinity management

In this paper, a spatially-explicit model to undertake catchment-level analysis of dryland salinity is discussed. The model uses a raster-based approach where a catchment is represented as a grid of neighbouring cells. Each cell is defined by a set of seven attributes: land use, elevation, soil type, rainfall, aquifer thickness, groundwater-table depth and groundwater salinity. Cells receive and transmit groundwater information to and from neighbouring cells through a simplified hydrology model. The hydrology model is combined with an economic model and can be used to analyse the effect of alternative spatial patterns of land use. The model is implemented in the MATLAB programming environment and is designed to allow users to test any arbitrary pattern of land use and explore its long-term consequences. This facility permits the analysis of tradeoffs between financial (profit) and environmental (salt-affected area, water yield and water quality) outcomes. The model is illustrated in an application to a small agricultural catchment in central-west NSW, Australia. Attribute maps for elevation and soil type are read directly from ASCII grid files generated by GIS software. Rainfall is assumed to be uniform across the catchment, and monthly time steps ensure water movements between neighbouring cells in this small catchment are mimicked accurately. The model is initialised by reading a look-up table for land uses and their respective parameters, and a look-up table for soil types and their respective parameters. The user can change the number and types of land uses or soils and their associated parameters simply by changing the relevant look-up table in a spreadsheet. Three experiments were run where the catchment was entirely planted to a single land use, either tree belts, grazing on perennial pasture or annual cropping. Preliminary results demonstrate that the model responds to the different land-use scenarios in accordance with apriori expectations, with tree and pasture land uses generally resulting in lower-groundwater tables. The model is in the process of being calibrated and validated using time-series data for groundwater-table depth for several piezometers located across the catchment and some issues are discussed.