Integration of paddock scale modelling and Source

Executive summary

The linkage between the Agricultural Production Systems sIMulator (APSIM) model used for paddock scale sugarcane land use constituent load estimation and the Source basin scale model is currently being undertaken by redistributing monthly loads generated from APSIM across the Source model hydrology.

This approach can result in unrealistic constituent concentrations in the basin scale model output if the paddock scale sugarcane is under irrigation. In these cases, paddock scale models tend to estimate higher runoff rates than the basin scale model. The paddock scale loads are then distributed across a smaller than expected runoff volume,resulting in unrealistic modeled constituent concentrations. This is a particular problem for pesticides, where the frequency of high in-stream concentrations, as well as long-term loads,are an important indicator of ecosystem health.

The current model linkage also limits the ability to directly translate some on-farm management practices simulated in APSIM to hydrology in the catchment outlet,such as irrigation management and recycling pits. These practices have the potential to influence both hydrology and constituent loads.

To address these problems, this study has investigated an alternative model linkage by constructing a Source plugin to directly link the sugarcane paddock scale model hydrology and loadstoSource by:

• replacing the Source-modelled sugarcane hydrology with APSIM-modelled sugarcane hydrology
• retaining APSIM-modelled sugarcane constituent loads delivered to Source
• incorporating Source-style routing and loss parameters to translate the paddock scale runoff and loads to the stream network.

Application of the new model linkage to the Pioneer (rain-fed), Barratta (Burdekin—irrigated sugarcane) and Tully–Johnstone (Wet Tropics) models for the water quality constituents dissolved inorganic nitrogen (DIN), atrazine and Diuron demonstrated:

• model hydrologic performance can be maintained or improved through the application of APSIM-generated runoff and drainage
• constituent load correlations with estimated observations (mean annual loads) can be maintained or improved
• modelled constituent concentrations in irrigated sugarcane areas can be improved to better match observations,while in non-irrigated sugarcane areas, concentration profile scan generally be maintained to those modelled previously.

The key parameters of the new plugin are the delivery ratio for drainage contribution to the stream,and the surface and drainage constituent delivery ratios. Adjustment of these ratios is available to the modeller as calibration parameters to achieve model fit. Following from this study, further work has been recommended to:

• expand the number of catchments and locations being tested by the new approach, particularly more testing of irrigated sugarcane areas
• provide guidance for selecting model parameters by investigating the applicability and mechanisms represented by the selected delivery ratios
• reduce reliance on delivery ratio parameters by understanding and adjusting constituent generation timing (application times) and magnitude at the paddock scale
• review the constituent load calculation methodologies to quantify the uncertainty in these estimates that are so important to water quality model calibration.