Evaluation of forest interception estimation in the continental scale Australian Water Resources Assessment - Landscape (AWRA-L) model

- Significant errors in continental scale model interception, which affect stream flow estimates.
- Errors are mainly due to underestimation of forest canopy storage capacity.
- Improvements can be made using better leaf area, rainfall and evaporation estimates.

SummaryThe AWRA-L model is a core component of the joint Bureau of Meteorology (BOM) and CSIRO Australian Water Resources Assessment (AWRA) system which integrates hydrological models and a variety of observations and satellite products to produce a National Water Accounting system for Australia. AWRA-L is a continental scale sub-model which provides surface water balance component estimates for all landscape types, including forests ranging from sparse woody vegetation in dry climates to dense tropical rainforest in wet climates. This paper describes the AWRA-L interception sub-routine, which is based on the widely used Gash model, and considers how its predictions compare with measurements of interception in tropical rainforests in north Queensland and sparse jarrah forests in Western Australia. The results demonstrate the importance of having accurate estimates of the three parameters which dominate the interception loss (as a fraction of rainfall), i.e. canopy water holding capacity, S, the average wet canopy evaporation rate, E, and the average rainfall rate, R. The current 'default' values of these parameters used in AWRA-L lead to significant errors in interception, which will also be reflected in the AWRA stream flow estimates. Analysis of how E and R values vary across the Australian continent have led to recommendations for how the interception calculations made by the AWRA-L model can be improved using values of E and R calculated from forest height and latitude. Revisions of the canopy storage parameters based on regressions derived from measured values at the rainforest and jarrah sites are also evaluated. Overall, the adoption of both updated canopy storage and E/R ratios outlined in this paper does provide a measurable improvement to the performance of the AWRA-L interception model. The approach developed in this study also has direct relevance to other applications of interception models for water balance modelling at large spatial scales.