AWRA‐R version 5.0 Calibration Tools

[摘要] This report described the calibration tools used for parameterisation of AWRA-R v5.0 model. A large number of parameters need to be calibrated for the implementation of the AWRA-R model in a river basin. One of the developmental goals of AWRA-R modelling system was to build a customised auto-calibration system for parameterisation of the model in a most computationally efficient way to roll out the operational model to large areas for national water accounting. The current AWRA-R modelling system includes two auto-calibration approaches: i) reach-by-reach calibration and ii) system calibration.This report describes the two calibration approaches, the guidelines on how to use the two approaches and details a case of study of their implementations in the Murrumbidgee region of the MDB.The background and objective of AWRA-R calibration tools are discussed in section 1. The technical details of the two calibration approaches are elaborated in section 2. A set of guidelines for implementing the two approaches are presented in section 3. A case study example of implementation of the two calibration approaches in the Murrumbidgee region of the Murray-Darling Basin is presented along with the results and performance analysis.The results of the reach-by-reach calibration in the Murrumbidgee show highly satisfactory performance of the AWRA-R v5.0 model. The median daily NSE was 0.60 and median annual bias was less than 3% for the period of calibration (1970-1999) for the Murrumbidgee region. Under the validation mode, the median daily NSE was 0.70 and median annual bias was 15% for period of 2000-2014.The system calibration approach produced better bias than reach by reach calibration since all mass balance errors were propagated to all downstream gauges during calibration, allowing the optimiser to select parameter sets that preserved mass balance better. System states produced by the system parameters were far less variable, showing reduced likelihood of localised over-fitting.The system calibration is much more computationally intensive calibration approach compared to reach-by-reach calibration.

[发布日期] 2015-10-08 [发布机构] CSIRO

[学科分类] 地球科学(综合)