Outcomes of The Living Murray Icon Sites Application Project Stuart Little Project Officer, The Living Murray Environmental Monitoring eWater CRC Participants Forum 23 rd November 2011

Context – The Living Murray The Living Murray aims to return an average of 500GL (long term cap equivalent) per annum of environmental water to 6 Icon Sites in the Murray system. Environmental works and measures are being constructed to support water delivery.

Scope of Project The Murray Icon sites Application will produce a tool which: –models ecological responses to watering scenarios; and –informs trade-offs and decisions when prioritising environmental water allocation under the MDBA's Living Murray program. Eco Modeller will convert daily time series data from MSM Bigmod scenario runs into measures of environmental consequence for each icon site. These environmental consequences can then be compared across sites for different environmental watering strategies.

The Plan Implement existing ecological response models from the Murray Flows Assessment Tool (MFAT) Link ecological response models to improved hydrological modelling developed for the Icon Sites Refine MFAT ecological response models using findings of monitoring and research at the Icon Sites since MFAT development Develop easily interpreted reports of modelling results to inform the prioritisation of environmental water allocation under the MDBA's Living Murray program

Tasks of the Project Phases Phase 1 – 2008 to 2010 Linking MDBA hydrological modelling to Eco Modeller Modelling system development Conceptual models of Icon Site responses to flow and ID key ecological assets Developing and implementing Ecological Response Models Compilation of research data and resources Designing and implementing methods of aggregation and reporting of model results Phase 2 – 2010/2011 Refine Eco Modeller as required Add to MSM Bigmod, to provide the input time series required Develop relevant models for the remaining Icon Sites Develop new models for pest species Model testing to calibrate and validate habitat prediction models Design and implement an aggregation and reporting format

Linking MDBA Hydrologic Modelling to Eco Modeller – Phase 1 Developed a data handler that: –Identifies Bigmod files –Allows users to select a time series for a particular point in the Murray system –Search function to make selection of time series easier

Modelling system development – Phase 1 Aim to fully implement the MFAT models into Eco Modeller, then upgrade the elements using additional science and data. Base models in MFAT had a level of complexity that was not able to be duplicated in the original design of Eco Modeller New ecological models relevant to the Icon Sites were also too complex Need to develop a structure that would allow various numerical model forms, that must also conform to providing output in a similar way

Summary Limitations Confidence Reporting Generate single metric to compare scenarios Metadata Plug –in model

Developing and implementing Ecological Response Models – Phase 1 MFAT contained models for up to 30 assets per icon site –an asset may have several life history stages; and –be calculated at several sub sites for each site –used a range of complex equations that also included scores from variables that remain static when comparing scenarios 30 assets x 6 Icon Sites x 2-3 habitat requirements/life history stages equals lots of models To handle the implementation of so many models a generic species plug in model was developed

The TLM Generic model principles A successful water event can be characterised as: Magnitude or Threshold –minimum depth, discharge, spatial extent Duration –how long does it last Timing –what time of year does it occur Period between events –how long has it been since the last successful watering Rate of Change –the speed at which the water rises and falls

Bigmod Output Event Duration Timing Return Period Result Threshold

Ecological Modelling – Phase 1 To demonstrate the application of Eco Modeller, two alternative flow scenarios were modelled, Natural and Current. Simple species based comparison showed that in most cases, the Current scenario had much lower average habitat suitability scores. The current scenario also had a marked increase in the number of poor years across most models. There were also some marked differences between species and even between the life history stages of the same species, demonstrating the value of modelling habitat requirements at the life history stage level.

Outcomes from Phase 1 The Eco Modeller software changed significantly, the changes show the benefits of eWaters Application Project/Focus Catchment approach to software development. The results provided an indication of the potential of the Eco Modeller software to inform water allocation at the flow-regime scale (i.e. >10 years). Changes to software functionality, the individual ecological response models and improved aggregation and reporting were recommended to further refine the product to address the original scope of the project.

Next Steps – Phase 1 leading to Phase 2 Compilation of research data and resources for all Icon Sites Testing and calibration of the ecological response models Review of model outputs for various scenarios and comparisons to actual watering events Creating aggregation, reporting and interpretation tools Creation of new and/or modification of existing ecological response models into the future as greater knowledge becomes available

Testing and Validation - Phase 2 In order to use the habitat suitability models to inform water resource management options, the models should ideally be well calibrated to observed habitat condition output. Limitations biological measures are not the same variable as that being predicted by the models lack of long term biological datasets against which to test The approach used here was to test the performance of the habitat suitability models against these small biological datasets.

Testing and Validation - Phase 2 Key Steps compile appropriate biological data, model areas of inundation for Hattah Lakes (for fish and vegetation analysis) and for all icon sites for the bird analysis, test habitat suitability models with the available data as well as relate the biological data to simple measures of inundation

Findings of Testing and Validation Waterbird models as they stand have limited use due to: lack of long term suitable data limited understanding of links between inundation and waterbird responses Floodplain vegetation models improved via the adaptive process: incorporation of rainfall and a decay factor improved correlation between field data and model scores However, the correlation for vegetation models were typically low. The limited frequency of field sampling is a limitation for validating the fish ecological response models, and indeed the waterbird and vegetation models as well.

Lessons Learnt and Recommendations Project outcomes are directly proportional to the ability of the coordinator to drive the project forward Calibration and/or validation of models requires long-term, appropriate datasets The development and maintenance of Eco Modeller model libraries needs to be treated like an adaptive management process Consider the possibility of incorporating uncertainty into the ecological response models, e.g. by a Bayesian network approach

Outcomes of the Application Project We did: Use a real-world application of Eco Modeller to improve the tool and add greater functionality that was relevant beyond the TLM Icon Sites application Develop a generic model format that can be used to build ecological response models based on flow requirements We did not: Successfully develop and validate models of ecological response for the TLM Icon Sites Develop an aggregation and reporting tool that could be used to inform the prioritisation of environmental water allocation under the MDBA's Living Murray program

Thank You For further information: Stuart Little Murray-Darling Basin Authority or