A Bayesian logistic regression model for...
URL: https://www.mdba.gov.au/sites/default/files/publications/A%20Bayesian%20logistic%20regression%20model%20for%20predicting%20the%20success%20of%20ecological%20flow%20indicators.pdf
Authors: Andrew Schepen and David E. Robertson
Describes the study area (Barmah–Millewa Forest and Chowilla), data used (inflow and streamflow data, site-specific flow indicators), methodology, results and outcomes of the models application.
Key findings / recommendations:
-
A Bayesian logistic regression modelling approach was developed for predicting site-specific flow indicators (SFI) achievement at two locations, Barmah–Millewa Forest and Chowilla.
-
This model uses inflow, lagged inflow and ewater volume as predictors, allows for independent intercept parameters for each indicator, and shares slope parameters across indicators.
-
The developed model skilfully and reliably predicts SFI achievement with estimates of the predictive uncertainty. The model is useful for predicting the SFI achievement with actual environmental water and also the counterfactual, e.g. for baseline diversion limits with no environmental water recovery.
-
The model may be extended spatially in the future, applied for more efficient policy planning in the Murray–Darling Basin and act as an alternative to expensive SFI generation with complex river system modelling.
You can copy and paste the embed code into a CMS or blog software that supports raw HTML