Operational Models

brahms

Making the models operational, to address extreme events (floods, droughts), and provide preventive alerts

     The region is subject to extreme events. Major storms come through, resulting in very high levels of rainfall, leading to flooding. Such flooding is partially dependent on existing levels of soil moisture. If soils are too dry, droughts can develop, and impact agriculture and water supply. It would be very useful to have “early warning systems,” that can track current conditions, and provide advance warning. The question is,

If the model is maintained in an operational mode, could the actual conditions of rainfall intensity and soil moisture be monitored, and could extreme events (floods or droughts) be anticipated via the development of a preventive alert system?

     The VIC and DHSVM models have been shown to be highly functional in modeling the hydrologic regime of Espírito Santo on both historical conditions and scenarios of the future. In principle these models could absolutely made to be operational, and provide alert capabilities. The core strategy would be to have the model(s) running, and continuously updated, “remembering” what “yesterday’s” conditions were, and needing only “today’s” weather to predict what would happen over the next time steps. DHSVM was made operational several years ago, coupled to the MM5 regional climate model.

     To do so, several requirements must be met.

     Requirement #1. The exact objective(s) would have to be specified. A system designed to address very fast time scales floods would operate under a different set of rules than a system addressing drought development. The same model could be used, but climate drivers would be required. A system focusing on floods would need the latest weather predictions, while a system looking at droughts would require weather projections.

     Requirement #2. For operational considerations, accuracy is of high importance. The models are currently set up with quite generic parameters. To improve accuracy, the base dataframe ought to be improved (e.g., more realistic values for the rooting depth of eucalyptus, and of soil hydraulic properties).

     Requirement #3. Depending on the exact objectives, and reliable and robust computer infrastructure would be required.