Linking Models to Data

In recent years, real time data has become more readily available from automated earth based sensors (e.g. soil moisture probes, hydrometric and weather stations etc. ) and air based platforms (e.g. drones, radar, satellites etc). Environmental prediction models can be built to take advantage of such information by including data assimilation technology with Bayesian based adaptive tools/ Kalman filters, among others. This technology can automatically ingest real time data as they become available, in order to improve forecasts. Apart from the data assimilation capacity, adaptive models can be designed to use feedback mechanism where errors in previous predictions are utilized to improve future forecasts. In addition, model parameters can also be allowed to change with time during the simulation period, a process particularly attractive when parameters are very sensitive to environmental variabilities. In particular hydrologic/ hydraulic/ agrometeorological models with automated adaptive features can be very useful in flood forecasting, reservoir operations, irrigation scheduling and other environmental predictive systems.