THE ENORASIS Meteorological Analysis Tool

To address the need for uncertainty – quantified rainfall forecasts for water management purposes within the ENORASIS project, high-resolution rainfall forecasts are adopted in a probabilistic manner using the state-of-the-art Numerical Weather Prediction (NWP) model WRF. The Weather Research and Forecasting (WRF) Model is a next-generation mesoscale forecast model and data-assimilation system designed to serve both operational forecasting and atmospheric research needs.

In existing implemented systems, information on weather is delivered by weather stations and weather forecasts are based on historical data rather than models using satellite images. This approach lacks in cost effectiveness and geographical coverage as it requires the installation of an adequate number of weather stations in the field in order to produce reliable data. ENORASIS uses the WRF model that for the purpose of the project will be set up with multiple grids of different spatial resolution over selected pilot sites.

On the whole, the high-resolution rainfall data is generated in a three steps approach:

  • Ensemble Pre-Processor The Ensemble Pre-Processor generates the short-term   ensemble members for our computational grid, based on available forecasts by   ECMWF. Further, it collects relevant information for data assimilation such as near surface winds, total precipitable water, water vapour, cloud-drifting      winds and water vapour derived winds from various operational satellites and instruments. An analysis of satellite imageries is performed in order to provide the      most accurate data to the    Ensemble Data-Assimilator.
  •  Ensemble Data-Assimilator: The Ensemble Data-Assimilator generates optimal    initial states for the WRF model to produce improved ensemble rainfall forecasts. It employs a particle filtering algorithm to account for the uncertainty in the initial state of the model.  
  • Ensemble Output: This process consists of the rainfall modeling for each   ensemble member under different configurations (initial conditions perturbations,         physics perturbations + data assimilation) and the statistical post treatment of the ensemble output.