ecPoint is a post-processing system that uses conditional verification concepts to compare NWP model outputs against point observations, and thereby anticipate weather-dependant sub-grid variability and biases at grid scale.
The main ecPoint outputs are provided in grib files and consist of:
- bias corrected rainfall forecasts at point scale (probabilistic forecast, provided in the form of percentiles).
- bias corrected rainfall forecasts at grid scale (quatitative, in mm, with the same number of members as the raw ensemble).
- diagnosed "weather type" indicators (provided for each ensemble member, for each grid box).
- ecPoint code (written in an ECMWF proprietary language called "Metview")
- Calibration "mapping function" files (computed using ecPoint-Calibrate)
Metview
Information about Metview and how to install it can be found here and in the Metview-Python GitHub repository.
Versions from Metview 5 are required.
Test Data
Before running ecPoint, the user might want to download the test data from Zenodo:
- ecPoint Vers1.0.0 and Vers2.0.0:
$ vi InParam.mv # Modify the input parameters as needed
$ metview -b ecPoint.mv # execute code in batch modeN.B: depending on the chosen settings, ecPoint might take a long time to run in series. In such a case, the user might consider parallel running.
ecPoint uses the SemVer standard for versioning. The available ecPoint versions are provided here.