New point rainfall products in ecCharts

Tim Hewson, Fatima Pillosu, Axel Bonet, Iain Russell, Cihan Sahin, Florian Pappenberger


New experimental ‘point rainfall’ products with global coverage have been available in ecCharts, a web-based application providing easy access to ECMWF’s medium- and extended-range forecasts, since 17 April 2019. They are the result of an ECMWF product idea which reached the final of the Harry Otten Prize for Innovation in Meteorology at EMS 2015. The idea was followed by considerable development work and the creation of a new post-processing suite called ecPoint. The post-processed products add value to existing products by accounting for the different degrees of sub-grid variability and grid-scale bias that one sees in different weather situations.

Example comparing raw ensemble and point rainfall charts. The charts show the 95th percentile for the raw ECMWF ensemble forecast (left) and for the corresponding ecPoint-Rainfall forecast (right) for 06–18 UTC on 2 April 2019 (T+90 to T+102). Although it is occasionally not the case, in this particular example the areas where there is a 5% chance of large accumulations, for example 60 mm or more in 12 hours in the Argentina–Paraguay border region, are more extensive in the point rainfall product than in the raw ensemble.

The new ecCharts products show probabilistic forecasts of 12-hour rainfall for any location (i.e. representing totals as measured by rain gauges). These complement the grid box mean values that the raw ensemble provides. Output for overlapping lead time ranges is available up to day 10 (T+0 hours to T+12 hours, T+6 hours to T+18 hours, ..., T+234 hours to T+246 hours). In the first release, users can display point rainfall in one of two ways. They can either select a percentile (1, 2… 99) and display the corresponding field of rainfall threshold values (mm), or select a rainfall threshold and show a corresponding field of probability (%, of exceedance). New colour schemes are available for both approaches, and these are duplicated for the raw ensemble, allowing users to easily compare the two. Potential applications include issuing warnings – e.g. for flash floods – and more accurate forecasts of dry weather at a given location. We would welcome feedback on the usefulness of the products. User guidelines are now available in ECMWF’s online Forecast User Guide (