...
This dataset was produced by forcing the GEFF Global ECMWF Fire Forecast (GEFF, https://git.ecmwf.int/projects/CEMSF/repos/geff/browse) fire forecast model with seasonal meteorological ensemble reforecasts. The forcing meteorological data are seasonal reforecasts from the European Centre of Medium-range Weather Forecasts (ECMWF) System5 (SEAS5), consisting of 25 ensemble members up until December 2016, and after that 51 members. The temporal resolution is daily forecasts at 12 local time initialised once a month with an horizon of 216 days (7 months) over the past period 1981-2022. The selected data records in this data set will be extended with time as SEAS5 forcing data become available. This is however not a real time service as real time forecasts are made available through the EFFIS/GWIS web services
...
A model integration at any nominal time will simulate the atmospheric conditions at a different local time depending on the location. A temporal and spatial collage of 24-h time model simulations is performed to produce a snapshot at 1200 local time. Thus, temperature and relative humidity fields are cut into, for example, 3-hourly time strips using the closest 3-h forecast output and then concatenated together so that the final field is representative of the conditions around the local noon within the 3-h resolution available. Using this method, the driving forcing are a composite of forecast outputs at different lead times in a 24-h interval and could therefore have different forecast accuracy. This inconsistency is assumed insignificant given the limited difference in forecast skills in a 24-h lead time range (Buizza et al. 1999)[1]. The most common way to achieve a 12 local time field is through a concatenation of fields. Adjacent stripes of forecasts can be sliced together as highlighted in Figure. The simplicity of the approach however implies that artefact can be introduced at the interface between two slices.
...
ECMWF has developed a new interpolation method performing a weighted average between the two closest timesteps. Below is a graphical explanation of the way the two methods are implemented is provided. If we extract the time series of two locations 1 and 2 near a change of time, the 24 hours forecast will provide the diurnal time for temperature (here chosen as an example) in those points. If we are interested in the value of temperature at 12 local time this will be the value of the prediction at 12 UTC for point 1 and 15 UTC for point 2 in the assumption of a 3-hour resolution forecast. The choice of the different forecast will create a discontinuity in the fields as depicted in the map. The new method instead will interpolate for both points between the value at 12 UTC and the value at 15UTC by weighting the two temperatures by their closeness to any available forecast. This new method provides a much closer agreement to the real diurnal cycle and removes the boundary artefacts that were presented above.
|
|
|---|---|
Interpolation methods to obtain a 12 local time composite fields. The "stripes" method implements a nearest neighbour approach, the "interpolation" method adopts a weighted bilinear interpolation between successive time stamps. The left-hand side figure shows the collated temperature fields for one sample day. The right-hand side figure shows the diurnal cycle for temperature in the two points indicated with “1” and “2” and the corresponding interpolated value in the two cases | |
Usage notes
Download global fire danger forecast maps
...