In the "About these data" for the "Thunderstorms" Layer provided by ECMWF, it defines the layer as "Defined as number of lighting flashes in the area of one square kilometer, in one day." However, it appears this definition is incorrect.
The unit is "l/km²", this seems correct, but seeing as the layer changes for every available time frame resolution (i.e. every one hour for the next three days, every three hours for three to five days from now, and every six hours for five to eleven days from now), it appears that it actually shows the number of lighting strikes per square kilometer per hour, or possibly per time frame.
Another thing I would add is that rather than stating that it is "per square kilometer", I would think a more accurate definition would be "...approximately within a radius of one square kilometer." I believe this seems to be a more accurate definition for two reasons. The first being that the model only has a resolution of 9km, therefore this must be approximate. Secondly, the data is not displayed in one kilometer "grids", rather, it is displayed in a gradient that can have intensities which vary in increments less than one kilometer.
Additionally, actual observations comparing real world conditions to the data presented in this layer appear to support this (i.e. when looking at the layer for the current time and location, during a thunderstorm, it seems to show the amount of lightening strikes stikes which occur within one kilometer of me [approximating distance from lightening strike using time between lightening and thunder] in a span of one hour).
So I am almost positive that this layer is displaying the approximate amount of lightening strikes occuring within one kilometer in an hour, at least for the next three days (three day forcast). However, what I am unsure of, is what the layer is representing past the three day forecast, when the time resolution is over one hour (three hours up to five days, and six hours for the rest of the forecast). When the forecast data is for a three hour period, does the data represent lightening strikes occurring within a given one kilometer radius in three hours, or does it still represent one hour during that period? When the forecast data has a time resolution of six hours, is the lightening data for lightening strikes in a six hour time frame? Or, is it always for a one hour time frame? Alternatively, it may be possible that lightening strikes are always for a six or three hour window- it is difficult to approximate, since thunderstorms are usually isolated and move fairly quickly, so while there may be three or four thunderstorms in a given day, the peak lightening that strikes a given area usually occurs within less than an hour.
So can anyone help ifentify exactly what this data layer actually represents (since it is almost certainly not lightening strikes in a given day)? Where can I find additional information about this layer? If we can determine the layer description is incorrect, the in should probably be fixed in the next update.
Thanks!