Are GRIB files consistently optimistic ?

I've been asked this question regularly, but until now I've never wanted to give my opinion on the subject publicly, so as not to offend anyone or create unnecessary controversy.. But after doing so many of my training courses, why not express it openly ?

My opinion, forged by observation and a long professional experience in pleasure boating (¹), is that this assertion of the systematic optimism of the GRIB forecasts regarding wind force, which unfortunately continues, no longer has a reason to be, and has not been verified in fact for some time.

Firstly, because weather models have made enormous progress in reliability over the past decades. Atmospheric and oceanic data collected has increased considerably, thanks to the multiplication of specialized satellites, to the increase in maritime and air traffic, the significant increase in data sensors of all kinds. Forecasting models have benefited from the exponential power of supercomputers and the growing skill of forecasters around the world (²). What if the wind force forecasts were always consistently optimistic, it seems obvious that this would not have been missed by the forecasters, the models would have been corrected for a long time (³).

Then the average winds at 10 meters high forecasted in the GRIB actually represent an average. And for any statistical mean there are deviations from the mean (standard deviations) more or less. Negative gaps in wind strength are of little importance, it is the positive gaps that matter to us. In this sense, systematically increasing the force of the GRIB wind was not totally wrong in the past, although the systematism is overstated. Because the amount of gap depends to a large part on the weather conditions in which we are, and is directly linked to stability. (high pressure) or instability (low pressure) of the air mass present. The gaps will be greater as the air mass will be unstable. A contrario, there is little correlation between the value of the gap and the strength of the average wind.

To overcome this inaccuracy, for a few years (2012) GRIB models provide, in addition to the forecast of the average wind at 10 meters high, those of risk of gusts (⁴). We therefore obtain a reasonable range between the average wind and the maximum winds that we are likely to encounter. It turns out, all models combined, highly efficient, especially in very disturbed situations (storm cells, cold and occluded fronts for example).

On the left the color gradient represents the average wind speed. On the right it represents the gap between the average wind speed and that of the gusts. In the meteogram the average wind curve is drawn in light blue, that of the gusts in dark blue.

In the example above, we notice, at the position of the target when a cold front passes, that the greatest value of the mean wind-gust gaps does not occur when the mean wind is the strongest, but when it decreases in its rotation behind the front (left window). The gap value then goes from simple to double (right window) : 21 knots up to 43 knots.

For the same average wind speed, the deviation from the gust speed is completely reduced in a stabilized system.

36 hours later, at the target location, the air mass being much more stable, the gap between the average wind speed, same as the previous situation, has considerably reduced (21 knots up to 27 knots).

Thus, the assertions still made by some recognized specialists in marine meteorology, despite all the respect I owe them, seem obsolete to me. For my own, since the availability of this "Gusts" data, both in the Atlantic and in the Mediterranean, I almost always found myself navigating within the range of these two values.

However, in navigation near the coast, care must be taken to use a weather model with high enough resolution to take into account the effects of thermal wind which can combine strongly with the synoptic wind. This is the case with thermal breezes, accelerations by the Venturi effect when passing certain capes, katabatic winds that can generate strong gusts, etc,. In France, only the AROME model can take these phenomena into account. But this does not mean that the forecast values ​​for the average wind are optimistic, even false.

To know more, I advise you to watch (in full) Christophe Asselin's excellent video on his YouTube channel : Why is it so difficult to predict tornadoes ? (and time in general).

–––
(¹) https://francis-fustier.fr/formations/#expertise
(²) https://www.noaa.gov/stories/weather-prediction-its-math
(³) See the ongoing evaluations by CEPMMT.
(⁴) https://www.navigation-mac.fr/weather4d-apporte-encore-plus-de-finesse/#rafales
–––