I'm looking to retrieve from weather forecasts (such as GFS or HRRR) the U and V component of the wind with respect to some altitude (meters) above mean sea level.

To do so, I've implement a interpolate following the same idea as in the link below. https://stackoverflow.com/a/61107634/6528830

However, I'm confused by the results it would seem like part of some pressure level would be underground.

If I look at the geopotential height @ ground or water surface. It seems to match with terrain elevation. Yet I don't believe it matched exactly, is this normal? enter image description here

However, it's the 1000 mb geopotential that really confuses me: enter image description here

How could some points be lower that in the gph@ground? What am I not understanding about this data and how can I use it to figure out what is the wind at specifics altitudes above mean sea level?


1 Answer 1


It makes sense, some locations have a surface pressure under 1000 mb... which means to get to 1000 mb you'd have to go below ground (the basic idea of the calculation is that pressure decreases as you ascend in the atmosphere... so to get it to increase from sea level pressure, you'd have to go the other direction, underground).

It certainly feels hints of being unrealistic... but I would think if you dug such a deep enough hole wide enough, you would get to the given pressure.

Here's the sea level pressure map similar to your 1000 mb map:

[1]: https://i.stack.imgur.com/POw1P.png (Source: weathernerds.org)

Those negative height areas indeed match a chain of lows with surface pressure below 1000 mb in the Southern Ocean and a 992 mb low over Europe. So if the station is 992 mb... you'd have to even lower in the atmosphere (= negative) to get to 1000 mb.

This is actually a common interpolation on pressure maps in meteorology... mean sea level pressure itself... probably the most common map we use... is found at a height that is actually below ground at most places, because almost all locations are somewhat above sea level. But by comparing on a flat surface it allows us to better identify the weather-causing features, factoring out the otherwise overwhelming factor of pressure decrease with altitude above sea level (your ears pop on mountains despite being at 0 ft above the ground because the atmospheric pressure is still less because there's less atmosphere above you)

Even at 850 mb, another common geopotential height for analysis... we can plot the heights above sea level... but we often will choose to mask out the model's winds in spots to show the 850 height is truly subterranean there, so the model's predicted winds are of imaginary quality, being below ground level:

enter image description here (From pivotalweather)... the grey is masked out wind.

So all your plots looks quite reasonable, and hopefully that helps make sense of the concept. Not every location has 1000 mb (or even 850 mb) in its atmosphere... both because some sites have sea-level pressures below that value (the cause of the issues on your map, as geopotential is based upon sea-level), and also because the elevation of some sites means lower surface pressure (not the cause of your negative geopotential heights, but the cause of the wind maskout on some such plots as shown).

  • $\begingroup$ Thanks for the clarification. When it comes time to mask out data, do you use terrain data or do you compare the isobaric level ghm to the surface ghm? $\endgroup$ Apr 8, 2022 at 20:30
  • $\begingroup$ I'd think they'd work out the same, whichever is easier for your coding flow? If I were doing it and just looking for what's easier, I'd probably just compare to the surface ghm if I already was dling that data. Then again, if I were doing it in an operational environment (did about 10 years ago), bandwidth and speed matter, so maybe I'd use a fixed terrain elevation file. $\endgroup$ Apr 9, 2022 at 3:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.