Monday, April 7, 2014

Changing Pressure Gradients

A post by Brian the other day led to some discussion about long-term trends in wind speed over Alaska and the lower 48 states - specifically, whether the observed weakening in wind speed is "real" or is at least partially an artifact of changing measurement procedures.  To look a little closer at the changes in Fairbanks, I calculated the mean annual gradient in sea-level pressure from the NCEP/NCAR global reanalysis, which extends back to 1948 at six-hourly time intervals with a horizontal resolution of 2.5 degrees latitude/longitude.  Admittedly the data is spatially coarse, but it does provide a consistent gridded history, and for sea-level pressure the quality of the synoptic-scale variations should be very good.

The chart below shows the results obtained for a reanalysis grid point close to Fairbanks: the black line depicts the annual mean pressure gradient, and the colored lines show the mean gradient in three-month seasons.  It is interesting to note that the mean annual gradient was slightly lower than earlier norms for about ten years beginning in 2000, but it was slightly higher again in 2012 and 2013.  The largest decline from earlier years occurred in winter and fall, whereas spring and summer showed no significant change.

The drop in mean pressure gradient since 2000 is consistent with lower mean wind speeds at Fairbanks, suggesting that the wind speed really has slackened in recent years.  However, note that the annual pressure gradient did not begin to drop off until 2000, whereas the reported wind speed dropped off quite precipitously in 1998, so I think it's likely that there is also an equipment or measurement change involved in the wind speed history.  Of course, we should remember that the surface wind speed is also affected strongly by the inversion and stability characteristics of the air near the ground, so the pressure gradient is not the only determining factor.

The spatial scale of the pressure gradient change is illustrated in the map below, which shows the difference in mean gradient since 2000.  Most of Alaska stands out prominently as having a reduced pressure gradient, while the Arctic Ocean has experienced higher gradients in recent years.  The changes in the lower 48 are rather small, except in New England where weaker gradients are consistent with Brian's map showing weaker wind speeds.


  1. Might the apparent recent changes in wind speed and direction (surface and jet stream), and pressure gradients, be due in part to the Arctic Amplification theory described by some (Jennifer Francis et al., for example)?

    Not to start a climate debate here, but just maybe there's a definable source for the changes observed.

    Gary (back to winter in Fairbanks, forget the early Nenana lottery guesses)

    1. Gary, there certainly seems to be a marked boundary in the pressure gradient trends near the Arctic Ocean coast, both in Alaska and Siberia, so I suspect Arctic changes have a lot to do with this. Mechanism unclear, however.

    2. Here's some references for interested parties regarding the current theory of Arctic Amplification and the presumed effects. The debates continue regarding the suggested mechanisms. There's more on You Tube via the following link.

      Francis et al. (2009) would be worth reading as well.

      This to me is a basic what and why debate regarding climate patterns. The what can be examined (as this Blog and others have), meanwhile the why remains a topic for active discussion.


  2. Thanks for going to the trouble to investigate this further Richard. A couple of thoughts. First, since wind is strongest during the coldest 6 months of the year (on average) the trend in pressure gradient force in those months would have the largest effect on wind speed. Also, since the EC reanalysis has a much finer spatial resolution, it might show a clearer trend (if one exists) for the SLP trend. I haven't figured out the EC reanalysis site yet so if anyone out there knows how to use it I would love a lesson. I think a good idea for a future study would be to identify a handful of stations with long records and look at their hourly observations to better tease out wind speed and pressure changes. Tundra sites with no exposed vegetation would be ideal since much of the hypothesized variation has been attributed to land use and land cover changes.

    1. Brian, winds aloft may be stronger in winter, but in Fairbanks winter is the least windy season despite stronger pressure gradients - see the climatology here:

      I have never gotten my hands on the EC reanalysis, but I would like to - it is undoubtedly superior.

  3. I was wondering if

    1. The red near the pole could be partially an artifact of gradient calculations. This since the reanalysis is simply taking real data and running it through a model.

    2. If real, wouldn't stronger recent arctic gradients mean a stronger polar vortex? But hasn't the vortex been weaker recently as evidenced by the wobbly jet stream.

    I really do like this blog. So much more relevant mental gymnastics than your typical climate blog.

    1. Thanks, Eric. To your questions:

      1. The reanalysis data is supplied on a lat/lon grid which gets very fine in the east-west direction near the pole, but I wouldn't think this causes a problem; the gradient calculation accounts for the varying grid spacing. The reanalysis model itself is spectral so doesn't operate on the lat/lon grid.

      2) At sea level there is not really a polar vortex as the annual mean pressure is much lower in the North Atlantic and North Pacific; there is a relative maximum in the Beaufort Sea. The polar vortex of recent fame is really a stratospheric phenomenon; it was strong this winter but presumably weaker in many recent years. I view the increased gradients in the Arctic as indicative of more frequent and stronger summer and autumn storms with loss of ice cover.