Thursday, October 30, 2014

Some Things We Know, Some Things We Don't

One of the most important papers in the atmospheric sciences was published by Ed Lorenz, the Father of Chaos Theory, in the journal Tellus in 1969.

The key finding of the paper is summarized in the second paragraph of the abstract:
It is found that each scale of motion possesses an intrinsic finite range of predictability..."cumulus scale" motions can be predicted about one hour, "synoptic-scale" motions a few days, and the largest scales a few weeks in advance.
Although we have learned that in some instances the predictability limits for atmospheric phenomenon may be longer than suggested by Lorenz in that paper, the general result that larger-scale weather phenomenon (e.g., cyclones, upper-level waves) can be predicted with greater lead time than mesoscale weather phenomenon (e.g., fronts and mountain waves) and especially smaller scale phenomenon (e.g., precipitation generated by convection) is critical to consider in weather forecasting.

A great example is provided by the forecast for this weekend.  We have known for a couple of days that an upper-level trough will be moving through Utah this weekend and that this would be accompanied by a dramatic airmass change and cooler weather.  Both the ECMWF and GEFS ensemble forecast systems have shown this for the past couple of days and continue to show it today, as illustrated by the GEFS forecast panels below.


The lower right-hand panel, in particular, is known as a spaghetti diagram.  It shows contours of 500-mb height from all 21 GEFS forecasts.  Where the lines are close, there is good agreement.  Where you see spaghetti, there is lower agreement.  All members call for a trough to be approaching northern Utah on noon Saturday, but the spaghetti is indicative of a range of forecasts for the position and intensity of the trough.

So, this is an illustration of Lorenz's theory in action.  The large-scale characteristics are relatively predictable, but the smaller-scale characteristics are less predictable.

Things get even more problematic if we ask how much precipitation will fall?  The precipitation forecast is dependent not only on the trough structure, but also the frontal structure and dynamics (a smaller scale process), the associated cloud processes, and interactions with the topography.  These smaller scale processes have a shorter predictability horizon than the large-scale trough.

This is why we can now anticipate a large-scale airmass change many days in advance, and indicate that precipitation is likely, but nailing down rainfall and snowfall timing and amounts is more problematic.  In addition, many storms in Utah are very small in scale.  As a result, the predictability horizon for rainfall and snowfall is even shorter here than found in other parts of the country where cool-season storms are larger in scale, especially the eastern United States.

This case is especially problematic since the various models and their ensembles have yet to tell a consistent story with regards to the gory details.  My best guess is that the mountains are going to see a few inches late Saturday and Sunday, but it won't be a huge event and it will probably leave us in hiker and skier Purgatory when the ridge builds back in next week.  

No comments:

Post a Comment