Saturday, March 16, 2013

Risks and Rewards of Resolution

Earlier this week, we started downloading and processing forecasts from the NAM high-resolution nest, which covers the continental US at 4-km grid spacing, which is three times the resolution of the regular NAM.

We will start to see if the greater resolution proves useful for Wasatch forecasting over the next few weeks.  During the 2002 Winter Olympics, we found that a 4-km model did produce more accurate precipitation forecasts over northern Utah than a 12-km model, but also with more false alarms.   The 4-km model got more forecasts right, but when it was wrong, it was a disaster.  Thus, like a well-designed golf hole, resolution offers up risks and rewards.

We have a dust-on-crust storm in the offing for tomorrow, so I thought we would do a quick comparison.  First the 12-km, which lumps all the major mountains of northern Utah, including the Stansburys, Oquirrhs, Wasatch, and Uintas, into a single broad mountain that is highest where the western Uintas sit.  As a result, the greatest precipitation during this event is forecast to fall in that area, rather than over the Wasatch range.  Alta gets .16" of snow water equivalent in the three hours ending at 1200 UTC (0600 MDT) and .06" in the 3 hours ending at 1500 UTC (0900 MDT).

The 4-km nest, however, begins to resolve the Stansburys, Oquirrhs, Wasatch, and Uintas.  They aren't well resolved, but they are more independent.  This leads to separate precipitation maxima over these ranges in the 3 hours ending at 1200 UTC (0600 MDT) and, after things taper off over the Stansburys and Oquirrhs, maxima over the Wasatch and Uintas in the 3-hours ending at 1500 UTC (0900 MDT), including a maximum over the Cottonwoods.

That certainly looks better compared to what we would expect climatologically, which features more precipitation over each of the mountain ranges with less precipitation over the intervening valleys and lowlands.  At issue, however, is whether or not the 4-km NAM can reliably produce those spatial patterns.  For instance, if this storm comes in a bit drier, or ends up ging to the north just a bit more than forecast, the added detail and precipitation will cause a bigger forecast bust than the 12-km NAM.  That's the risk and reward of resolution.  

One case isn't a big enough sample, but let's what happens late tonight and tomorrow morning and during storms in the future.  Personally, I hope it verifies and perhaps even underdoes the precipitation.  We need every flake we can get.  

No comments:

Post a Comment