The storm that won't form

Looking out into the tropics today.  Last week, Tropical Storm Don failed to live up to expectations in south Texas.  The rainfall associated with the storm dissipated rather quickly, bringing little relief to that drought-stricken area.

However, we have a new disturbance in the tropical Atlantic, labeled by the National Hurricane Center (NHC) as Invest 91L.  Here's how it looked early this afternoon on visible satellite.

GOES-E visible satellite image of invest 91L, 1815Z, Aug 1, 2011.

It's definitely a somewhat-organized area of convection and the NHC says that tropical storm force winds are already being reported in the vicinity as it approaches the Lesser Antilles.

However, this storm has been on the watch list for days--since Friday, the NHC has maintained on its website that this disturbance has an 80-100% chance of developing into a tropical depression or storm in the next 24-48 hours.  Three days later, they still haven't officially upgraded it yet.  Why not?  Hurricane hunter airplanes flying through the storm failed to find a closed circulation about a clear low-pressure center.  Without that level of organization, the NHC doesn't assign storms a unique identity as a depression or as a storm, even if the winds are of tropical storm strength.  So, we continue to wait...and watch.

The tropical cyclone modeling community is ever-expanding, and they're trying to solve what is becoming one of the classic dilemmas of current numerical weather prediction.  As our computer models have become more advanced, our ability to forecast cyclone tracks has increased, while our ability to forecast cyclone strength really hasn't changed much at all.  Many, many scientists are working on that problem, and the result is many fancy model configurations that try to get it right.

Here are three examples of advanced hurricane weather models and what they are predicting will happen with this storm.  I'm going out to the end of their model run at 126 hours--that's over 5 days.  Already we're stretching the limits of predictability even in relatively quiet weather conditions.  But, I wanted to illustrate the differences in these models in both track and intensity over this time frame.

I got these images from Bob Hart's Tropical Cyclone Genesis Potential Fields page--my absolute favorite page to visit to get a nice summary of what all the models are saying with respect to possible tropical cyclones.

First, here's the Geophysical Fluid Dynamics Laboratory (GFDL) model forecast for 126 hours out from 12Z this morning:

GFDL 126 hour forecast for incest 91L at 18Z, August 6, 2011.

The left panel shows a wider view of the tropical Atlantic and eastern US.  You can see that by August 6th (Saturday), the GFDL model forecasts the storm to be well off the coast of Florida and rather small in size.  The right-hand panel shows a moving nested model grid inside the large model.  This shows the output of a second model over a small area centered about the storm at much higher resolution.  In theory, the overall synoptic pattern in the larger-scale model will help determine where the storm will go and the higher-resolution, moving nested grid will help determine the storm strength.

One thing I really like about the graphics on this particular website is that they've color-coded their wind contours to match the different levels of tropical cyclone strength.  For instance, any areas in cyan are where winds are of tropical depression strength.  Dark blue is tropical storm strength.  Green corresponds to category one hurricane strength, yellow for category two, and so on from there up.  This lets you see at a glance what the structure of the wind field is forecast to be by this model, and also what the maximum intensity would be.

In the GFDL case above, there's a significant area of green, signifying category one strength, with a few small pockets of yellow (category two) strength.  So this forecast offers the possibility of a category 1-2 hurricane well off the southeast US coast by Saturday. The minimum pressure is given as 978.9mb.

Now here's the Hurricane-WRF model (the HWRF) forecast for the same time:

HWRF 126 hour forecast for incest 91L at 18Z, August 6, 2011.

There are significant differences between this and the GFDL model.  First, the storm position is different.  The HWRF is predicting it to be much closer to the Florida coast on Saturday.  Also, in this model the storm is predicted to be much larger in area, with tropical-depression- and tropical-storm-strength winds over a much larger radius from the center.  However, in terms of strengh, I don't see any signficantly visible areas of category two strength winds in this output.  The minimum pressure is forecast to be lower--971.4 in this model as opposed to 978.9 in the GFDL model.  So lower pressure, but also weaker winds.  Still probably looking at a high-end category one hurricane in this model.

And now, another version of the HWRF, but this one with modifications from NOAA's Hurricane Forecast Improvement Plan (HFIP) initiative.  This group has modified the HWRF model somewhat and also runs their smaller nested grid at very high (3 kilometer!) horizontal resolution.  Here's their forecast for the same time:

HWRF-HFIP 126 hour forecast for incest 91L at 12Z, August 6, 2011.

This forecast is actually from 6 hours earlier, but it paints an even more sinister picture.  In this model, the storm is very close to the Florida coast by Saturday morning.  The storm also is forecast to be much stronger--there are significant areas of orange in the wind field, indicating category three strength.  That would make this a major hurricane.  The forecast minimum pressure is also much lower than the other two models at about 940mb.  If this were to happen, then south Florida would really need to watch out.

So, you can see that we still have a long way to go in our hurricane models.  Granted, these are forecasting for five days out, and even in non-hurricane situations our weather models do poorly at that range.  These only represent possible futures for this storm.  If my experience in watching these models is any indication, the end result will be something that avoids the extremes.  In fact, usually these storms turn out weaker than forecast.  But not always...  So, just be aware of this storm as it develops and moves eastward.