Forecasting Sandy

I've been following the progress of Hurricane Sandy pretty closely.  This is a rare scenario - a late-season hurricane, with high pressure over the Atlantic, and an arctic cold front pushing into the Eastern Seaboard.  The result?  A storm that will have tremendous impacts over a wide area.  While there won't be the massive, wholesale destruction of a Category 4 or Category 5 hurricane, the effects will be spread much farther afield.  Tropical storm force winds will likely affect the states of North Carolina, Virginia, Washington DC, Maryland, Delaware, New Jersey, Pennsylvania, New York, Connecticut, and Massachusetts, and in many places, those strong, sustained winds of at least 35 mph will persist for 12 to 24 hours.  On top of that, several inches of rain will fall over a wide swath of territory, saturating trees' roots and making them more susceptible to topple in the winds.  Power outages will be common and widespread, and it will likely take days to bring it back online everywhere.  But much of this is common knowledge by now.

I want to give you a look at the process of forecasting an event like this.  First, you have to know what's currently going on...forecasting starts with observations. In the US, forming the core of the observational network are the automated weather stations that are located at airports and other significant sites across the country.  These stations measure temperature, dewpoint, winds, clouds, and precipitation, among other things, and they are reported every hour.  There are other, usually local or regional, networks of automated weather stations as well.  It's important to understand what's happening not just at the surface, but also through the depth of the atmosphere.  Weather balloons are launched twice daily (morning and evening) from National Weather Service offices and other sites across the country.  In preparation for Sandy this weekend, over 200 extra balloons were launched in the afternoon to gain additional data about the state of the atmosphere...a first!  Of course, it's also important to have observations of precipitation, and for this there is a network of cooperative observers who report 24-hour rainfall every morning.  Much like for surface observations, there are numerous local, regional, and national rain gauge networks to measure precipitation at time steps from every 5 minutes to every 24 hours.

In addition to the measurements that are taken in situ (directly in the environment), there are two major means of remote sensing used by meteorologists.  Radar detects precipitation and estimates rainfall totals.  The national radar network is currently undergoing an upgrade that will improve rainfall estimates, improve detection of precipitation type (snow, sleet, hail, graupel, rain, etc.), and have other benefits to forecasters and researchers.  Radar can also estimate the direction of the wind field...this is what allows forecasters to find "Doppler-indicated" tornadoes.  In addition to radar, satellites provide information about clouds, precipitation, moisture, and even surface temperature.  For a storm like Sandy that has spent much of its life out to sea, satellite data is invaluable.

Once observations have been collected, they're fed into computer models that use the fundamental equations of the atmosphere to forecast future conditions.  There are several models, and each model is run several times with small changes to the observations and with slightly different ways of estimating atmospheric processes to provide insight into the different ways that the atmosphere might evolve.  The more similar the different model runs are, the more confident forecasters can be in what is likely to happen.  Early in the process of forecasting Sandy, the models disagreed on where she would go.  Most turned her westward and brought her onshore, but they disagreed as to where - anywhere from Virginia to Nova Scotia.  A couple of models took her out to sea and kept her away from the US mainland.  The following few figures show a series of predictions of the models' projected path for Sandy from late last week.  You can see that as time went on, the models tended to agree on a US landfall, and they began to narrow the location down.  Sorry for the blurry first figure!

Of course, forecasting is not left entirely up to the models.  Human forecasters look at the model output and the current observations.  Using their own knowledge of the models' tendencies, the trends from the current observations, and their knowledge of past storms, the forecasters make predictions about the most likely path for Sandy and the weather that she will bring.  The National Hurricane Center is charged with forecasting hurricanes, and they produce graphics like the ones below.  The dots indicate the most likely position of the storm's center, and the hatched area around it shows uncertainty in this path...the storm center is expected to stay within this cone.  For a gigantic storm like Sandy, though, the storm's effects are felt well outside the cone.  Take a look at the series of track forecasts below to see how the forecast has evolved over the last few days.

Of course, forecasting the path of the storm's center is only a small part of gauging the storm's
impact.  The Hydrometeological Prediction Center makes forecasts of rainfall totals across the nation every day, and a storm like Sandy keeps the forecasters especially busy.  Below is a series of maps showing HPC's long-term (5-day) forecasts for total rainfall.  You can see how the rainfall forecast changed as predictions of Sandy's strength and path changed.  Here in central New York, we're looking at 1-3" of rain over the next couple of days, though more than that is possible in local areas, especially on higher ground.

Here's tonight's rainfall forecast for the storm, issued by the local Weather Service office, located in Binghamton, NY.  These offices are staffed 24/7, and they'll certainly have extra help around for the duration of the storm.  The local offices are in charge of issuing watches and warnings (such as the Flood Watch and High Wind Watch that we're currently under) for their forecast areas.  They also coordinate with local emergency managers, storm spotters, cooperative observers, broadcasters, and the public to stay informed of current conditions and their impacts, and to alert the public to hazards.  Needless to say, many forecast offices across a wide area will be very busy over the next couple of days.  Here in Tompkins County, Ithaca is forecast to receive 1.42", and right here in Freeville, we're expected to get a tad less than that.  The forecast has gotten better over the last couple of days.  For my dock's sake, let's hope it stays that way.