Monday, August 27, 2012

Anyone can be this right

Apparently it is really easy to forecast (e.g. weather, airplanes, objects, finances, emergencies, crises, etc). So easy any one can do it.


There are few things to get straight. There is alot of data out there and that includes forecasts. What can be encapsulated in a forecast is sometimes data and sometimes information. It can be expressed explicitly (explained) and implicitly (inferred; typically in the form of specialized vocabulary).

Another aspect is the relationship between space and time, and no I am not talking about Einstein. This relationship is very specific to the phenomena you are trying to forecast. For thunderstorms, we are usually talking small. For organized thunderstorms we are getting bigger, perhaps to the size of an average county. For larger thunderstorm complexes we are talking the size of Iowa-ish. Now each of those phenomena have different time scales (i.e. how long they live, and also importantly how long they "breath"* - compare to humans: we live 80 yrs and yet have a heartbeat that occurs every half-second). The implicit part is the short time scales inferred when one uses certain terms like thunderstorm (15-30 minutes) or supercell (30minutes to 6 hours) or MCS (up to 6-24 hours). These thunderstorms can drift with the wind, deviate from the wind (supercells and MCSs) conditional upon their strength and type (not all MCSs are the same).

Hurricanes in contrast are composed of thunderstorms yet can have lifetimes of weeks, and attain a size of a much larger state than Iowa, assuming they stay organized. Because of that longer lifetime and their larger sizes they have similarities to MCSs. They are simultaneously at the mercy of their surroundings and dependent on their inner structure (i.e. the processes that act to maintain them) for where and how they move.

A 5 day forecast for hurricanes, tropical storms or depressions is rather difficult. You are projecting both the location and the intensity. They are not separate or distinct elements. A shallow storm (such as a depression or under-developed or weak tropical storm) will move with the low level winds while a deeper system will move with the winds through a much deeper layer (sometimes slower, sometimes altering their own local environment!).

So when some of these terms are used, meteorologists are implicitly referring to these core characteristics, utilizing this foundation of knowledge, to apply to current data and models. We use words like uncertainty because we know how inherently difficult it is to make predictions at long forecast times which have little skill in the aggregate, but under very specific circumstances may have much better skill than normal.

The purpose of this information is to keep you updated in advance, so that there is awareness of an impact to you and yours. That is, forecasts are issued for their information content and their perspective. Knowing is half the battle. And they are issued because there is an expression of need because some organizations or activities take many days to be prepared. A 5 day hurricane forecast may not be accurate (i.e. specific or reliable), but if you know the typical errors you can make an educated guess on closeness (both space and time). Thus you can infer value, or if you like make plans that can capitalize on that information in small steps. It is all about maximizing your odds given the uncertainty.


In the end it may seem like anyone can do it given the errors or inaccuracies. But this is the state of the science. It has been improving quite nicely, but that doesn't mean there are no flies in our forecast ointment. It just means we have been improving in the aggregate (over all storms). True skill comes from being accurate over most (or at its best: all) storms. But this trend can only improve when we learn more about how these thunderstorms work and how best to simulate them and how best to communicate that forecast information which is most valuable at the right time to the right people in the right places.