The Unpredictable Weather

During my brief stint in the mainstream public weather forecasting, I have had “genuine” requests from friends, users and even some high-profile personalities who have been planning their activities and hence need to know if a particular day (of their interest) will be raining and at what time in a specific place.
Recently, I have had friends ask if the wet spell we had recently will continue until the end of the season?
I have attended workshops (mostly local) where people recommend that the Meteorological Office need to buy modern equipment in order “to enable them to tell what the rainfall season will be like in the coming three years”.
We have heard a number of churches, wedding planners and entertainment promoters come to the Met Office to check if they can get the weather forecast for their interested area on a given day and time.
Indeed, they are aware that harsh weather conditions can easily turn away their prospective clientele resulting in a much publicised event being a flop unless contingency measures are put in place.
While the meteorological community has diverse customers, it is crucial for the users to be aware of what is possible in terms of the prognosis/forecast lead-time, time and site-specificity of the weather forecasts in view of the “unpredictability of the atmosphere” beyond 10 days.
It was at the University of Zimbabwe, some six years back when I first fell in love with weather forecasting.
I quickly dashed to the library’s online catalogue and looked for a few forecasting books, one of which taught me that “we are all forecasters and also weather forecasters of some sought”.
The fact that we carry umbrellas sometimes without hearing the forecast or do the reverse cemented that notion that sometimes we do our “very short-range forecasts” which enable us to dash to the shops and back before the rains fall even without an umbrella.
What then differs with the professional meteorologists is that in addition to the physical observations (which we all make), they have more computing equipment, satellites which see larger parts of the world at the same time as well as complex computer models which try to represent the atmospheric system and how it will evolve with time.
This together with knowledge of atmospheric physics, geography of the place, other local systems and expert judgment give us the expected weather conditions.
Many of us have heard of Lorenz’ chaotic theorem which shows that, for the atmosphere, small differences in initial conditions will eventually result in big differences in the evolution with the classic example often (mis)quoted is that “a butterfly flapping its wings in Brazil could cause a tornado in Texas”.
The theorem talks of a very small perturbation or small difference in initial conditions (perceived state of the atmosphere at the beginning of the forecast period) can cause a much significant changes in the weather condition thousands of miles away or as the atmosphere evolve in time.
Thus people would all want to know whether it will be raining on Good Friday of 2013 as early as today so as to plan their Easter conferences, way of the cross processions, travels and other events, the atmospheric science (inclusive of research, computer-generated models and equations which represent the atmospheric conditions) has not yet developed enough to give a reasonably accurate forecast beyond 10 days.

• Scientific Limitations

It is crucial to know that this has nothing to do with the staffing or type and age of equipment any met office has but simply a limitation in the laws of physics, limited level of scientific understanding of the atmospheric system as well as the complex systems that control our weather.
Even the leading met offices with massive super-computing power, hundreds of professors working on weather in their centres at any given time, in addition to being supported by more hundreds of renowned scientists in universities have limitations in giving the customers the weather product in the lead-time they require.
My analysis and little experience with such centres as the Japanese Met Agency, the Chinese Met Administration, the French, British, German and American’s National Weather Services is that their customers seem to have better appreciation of the capability of their met offices in view of the scientific limitation I have highlighted earlier on.
Despite its high destructive potential in the form of strong winds, tornadoes, lightning and heavy rains which usually trigger serious flooding and downstream challenges, scientists with their complex computer generated models, massive computing might have only a few days up to a week to know where a tropical cyclone may form. They have even fewer days to predict the possible trajectory with reasonable accuracy.
That basically means the preparedness for a tropical cyclone strike has to be less than a week. What makes tropical cyclones even more complex to monitor and predict is that there is even much limited knowledge of its structure and behaviour as well.
Detailed research on these over the years has been hampered by the mere fact that there are few observation stations, ships, planes and persons who can risk withstanding the tropical cyclone lashing to be able to measure its full structure and behaviour in real life.

• Predictions, predictions,

Despite the common occurrence of thunderstorms often accompanied by lighting and hailstorms, it takes only a couple of minutes to develop a fully-grown system which can be destructive and hence forecasters usually have a few hours to a day to tell if thunderstorms are to form.
In countries where weather radars are unavailable, the nearly real-time forecasting or “now casting” which is crucial in this case is almost impossible hence the intensity, movement and prediction of the storm is often missed in such instances, with the usual consequences.
Heavy rains and floods can be predicted with reasonable accuracy some three to 14 days in advance.
Given that situation, it therefore implies that the meteorologists cannot tell (this early) with confidence what the weather will be like on the day in question (Good Friday of 2013) as it is 50 days away.
In such instances the only available option is to look at climatology (to say of the past).
For example on how many April 29s  did it rain in the past 30 years? If 20 rained, it means that there is a 2/3 chance that it rains on such a date.
This is, however, also problematic in that it assumes a linear relationship between the past and the future thus assuming that the future will behave just the same way the weather behaved in the past — which can be very misleading especially in view of the fact that climate is changing and this is affecting the rainfall patterns too.
In view of these circumstances, it will be more helpful to prepare for any type of weather (rain or no rain on the day in question) and most importantly check with the Met Office for a more reliable forecast a week or less towards the day.
One of the key information often requested (almost every year) by farmers is the occurrence of dry spells within a season. While the occurrence of drought spells has a large bearing on the agricultural drought and productivity, the current level of scientific advancement only allows a drought spell to be reliably forecast at most two weeks in advance.
This is against the possibility of a meteorological drought which can be reliably forecast three months in advance. It is therefore meteorologically impossible to tell if a certain place is going to rain or not as long as it is more than 14 days.
Even the cutting edge technologies and models which attempt to give some guidance on the prospective forecast become relatively unreliable at those far-away lead-times.
As for ground frost forecast lead-time in three days to one week.  Be informed! – The Herald
• Elisha N Moyo is a senior meteorologist in the Met Services Department of Zimbabwe’s Climate Application Branch. The views expressed are those of the author.

February 2013
« Jan   Mar »