Weather forecasting is not a profession for those who need regular praise. When was the last time you called up the weather service to compliment them on an accurate forecast? Predicting the actions of Earth's atmosphere is a challenging task. Yet, when it comes to weather forecasts, offshore sailors expect meteorologists to get it right. On average, forecasters can predict weather three to five days in advance fairly accurately. However, as any voyager knows, there are times when weather forecasts aren't much good after only a few hours. What can meteorologists do to improve forecasting? One intriguing approach uses a form of chaos theory to analyze atmospheric "hot spots". Changes in these areas can indicate much larger changes in the weather. Should the research on hot spots continue to prove fruitful, the result could be improved forecasts for mariners (more on these chaotic hot spots below).
With powerful supercomputers, sophisticated software models of global circulation patterns and the vast amount of data collected by ground stations, weather balloons and satellites, you'd think that weather forecasts would be highly accurate and reliable for long periods. But forecast weather and actual weather tend to diverge, sometimes rather quickly. "The forecast gets worse and worse with time," Eugenia Kalnay, a professor of meteorology at the University of Maryland and one of the members of the team performing research on the chaos hot spot approach, said. "There's a theoretical limit of predictability of about two weeks," D.J. Patil, a graduate student at the University of Maryland who participated in the chaos hot spots research, said when referring to weather forecast accuracy. In practice, of course, forecasts are often much less accurate than that. One of the reasons often given for the unpredictability of weather is that the atmosphere is a chaotic system. In other words, small errors in weather observations grow over time. Eventually these errors become so magnified as to make the forecast unreliable. A well-known illustration of chaos theory is a hypothetical butterfly flapping its wings on the equator. The flap puts air in motion that, through a series of chaotic magnifications, results in formation of a hurricane in the subtropics.
With this chaotic magnification of events, perhaps a bit exaggerated in the butterfly example, it's no wonder that weather forecasting can be a challenge. To make things tougher, there are other factors over and above the chaos of the atmosphere (isn't that enough?) that make weather prognostication ticklish. One is the dearth of reporting stations. In this age of sophisticated weather technology, we might like to think that the entire globe bristles with weather instruments, the truth is that there is no local weather gathering capability in many parts of the globe. There are no weather instruments on the ground, and no weather balloons laden with instruments are launched to gather a daily weather profile by altitude. Only in developed countries are there networks of automated weather stations and regular balloon launches. "The U.S. is pretty well covered," Patil said. "But go to Africa, for example, and there's very little [weather station coverage]."
The situation gets worse for mariners. While land coverage is spotty, there's an area of the planet with even worse weather station coverage: the oceans. Meteorologists receive helpful weather data gathered and sent in by merchant ships on ocean passages; but ships tend to stay in narrow shipping lanes. They also tend to avoid storms and other unsettled weather – the very conditions meteorologists are most interested in studying. One valuable area of coverage in the Pacific Ocean is provided by the Tropical/Ocean Array. This is an array of approximately 70 moored buoys, each equipped with a range of weather and ocean-sensing instruments. The buoys, put in place to study the ocean and atmosphere during El Niño/La Niña southern oscillation cycles, provides on-the-surface weather data that are important for marine forecasters. The vast majority of the ocean surface area, however, is not covered by ships or buoys.
To a certain extent, some of this missing coverage can be made up using satellites. While they are not a substitute for ground stations, satellites can provide information on clouds, temperature and winds over the oceans. NASA's polar orbiting Quikscat satellite, for example, uses a radar scatterometer to gauge wind speed. Scatterometers read the wind-induced ripples and waves on the sea surface. From these surface features, computers can calculate wind speed and direction. Weather imaging satellites also gather pictures of weather in visible and infrared light.
Interestingly enough, another challenge facing weather scientists becomes larger as the number of observations from around the globe increases. The issue here is calibrating the observations from a wide variety of instruments and times. For example, weather observations taken from instruments in North America at one time and India at another time have to be assimilated into one coherent data set. Processing all the raw weather data is a huge number crunching problem. This assimilation process can take reportedly as much as three times longer than running the weather prediction model itself.
One technique forecasters have devised to deal with the unpredictability of the atmosphere is something called the ensemble forecast. In this method, a variety of different computer forecasts runs with slightly different starting conditions. Meteorologists then compare the outcomes and how much they agree. Where there is disagreement, the meteorologists assign a confidence level to the various forecasts based on a variety of factors, including their own judgment and experience. What they get is a weighted blend of forecasts that is issued as the ensemble forecast for the next five days.
With the chaos hot spot theory the researchers took a similar approach to the ensemble forecast. The team, composed of physicists, mathematicians, meteorologists and computer scientists, combined a series of forecasts and studied those areas on the map where the forecasts were diverging the most. "We used the ensemble forecast to measure the uncertainty by looking at differences in the members of the ensemble," Prof. Kalnay, who was one of the inventors of the ensemble forecast concept when she was at the National Center for Environmental Prediction in Camp Springs, Md., said.
The team discovered a technique to classify the complexity of the areas where the forecasts diverge rapidly. In these regions small changes in conditions are magnified quickly into large changes in the weather – the butterfly effect. The surprising result of the research is that in some of these regions the complexity, or level of chaos, is not as large as scientists have traditionally believed. These areas of separation are regions of low dimensionality according to the researchers. "The low dimensionality areas are where the forecasts are breaking down, but not into a complete mess or chaos," Patil said.
Since changes in the hot spot areas will have a large effect on the coming weather, researchers believe that making additional measurements in these key areas of low dimensionality will improve forecast. Some ways to make additional observations might include flying a sensing aircraft into an area to take measurements or pointing a weather satellite at hot spot region for several hours.
While this chaos hot spot theory on weather prediction accuracy won't change marine forecasting in the immediate future, it is an example of the research focus on improving forecasting. For voyagers, the hope is that this and other areas of study will result in better long term forecasting, easing the lot of the ocean sailor.