In similar fashion, weather forecasters look for patterns on weather maps in order to predict the tendencies of potentially offensive weather. Such pattern recognition can play an instrumental role in short-term forecasts (one to two days ahead), medium-range forecasts (three to ten days) or long-range outlooks (months to seasons).

For an example of pattern recognition, suppose it's mid-September, the heart of Atlantic hurricane season. If upper-level winds resembled the pattern shown in Figure 15.1, forecasters at the Tropical Prediction Center in Coral Gables, FL would watch for the possibility of a tropical system running a northward "fly pattern" up the East Coast. Indeed, with a lethargic trough of low pressure drifting over the Middle West and the Bermuda high lined up in its typical formation, high-altitude steering winds (which, like football pass patterns, are marked by arrows in Figure 15.1) would tend to direct ny tropical system over the Caribbean or the tropical Atlantic right up the Eastern Seaboard.

Another example of pattern recognition is used by forecasters at the Storm Prediction Center (SPC) in Norman, OK to predict large outbreaks of tornadoes from late winter to early summer (see Figure 15.2). When SPC forecasters see such a pattern developing on a given day, they don't hesitate to issue tornado watches within the hatched area in Figure 15.2.

When composing seasonal outlooks, meteorologists look at (among other factors) "teleconnections" from either El Nino or La Nina, if one of these is present or imminent in the tropical eastern Pacific. In essence, a teleconnection means that an atmospheric or oceanic anomaly in one part of the world is correlated to an anomaly somewhere else. Both El Nino and La Nina involve significant departures from average ocean temperatures. These anomalies incite anomalous exchanges of heat, moisture, and momentum between the ocean and the atmosphere that can have far-reaching effects on seasonal weather patterns in other parts of the globe.

As an example of a seasonal outlook based on teleconnections, Color Plate 15.A shows the probability of the occurrence of temperature extremes during meteorological winter (December through February) across the United States during a La Nina. For example, suppose that the typical probability of a cooler-than-average winter in northeastern Minnesota is 20% (based on climatological records). During La Nina, the probability of a cooler-than-average winter goes up to 2.0 x 20% = 40%. If La Nina conditions are expected, the winter outlook for temperatures in this region should be composed with this tendency in mind. A cooler-than-average winter is also favored in the Pacific Northwest and the rest of the northern Plains during La Nina. Not much of a temperature signal from La Nina is observed in the Northeast and Ohio Valley, so forecasters in these areas cannot statistically skew their temperature forecast toward a warm or cold winter based solely on La Nina.

There are many other pattern recognitions that meteorologists use to make short-term, medium-range, and long-range forecasts. We will further discuss the details of pattern recognition later. First, we need to lay some groundwork. Just as some coaching staffs rely on computers to quantify the tendencies of opposing teams, forecasters rely on computer simulations of the atmosphere for guidance in molding their forecasting strategy.