A word about long range forecasts

There's been a lot of bad press around here lately regarding the reliability and accuracy of EC's long range forecasts.. the latest being a less than flattering article in today's Winnipeg Free Press. This public criticism is understandable, given that EC long range forecasts lately have been giving some rather optimisticly warm forecasts in Winnipeg where there's still considerable snow on the ground. Surely EC meteorologists in Winnipeg know this and would issue more conservative forecasts to account for this, right? Well, they do, but they don't. And this is where some explanation is necessary..

Canada is a big country. Lots of territory to cover, not very many people. Like most federal departments, Environment Canada has been under tremendous pressure over the past decade or so to cut costs, and work more efficiently with fewer resources and fewer staff. The net result is that Canada now has only 7 weather offices in the country to produce every single weather forecast and warning for the entire country. The local Winnipeg weather office for example is responsible for the forecasts from southern Saskatchewan to Baffin Island. That's a huge territory for one office! Compare this to the United States, where they have over 100 weather offices for a smaller geographical area. (There are 13 offices covering Texas alone!) Obviously given such a large area to cover with such limited staff, EC has had to rely more and more on automation wherever possible, from downloading weather information on phone lines and weatheradio, to actual production of the forecasts.

As a result, weather forecasts in Canada are now automatically produced from a central supercomputer at the Canadian Meteorological Center (CMC) in Montreal. They come in two parts.. a short term forecast (today and tomorrow) and a long range forecast (Day 3,4,5) Twice a day, these forecasts are sent to the 7 regional weather offices where regional EC meteorologists start with these computer generated forecasts as a first guess. Generally, these forecasts are quite good most of the time..(most people just don't realize how good weather models have become lately.) However, regional meteorologists will adjust these computer-generated forecasts if and when necessary, based on their skill and experience. The best return on this investment of human intervention in the forecast process is in the first 24 to 48 hours, and increasingly, EC meteorologsists have been concentrating their efforts on improving the short-term forecast (i.e. today and tomorrow's weather) which has a more immediate impact on the general public. This means that the long range forecast (Day 3 and beyond) is now basically left untouched from the computer generated version sent from Montreal. In other words, local Winnipeg meteorologists are becoming less and less responsible for the content of the forecast beyond tomorrow! Now generally, these automated long range forecasts are no worse than what a human meteorologist would produce (in fact, quite often, it's better) However there are times when these forecasts can be quite poor, especially in the transition seasons of spring and fall when more variable weather and snowcover can have a tremendous impact on long range temperature forecasts (as recent forecasts have demonstrated) But regional EC meteorologists only have so much time and resources.. and increasingly, their efforts are being invested in the short-range forecast, which sometimes comes at the expense of a less accurate long range forecast. So the next time you hear that it's supposed to be 17C on the weekend, and there's still 30 cm of snow on the ground, don't blame the local Winnipeg meteorologist.. he's too busy looking at today's weather from Swift Current to Iqaluit!

More recently, the long range forecasts from EC haved been showing a warm bias due to a change in the model last year. Winnipeg meteorologists have been active in pointing out this problem to CMC headquarters in Montreal where the weather forecasts are generated from. Eventually, the model statistics will catch on to this bias and correct it, so that future long range forecasts will be more reliable temperature wise. This will be especially evident when we lose our snow cover. In the future, long range forecasts will be based on an even better forecast approach called ensemble forecasting which will be more reliable, and less prone to wild forecast swings. Still, as long as the long range forecast is totally automated, there will be occasions where the forecast will not be as accurate as the public would like. But that's the chaos of weather.. people can't expect Day 1 accuracy out to Day 5.

So there you have it.. that's why these long range forecasts are sometimes questionable (and it's not only EC.. check out Accuweather or TWN long range forecasts, they're not much better!) It's unfair to criticize local Winnipeg meteorologists for these questionable long range forecasts of late. Increasingly, it's not their responsibility, the short term forecast is. The bottom line is that you get what you pay for. If the public wants more accurate long range forecasts, then they should let their MPs know that a better funded weather service is a priority for them. In the meantime, it's always a good idea to check other sources for long range weather information (including Rob's blog!) for additional insight into future weather.