Prairie Weather (Rural Roots Canada) – Applied local weather data reduces costs and increases efficiency.
While attending a recent conference, I was asked by an agronomist about the methods used to develop weather forecasts and the role that local, on-farm weather stations and sensors play in forecasting.
As the networks of private weather stations become more dense and more reliable, their roles increase. All forecasts start with analyses of current conditions, and a well-sited, reliable, calibrated weather station provides the kind of information essential to forecasting. Regardless of whether the local stations are tied into data networks used by forecasters, their information is both seen and valuable.
Weather tech on the Prairies has come a long way from just watching the sky and hoping for the best.
While it’s always important to stay updated on the forecast for the nearest town or city, many prairie farmers now pull in data from dense weather networks plus their own on-farm stations. In Alberta, for example, the ACIS provincial network ties together hundreds of stations to give hourly updates on temperature, rainfall, wind, and soil moisture that can be checked online..
More farms are putting in their own automated stations and simple sensor networks that capture what is actually happening on their fields, not wherever the nearest town happens to be. One of the more widely used sensor networks is Tempest, which has a dense network of weather stations across the agricultural regions of Alberta, Saskatchewan, and Manitoba.
Some of the more elaborate sensing systems can provide dense data that can be used for better decision-making. Farmers can respond to moisture and yield potential acre-by-acre instead of treating a whole quarter the same.
AI is quietly emerging as a forecasting tool. AI weather models are much different from the traditional forecast models that have been used for decades. Traditional models take current weather conditions and run them through several algorithms that carry out trillions of calculations to produce estimates of future weather. It can take several hours after current conditions are inputted before forecast model results appear. AI models are 40,000 times faster, and rather than use multimillion-dollar supercomputers, some can even be run on a standard desktop. AI models use weather data from the past in order to predict the future. They will find periods of time when weather conditions are nearly identical to the present day. The AI model will then analyze what happened on those past dates and use that data to forecast the future. This takes a matter of minutes, not hours.
AI models trained on years of weather, soil, and yield data can estimate how a crop is likely to perform under different conditions and management choices, helping growers move away from “we’ve always done it this way.”
AI models can be used for targeted spraying and weed control. Others aim at better yield predictions and risk assessments so farmers can make marketing and insurance decisions with a bit more confidence.
A mix of hyper-local monitoring, shared knowledge, precision equipment, and AI-driven insight is helping Prairie farmers stay on top of the weather.
Still, there is no substitute for the human meteorologist with the professional experience to analyze and interpret this vast amount of weather data to create forecasts that AI still can’t match.
Environment Canada is there for daily forecasts. For weekly look-aheads, we’re here for you at Rural Roots Canada with human-created weekly outlooks on Monday mornings and updated Thursday middays.
