A Gardener’s Guide to Weather Prediction

It’s about time, placeand data.

by Beverley Wagar, Halton Master Gardener

(This article was originally published on April 8, 2022, on the blog of the Halton Region Master Gardeners)

The end of March taught us, yet again, that climate change has turned the words “normal weather” into a quaint reminder of the good-ol-daysbefore CO2 levels began their steep rise, before we started wanting a second opinion on how soon to put the tomato seedlings outside. But now, with the weather weirdness getting weirder, do the “normal” temperature patterns of the past have any relevance to the present?

Yes—sort of. With weather extremes growing more extreme and occurring more often, the act of averaging the highs and lows does not convey a true picture of the chaos. But averages are still the best tool we’ve got for understanding how fast our planet is warming. How long they remain helpful for gardeners is another question.

Garden under snow ( CCO )

“Climate normals provide a baseline to compare yesterday’s weather and tomorrow’s forecast to a standard for each location and time of year,” says Mike Palecki, project manager for the US National Oceanic and Atmospheric Administration (NOAA). Weather averages (“normals”) are available on various time scales: yearly, monthly, daily and even hourly. They are useful to energy companies, transportation schedulers, vacationers, and anyone who plans their activities in coming weeks or months based on what the weather is likely to do. Of course that includes farmers and gardeners.

We gardeners need to have a good understanding of the concept of temperature averages because that’s the way our springtime planning and planting protocols continue to be done. Understanding the meaning of “average last frost” allows us to interpret and understand the increasingly conflicting advice available to gardeners throughout the internet-connected world.

“Last frost” calculations start at the weather station. Environment and Climate Change Canada (ECCC) collects data from 1,735 weather reporting stations throughout the country. Some of these stations have been recording weather data for more than a century. Hamilton began reporting daily weather information in 1866; Toronto started in 1895. The Hamilton / Halton area has ten stations for which normals are calculated:

  • Brantford
  • Georgetown
  • Burlington TS (near QEW and 403)
  • Hagersville
  • Hamilton Airport (Mount Hope)
  • Hamilton RBG (Southwest Burlington)
  • Middleport (near Caledonia)
  • Millgrove
  • Oakville (Joshua’s Creek Trail)
  • Vineland (near Grimsby)

Using Hamilton RBG data for our example, the average last-frost date (calculated by ECCC) is April 21.

graphic of April calendar with 21st highlighted

Average last frost

But what exactly does “last-frost date” mean? (The phrase is more accurately “last 0-degree day” but more on that later.) To really understand the concepts of average and probability, we must tiptoe into the field of Mathematics.

Imagine a big spreadsheet that lists the lowest temperatue on every day between March and June, for the period between 1959 (when the RBG began recording weather data) and 2010. Now, highlight all the days it went below zero. Now, within these below-zero dates, for each year, find the one that occurs latest in the spring. Some will be in April and some in May. Now find the arithmetic mean (the “average”) of these last-frost dates. Et voila! April 21!

So, is this a safe date for a gardener to plant out? That depends on the amount of risk we’re willing to take. Remember that the concept of “average” means half of the last-frost days were before April 21—and half were after. Statistically speaking, it’s not quite correct to say there’s a 50-50 chance of frost on April 21, but you get the drift. And it may be a snow drift, so the wise gardener will wait for the risk to decrease.

How quickly does the risk increase or decrease before and after April 21? Analysts at the ECCC have calculated risk levels by plotting last-frost dates on a graph and determining the probability of frost on a given date. Looking again at the data from the RBG weather station, here are the probabilities that the last temperature in spring of 0°C or lower happens on or after indicated dates.

May 15: 10%
May 8: 25%
May 3: 33%
April 27: 50%
April 22: 66%
April 20: 75%
April 14: 90%

A couple of things jump out. First, why did they stop at 10%? On what date is there zero probability of freezing? We’ll look into this later.

And second: why is there a six-day discrepancy between the 50% probability date (April 27) and the average last frost date (April 21)? According to ECCC Senior Climatologist David Phillips, it’s due to how the statistics are calculated. It could also be related to the dataset used; it’s not clear whether the probability numbers were derived from normals (30 years) or all the data available from the RBG (51 years.)

When is there zero risk of zero?

So when is there absolutely no risk of freezing in the spring? On what date has the temperature never gone below zero in Hamilton? The answer is: May 31. This date was obtained not from ECCC but from WeatherSpark, a software company that makes interactive weather maps. Their calculation of “growing season” uses data from the Hamilton International Airport from 1995 through 2021. Yes it’s apples-to-oranges, but from a gardener’s perspective, it’ll do.

screenshot from WeatherSpark, graph of growing season showing May 31 as 100% frost free date in Hamilton Ontario


Unfortunately, math is not the ultimate crystal ball and air temperature is not the only predictor of whether your seedlings will survive outdoors overnight in mid-May. Remember that your garden is a unique microclimate. Cold air is heavier than warm air—it tends to sink into valleys. So frosts usually come first in valleys while hillsides can remain frost-free.

As well, your property’s microclimate is influenced by the amount and type of hardscape you have. Pavement, stone, and buildings collect heat during the day and slowly release it at night, warming the surrounding area. Swimming pools and ponds also create a buffering effect. For an in-depth look at microlimates, visit https://pss.uvm.edu/ppp/articles/frosts.html

Using a thermometer that measures temperature changes over time (sometimes called a temperature data logger) you can compare the actual overnight lows in your garden to the predicted lows. Of course you may not want to go full nerd on this stuff; simply observing your garden’s microclimate and general variance, either colder or warmer, from the forecast will help you know whether to schlep your seedlings indoors or to simply throw a sheet over them. (To learn more about frost damage, check out OMAFRA’s excellent page on the effects of extreme temperatures on tomato and pepper plants.)

Frost vs. freeze

Just because the thermometer stays above zero does not mean our plants are safe. Understanding the difference between “frost” and “freeze” helps us understand why the TV weatherperson gives stern frost warnings even though the temperature is not expected to dip below zero.

Frost damage on tomato plants. Image (C) Gary Pilarchik. All rights reserved

Frost is the white stuff on the ground. Freeze is the air temperature dropping below freezing. Sometimes we get frost when the temperatures are above freezing and we often have a freeze without frost. It all has to do with the dew point—the temperature at which water vapour in the air condenses on the ground and other surfaces (our plants!) as dew. If the dew point point is below zero, the water vapour condenses as ice, freezing as frost. The more moisture in the air (humidity) the higher the dew point. This explains why frost will happen on a dry, crisp night even if the temperature remains above freezing. For a more detailed explanation, here is an article by meteorologist Jeff Haby helping his peers educate the public about dew point: https://www.theweatherprediction.com/habyhints/190/

Shorter short-term forecasts

It’s evident that predicting the arrival of damaging cold can be tricky. So we might want to fall back on that old stalwart: the short term forecast. Surely we can rely on the Weather Channel, AccuWeather, TV Networks, or even the crowd-sourced Weather Underground to give us accurate predictions.

Actually, we can’t. By now we’re all aware that climate change is making things more unpredictable: hotter hots, colder colds, bigger storms, longer droughts, faster changes. But now, scientists have actually measured the extent of this unpredictability.

A November 2021 Stanford University study shows “rising temperatures may intensify the unpredictability of weather in Earth’s midlatitudes.” Lead author atmospheric scientist Aditi Sheshadri states: “Errors propagate through weather models faster as temperatures rise, and there don’t appear to be any temperature thresholds where the trend shifts.”

But let’s not backspace on all the algorithms. According to Alannah Campbell, a meteorologist with Instant Weather Ontario, “We can still predict weather on a short term basis, but weather and climate models can struggle with extreme events, making forecasting difficult.” She adds, “There are some models that do better with extremes than others and model developers are constantly trying to upgrade.” The NOAA has upgraded it global weather model, but Canada has not.

Time for the Ouija board?

Not yet! Science, although its power to predict weather is diminishing, is still the best tool we’ve got. Temperature predictions useful to gardeners, whether based on the data of historical norms or the skill of meteorologists, come with caveats based on our understanding of not only climate change and microclimates but also basic Math and Geography. We can make informed decisions that result in a tomato harvest. It might not be the earliest in the neighbourhood but, year over year, it will likely be the most reliable.

Canada is getting set to release new data on climate normals
Every decade Environment and Climate Change Canada (ECCC) re-calculates normals for the previous 30 years. Current (April 2022) normals are derived from the 30-year period between 1981 and 2010. A quick bit of mental math tells us we’re due for an update. The USA completed its update (see graphic below) in May of 2021 but ours is still in progress (as of April 8, 2022.)

David Phillips, Senior Climatologist at the ECCC, reassures us that the new normals data will be released in the summer of 2022, for the period from 1991 to 2021. Phillips is “anxious to see how the updated set of data, which will include the warm decade 2011 to 2020, will change the normals.”

As gardeners we can expect the average last-frost dates and the probability charts to shift. As well, there will likely be changes to Canada’s Plant Hardiness Zone Maps. And, of course, we’ll need to adjust our expectations for the 2023 gardening season.

map of USA showing warmer average temperature zones
Annual average temperature change from the old 1981-2010 normals to the new 1991-2020 normals. With the exception of the north-central U.S., the entire country trended warmer” (from https://www.weather.govh/ict/newclimatenormals )