All About Is My Weather Weird?™
A Weather App with a Twist
Is My Weather Weird?™ is a weather app with a twist — it offers a simple answer to a common question we’ve all asked. To do this we look at how often weather like today’s used to happen in the past, using the best data available.
Fun Fact: every day of weird weather calculations involves over 10 million data points
Define “Weather”
This might seem obvious, but there are two important points to understand.
First, weather refers to conditions right now, or on a particular day, at a given location. Examples of weather include the amount of rain in Tulsa today, or the current temperature in Seattle. This differs from climate, which is the average weather at a given location, such as the typical number of days of rain in Tulsa in May, or the average high temperature in Seattle for the second week of January.
Second, Is My Weather Weird?™ (or IMWW for short) measures daily weirdness for three specific aspects of weather: high temperature, low temperature, and total precipitation. There are other weather measurements we’d love to use, such as humidity or wind speed, but a large majority of historical weather measurements tracked only those three items, so that’s what we have to work with.
Weather is what’s happening outside any given moment or day; climate is average weather for a given location.
Define “Weird”
Basically, weather is weird if you wouldn’t expect it to happen at a given location and time of year. A high temperature of 90 F in January would be bizarre in a place like Chicago, but not too weird for somewhere like Miami.
IMWW uses a weirdness score that corresponds to how often a given type of weather used to happen, as described below.
Define “How things used to be”
To say weather is weird, we need to compare it against normal. For this purpose, we use the 20th century as our definition of normal, more specifically the weather before 1990. This is a fairly common date to use for before & after comparisons of weather and climate, because this about when changing temperatures became impossible to ignore. Older folks will remember pre-1990 as the normal that they grew up with, and the timeframe is further relevant because our built world, such as homes and infrastructure, was largely designed for 20th century conditions.
IMWW compares today’s weather with weather patterns from before 1990.
The Data
Historical Weather
IMWW uses historical weather records from the Global Historical Climatology Network (GHCN) the most comprehensive database of daily weather data available, with measurements from over 100,000 weather stations in 180 countries going back as far as the 1700s.
IMWW uses daily high temperature, low temperature, and precipitation depth (rain and snow water equivalent) from approximately 6,000 of the GHCN weather stations that have particularly high-quality measurements, with an average of 47 years of pre-1990 data per station.
Today’s Weather
IMWW current conditions and forecasts are provided by Aeris Weather, a professional-grade weather service trusted by industries including aviation, agriculture, and logistics.
IMWW uses the best data available for historic measurements and today’s weather.
The Method
For more detail see the accompanying technical description (TBD).
1. Calculate Historical Weather Patterns for each Station
First we determine what weather looked like prior to 1990 for each day of the year at all of the 6,000 weather stations. We do this by lumping together all of the data, across all years, for a particular day of the year. In fact, we also lump in data from a few days before and after. This provides more data to work with for generating statistics, and it makes intuitive sense because typical weather doesn’t change much in very short timeframes. For instance, on average the weather March 15 isn’t much different than on March 16, or March 13. Seasonal changes in weather become more important on the timescale of weeks or longer.
This approach gives us around 600 days of historical measurements per day of the year, representing what the weather was like on (or near) that day in the past. Then, for each day, we analyze the distribution of measurements, which essentially means counting the number of times each weather type was a particular value — how many times the high temperature was 55, 56, 57, … 90, 91… etc — to determine what kind of weather was common and what was rare.
2. Compare Today’s Weather with Historical Patterns at Stations
We now have a description of historic weather patterns for each day of the year at each weather station. Next we grab today’s weather for any of those station locations, compare it to the historic pattern, and determine how weird the weather is at that location. This allows us to conclude things like, “A high temperature of 73 F on April 15 only happened 5% of the time at this station prior to 1990.”
It’s again slightly more complex than this because we actually compare the current weather with a statistical fit to the historic data, using distribution forms that are suited for describing extreme values, i.e., the “tails” of the data.
3. Calculate Weather Weirdness for All Locations
If you lived very close to one of these 6,000 weather stations we’d basically be done. But since most of us don’t live within a few miles of these locations, there is another step to take to determine weather weirdness for all locations and not just at these particular stations
To do this we use a simple idea: if the weather is weird at point A, and it’s weird at point B, then it’s probably also weird in between, as long as A and B are not too far apart. To put this to practice, we first calculate weather weirdness today for each of our 6,000 weather stations. Then we use this information for every station to estimate weirdness for a nearly continuous map of the covered area, currently most of North America. We say nearly continuous because we can’t accurately calculate weirdness for places with no historic weather stations nearby, so we don’t return a result for those areas. This is also the reason we’re currently limited to North America, which has a particularly dense network of weather stations vs. other parts of the world.
Recap
Is My Weather Weird?™ is a weather app that provides a simple answer to a common question. We do this by comparing today’s weather with historic patterns, taking steps to ensure the best possible answers. We hope that this context about your daily experience of the weather is fun and thought provoking, and we’d love feedback at info@ismyweatherweird.com. Thanks!
FAQ
Is there an Is My Weather Weird™ app for download or purchase?
Not yet, but we’d like to do that soon. If you’re interested in this, or if you have ideas for features, let us know at app@ismyweatherweird.com
Is this an open source project?
Not currently, but we’re open to the idea — it just takes quite a bit of effort to properly open source a project, and this is a side project for now.
Does this app prove that climate change is real?
The purpose of this app is to answer a simple question about today’s weather using the best available information. I hope you’ll find it interesting and fun, and it may also be concerning. Weird weather on any particular day is not by itself evidence of climate change, which is about long-term changes in weather. The long-term trends in our weather weirdness calculations show the same conclusions as other climate research, which is that since 1990 temperatures are increasing on average and that precipitation patterns are changing. A future version of the app will provide this climate analysis, so that you can see trends in your own hometown.
Can I get Weird Weather data for my app or data analysis?
Yes, we offer the information from the app and much more via a programmatic API that can be integrated with your own weather app or data processing. Contact us at data@ismyweatherweird.com
What do the different types of weird weather mean? Cool Highs? Warm Lows?
IMWW uses the three types of weather measurements that are broadly available with a long history of measurements:
- High Temperature: The daily high can be weirdly warm or weirdly cool. Warm highs are often more noticeable because they can make our days unbearably hot (in the summer) or perhaps pleasantly warm (in the winter). Cool highs can mean jackets when we expect to be wearing short sleeves, or bitter cold winter days when we prefer not to leave the house.
- Low Temperature: The daily low temperature can be weirdly warm or weirdly cool. Most of the time the low temperatures happen overnight, so warm lows might feel like a surprisingly pleasant autumn night. Cool lows might mean unexpected frost in the warm season, or dangerously cold winter nights.
- Precipitation: Weirdness is determined by total daily precipitation. For rainfall this is simply the depth of rain, while for snow this refers to snow water equivalent, which is the depth of water that would be left if you melted the snow that fell that day. Interestingly there are very few places on the entire planet where it is weird to have no precipitation on a given day. Even where daily rainfall is very common, such as the tropics during the wet season, it is statistically not that weird to have the occasional dry day. For this reason we only measure weirdly wet conditions.
Are there other weird aspects of weather that can be calculated?
Definitely. Even if we’re limited to temperature and precipitation measurements there are many other interesting and important aspects of weather weirdness that can be calculated. Examples include heat waves (consecutive days above a given temperature), cold spells, and dry spells or droughts. We’ll roll out more of these as soon as we can.
