Weather forecasting has a dimensional problem.
Let me explain.
When it comes to our relationship with weather, the majority of our days are spent in one location. We move around to different places, taking trips, exploring, and getting things from one point to another. But on the average day, we are going to be in a single geographic area.
This has led us to look at weather as a one-dimensional construct. What is the weather in X city? What will the weather be like today, and specifically where I’m at right now? This feels natural to us, and the vast majority of weather services, apps, etc. are built around this model.
We search the weather based on the city name, because the service assumes we will be at or near some type of settlement. And for the most part, this makes sense. We are not continuous nomads, only “occasional nomads”.
That said, when we are playing the role of a nomad, we can’t look at the weather as a one-dimensional construct. As we are covering large distances over the course of a day, a week, a month, weather becomes significantly more dynamic. And for these traveling activities, understanding the weather and being able to predict it two-dimensionally is critical.
Let’s consider the variables needed to turn our one-dimension mindset into two, and what it takes to create a dynamic, geospatial weather plan.
The Weather Forecast Gap
Understanding and predicting weather has improved greatly over the years, becoming exponentially better as more and more data collection points are installed via weather satellites, terrestrial-based posts, and even hyperlocal sensors.
This has expanded what we know about the weather at any given point, and allows us to look beyond the weather stations set up in several places across a city. Better weather prediction has been aided by more data, and also by better algorithms.
Artificial intelligence, our ability to handle vast amounts of data at once, and our better understanding of the interconnected weather variables have all made a difference. It’s possible now for anyone to access historical weather data via API to better understand a specific area’s weather patterns.
The weather forecast gap occurs when we look at a route, and the gap grows the more and more remote this route is. And with our world growing more and more interconnected, we are creating more use cases for tracking weather not at a location, but across an entire route.
Transporting goods is a major use case, whether it be by land, sea, or air. But there are more personal use cases as well. A group going on a backpacking adventure may cover impressive distances each day, encountering diverse weather systems as they hike. Sailing routes are especially important, and optimizing sailing routes based on weather is a challenge we continue to pursue. This is especially important as foul weather is not just a source of potential danger for sailing, but the amount of wind being produced is crucial for the entire experience.
The key issue has become this: our technology improvements have created more and more data that expands well beyond cities. We are able to track and predict weather across routes in a way never experienced before. However, our mindset is still location-based. Apps are based on cities, and only specialized software focuses on tracking an entire route. It is true, on a basic app one could list the various cities the route will cover, being able to see the weather in each simultaneously.
But what about the space between, and what happens if the route is more remote?
Thinking Dynamically About the Weather
Thankfully, the solution to this issue isn’t about the lack of data, which is the most expensive and difficult part of the problem. We have the data. With platforms like Tomorrow.io, we can pull historical data via API to better understand the weather dynamics across an entire route. We can also use the same platform to feed (again, via API) real-time forecasts and updates.
If I need to look at the weather across a given route, I need to expand the dimensions and variables alike. If my goal is to, for example, go on a week-long backpacking trek, I would want to first pull the historical weather data for the route to see what could be expected. This, along with the current forecast of the route, will help determine what route should be taken in order to experience the best and safest weather.
However, this is only part of the solution. With real-time API connections, an ideal analysis would update the weather forecast for this route in real time, creating alerts when the forecast detects weather events we didn’t expect. Heavy rain, thunderstorms, temperatures spikes and dips, and extreme weather events are all elements we would want to know about immediately.
The last step in this solution would be quickly calculating alternative routes based on our preferences. In this case, our analysis would take all alternate routes (likely decided ahead of time), overlay the current weather forecast of each compared to our weather preference, and suggest the best route to take.
This is a much more dynamic view of route-based weather analysis, and is akin to what we are already used to with GPS navigation: We plan a route, the software alerts us if traffic or some other issue causes the forecasted elements (time to destination) to move out of expectations, it reviews forecasted traffic based on each alternative route and suggests the best one to take. While this is a very common process for us, the same process for weather-based routing is not.
What’s Next?
The next steps are fairly obvious. There is a gap, and with it, an opportunity for us to create better weather-based optimization tools for the many use cases for which we need them. We’ve done without up until this point, but there are massive benefits if we create the proper tools.
Thanks to massive infrastructure improvements to our weather monitoring systems, we have much more detailed geospatial information, elements such as API’s to make the data readily available, and analysis tools to help us understand, identify anomalies, and make recommendations.
