OpenWeather at Cold Comfort 2026: Road Weather Intelligence for Winter Maintenance Interview with Daniel Johns, Senior Strategic Consultant at OpenWeather

1. The industry has relied on regional forecasts and professional judgement for decades. What finally made that approach unsustainable?

I would say that it isn’t so much unsustainable, rather that technology is moving at such a rapid pace that now alternative or enhanced approaches exist. Augmenting these tried and tested methods with new innovations in both sensing density and Machine Learning/AI techniques is leading to a step change in how we look at road weather and the operations needed to keep travel safe and ensure that mobility is maintained at its maximum.

 

2. You talk about highway authorities "flying blind" when it comes to road surface conditions. How widespread is that data gap in reality, and does it surprise people when they hear it?

Let’s be honest this is probably the oldest question of them all in road weather! The cost of a full Road Weather Information Systems (RWIS) has always prohibited high density observation networks to be deployed meaning vast stretches of road have no “ground truth” to be certain of what is actually happening at any given moment. Modelling the weather between points is an obvious answer, but then how do you verify it is correct if there are no observations? Here at OpenWeather we blend affordable road surface temperature observations with the latest Machine Learning techniques to create a truer picture of the networks response to ever changing weather conditions.

 

3. Standard NWP models are calibrated for airports and urban centres, not rural A-roads or exposed motorway sections. What does forecast bias actually cost an authority in practical terms?

Forecast Bias is something you can only discover through observation. The feedback between what was forecast for a location and what actually occurs is vital for long term accuracy of information. If a forecast model is trained on an urban centre then it is open to having a warm bias. This means practically that it will under-forecast road frosts and as such leave the travelling public open to a higher risk of accident. However, if for some reason the bias is the other way then regular colder than reality forecasts will mean a big impact of council resources as well as a negative impact on the environment through over-salting. Clearly getting observations to ensure accuracy should be a priority!

 

4. Walk us through what happens in your system when a potential icing event is developing. What does a duty engineer actually see, and when do they see it?

At OpenWeather we can supply a fully integrated service, or what you might term a one-stop-shop, for winter maintenance operations. Through our dashboard you can quickly and intuitively look at real time observations from both our own (and any third-party sensors) on your network. You can then get high resolution forecast services trained by those observations to ensure a high level of accuracy. Then, just in case, if you need to discuss an upcoming event, we have a team of on-call meteorologists waiting to take you through the finer points of the forecast and critically discuss the inherent risks in the unfolding weather event that you should be aware of. All this leads to consistent and appropriate decision making.

 

5. The OWHL model fuses ground-truth sensor data with atmospheric forecasting. In plain terms, what does that correction process look like, and why does it matter for a specific stretch of carriageway?

Think of the process as a virtuous circle. A site-specific forecast is given for a specific location, and then we have the actual conditions monitored at that site. By placing a machine learning module in between, the system gradually learns over time about the site-specific characteristics and adjusts the forecasts accordingly. I mentioned forecast Bias earlier and this is a clear way that the forecast can be improved. However, Machine Learning adds a new dimension in that it “learns” many of the local attributes such as variations in shading as the day progresses (important to understand how much energy gets absorbed by the road surface during the short winter daylight hours). It also begins to understand how traffic impacts at differing times of the day and week. All these things add up to a much better view of what is happening that then translates itself into more accurate forecasts for that location.

 

6. Your case study with Cumberland Council shows a shift from broad assumptions to route-level decisions. What changed operationally for their duty teams once they had that level of granularity?

When the granular, high-resolution data is delivered, it supports detailed information around urban corridors and junctions, where small temperature differences can alter frost risk and treatment needs. It also captures wind and precipitation micro-variability on exposed routes where drifting snow and rapid freezing are a potential risk. This information improves confidence in treatment timing and targeting. We’ve got very positive feedback from Cumberland Council about our solution: "The platform has proven to be highly reliable and is excellent for gaining a clear, overall picture of our performance metrics. We have been particularly impressed with the team's responsiveness; they have been quick to implement specific customization tweaks that have greatly benefitted our workflow. Furthermore, features such as the diary aspect are incredibly useful and intuitive to use."

 

7. Salt usage is under scrutiny from both a cost and environmental standpoint. How much of an impact can better road weather intelligence realistically make on a council's seasonal salt bill?

This is a very delicate balancing act. On the one hand the costs to a local authority for an incorrect decision can be huge – depending on the situation even into the £millions. However, on the other hand wasting salt and the resultant operational costs have an impact both on the councils spending but also the environment though salt seepage into water courses and the effect it has on the local flora and fauna. So, you can’t take one issue in isolation. It’s all about getting the balance right and that can only be achieved through quality of information to get efficiency of resources correct with minimal impact on the environment.

 

8. There is a significant difference between what the sky is doing and what is happening on the road surface. A shaded bridge behaves very differently from an open carriageway at the same air temperature. How does your sensor network account for that?

Road meteorology is a science in its own right and it’s all about the balance between incoming radiation (heat) to the road surface through mainly sunlight and traffic and that heat leaving the road mostly overnight, the speed of which is determined predominately by cloud cover. At any given location multiple factors affect that process including road construction, nearby vegetation, local topography, proximity to water and so on. As a result, trying to model these inputs directly can quickly become very complicated. However, an observation, by its very nature, takes all these things into account hence we see the real behaviour of the road surface at that location and as such get a truer picture of the reality.

 

9. You offer 24/7 meteorological support alongside the platform. How does that human layer work alongside the automated intelligence, and when does it make the difference?

Although the industry has discussed fully automated decision making for many years the truth of it is that there is still a “human in the loop” who is ultimately responsible for all decisions made on their network. Some nights these decisions are straightforward, but given the very marginal nature of our climate, many nights are far more complicated. The weather itself is a chaotic system and likewise some forecasts are straightforward, but many have a range of possible outcomes, especially at longer lead times. It is when this happens that the ability to discuss directly with a meteorologist comes into its own. They can delve into the forecast and talk about risks of the differing, but possible, outcomes to help the decision makers understand the risks in a totally different way to an automated forecast. This is true decision support!

 

10. Cold Comfort brings together the people responsible for keeping roads open through the winter. What is the single thing you most want that audience to take away from your stand this year?

I would say that we would want people to remember that we provide a fully integrated solution utilising the latest technological advancements to create a virtuous circle of observations leading to increasingly accurate forecasting all delivered through modern dashboard interfaces with on-call meteorologist support.

 

-------END-------

 

Meet the OpenWeather team on Stand B42 and get more information LIVE at the Resilience Theatre: 

OpenWeather Integrated Solutions for Road Maintenance: How to Make the Most Accurate Forecasting
11:40AM - 12:00PM | Thursday 21st May

Speaker
Dmytro Chupryna
Head of Business Development
OpenWeather