We have only just started to understand the potential impact of variable weather conditions on eVTOLs and drones, says Don Berchoff, CEO, TruWeather Solutions Inc.
Where do you sit in the drone/advanced air mobility ecosystem?
Our aviation weather system is still stuck in the 1990s; even though our modelling has improved, our core data has not. Yes, we have better satellite data and that’s why we’re able to do long-range forecasting better, but we haven’t appreciably changed our data collection below 5,000 feet. We have not advanced aviation weather measurements beyond airfield weather systems, two balloon launches a day, and weather data collected by aircraft that pass through 5,000 feet at landing. Between airports, generally, the actual weather and wind is a mystery.
The layer below 5000 feet is the turbulent layer that we have trouble measuring and understanding. And that is where advanced air mobility and drones will be flying. The other part of the problem is the uncertainty. We still haven’t figured out how to use uncertain weather data in a way that can maximise its value.
Advanced Air Mobility (AAM) and drones are the best business case for micro weather we’ve ever had. I view this as a beachhead in solving a problem that won’t just help AAM and small drones but can help with monitoring air quality, the transportation of pollution and other environmental needs. My focus is on solving the problem and finding a way to monetise the service which means going to where the action is and demonstrate the value of it to the end user.
Weather is uncertain. It’s going to take a diversity of sensors, and smart deployment of those sensors to reduce the uncertainty and it’s going to take a business case that justifies the infrastructure investment.
TruWeather has built the software that can take diverse low level weather data and present into simple format which can be tied into UTM systems. We are focused on how to bring more investment through various business models to pay for the infrastructure – a state could pay for it and make the data available; the municipality could buy it and let a private company run the systems as some examples.
What is the scale of the challenge, say between Florida, where it’s warm, versus New York, in terms of the number of sensors, complexity of the problem and the variables that these environments face?
What are seemingly ”good” weather conditions for crewed aircraft may not be “good” weather conditions for drones. We always break view through the lens of use cases. In Florida, the problem is pop-up thunderstorms in the summer. We don’t have good tools right now to understand where thunderstorms will form in the next hour or two. A thunderstorm pop-up could mean your flight from the vertiport is cancelled and you can’t get to the airport on time. The thunderstorm problem is for us the biggest challenge in Florida because it’s the one that we have the least amount of data on from the boundary layer below 5,000 feet. I believe it’s solvable, but we need a lot of data to solve it and there’s still a lot of science. The other problem you’re going to have in Florida is the wind, a sea breeze front, location, for example.
The problem in the northeast is fog, wind, precipitation and icing in clouds. We need to be in the room during the vertiport design phase; we need to look at every vertiport location. Not every vertiport should have the same weather instruments. Instrumentation should be based on data performance standards to meet each vertiports need.
How do we get to where we need to be for accurate micro-weather forecasting for drones and AAM?
With small drones we’re at 30% of where we need during marginal weather conditions because we have no measurements between airfields and there’s a lot of local climatic effects. On a beyond visual line of sight (BVLOS) flight of 10 miles you can go through three patches of fog without knowing it and the wind can be surprising above tree level. Quite often drone operators in North Carolina, for example, have to cancel flights up to 40% of the time due to weather, and approximately 30% of those cancellations are because they don’t know with certainty what the conditions are where they will fly. The economic impact to them if they don’t fly is huge.
With middle-mile drones, we’re probably at 50%, maybe 60%, when weather conditions are marginal because they’re fly a little higher and once you get above 2,000 feet you start to get into more homogeneous weather and wind flow. Our models do a little bit better there. The biggest challenge they’re going to have is icing, because if you’re going to 5,000 feet in wintertime, you’re going to be grounded half the time just for the potential icing in northern tier locations. The other reason is if they have to fly VFR. We might be able to get above the clouds, but they’re going to have to punch through at times and they have to be allowed to do that.
With eVTOLs I would say we are at 30% in urban areas during marginal weather conditions, because of the winds around urban canyons. We have no way of knowing how those buildings are influencing wind and if you use anemometers readings at the airport, they are not going to be relevant. At an airport, on the other hand, we are closer to 60%.
How can we make sure the people who are running the ecosystem along with drone and eVTOL operators get to see the same picture?
I think we can say there will be a minimal viable common weather picture that is based on traditional government aviation weather sources. But because it is likely the governments will not have the resources to fund all weather infrastructure needed to capture all the micro-climates, there will be a dependency on third party weather providers to fill the gaps, and so I imagine a future where some operators will require a better weather picture than the minimum viable picture. Unfortunately, weather is always going to have some uncertainty about it, and the uncertainty in any weather picture will depend on how many sensors we have and how the sensor density improves the weather picture. Keep in mind, the weather pictures shared today all have a level of uncertainty, because they are based on a model of our best estimate of what we think is happening based on very limited weather measurements below 5,000 Feet. My concern is the industry is seeking full automation of weather systems. But weather will never be a 100% accurate, even with more sensors. On some days, there will still be a requirement for a human meteorologist if you hope to achieve a higher number of flights than automated weather alone will allow.
I believe 80-90% of time automated weather will be adequate for an operator’s business model with acceptable risk.
How much would it cost to get the kind of system that you’ve just described to support a route between Miami Centre and Miami International or New York and LaGuardia?
The question should be “what does it cost if you don’t have this?” If an eVTOL operator wants to fly one aircraft eight times a day from LaGuardia to a Manhattan vertiport, and the assumption is that they would generate USD14 million a year, the question I would ask is does that include “weather tax?” I have come up with a concept I call “the weather tax” – the price mother nature imposes upon your business, for both real or perceived weather. And in New York City it’s about 30%. So, under this scenario, you would likely generate closer to USD 9.8 million We are targeting, with investment in weather infrastructure, recovering USD1.2 million of the lost revenue per aircraft per year. What would a business owner pay to get that type of return on investment? This is why I prefer we focus on the requirement and the value, rather than the cost.
The return on investment I’m striving for is four to one. So, if you’re looking at Miami Centre to Miami International, it’s about 20 miles, assuming that they’re going to fly within a 20-mile corridor area with room to move around a thunderstorm, I would lay down an array of wind lidars, micro-weather stations, with appropriate processing and data distribution capabilities that may cost USD20 million over five years, or USD4 million a year. If you start flying 100 aircraft in that 20-mile area, for just those 100 aircraft, the cost is USD40,000 per aircraft a year to get that predictability in the winds and weather to the microscale level. The key though is to amortize the cost across multiple operators flying in the area. With scale, the cost per aircraft drops, and if you use the data for other purposes like air quality monitoring, achieving a four to one return on investment for each operators become feasible,
I believe that this is no-brainer in high density areas.