#011 – Drone Remote Sensing for Crop Insurance with Cassidy Rankine
[00:00:00] [Introduction] Welcome to commercial drones FM, the podcast that explores the commercial drone industry. The people who power it and the concepts that drive it. I’m your host Ian Smith.
[00:00:15] IAN SMITH: Hey everybody welcome to Commercial Drones.Fm, tonight we’ve got Cassidy Rankine joining us all the way from Alberta Canada. Cassidy is a Ph.D. in Earth and Atmospheric Sciences from the University of Alberta.
[00:00:31] He’s the CTO of Skymatics and his work pioneered the use of wireless sensor networks and UAV technologies. For monitoring spatially and temporally explicit patterns of land atmosphere, carbon and water dynamics in forests revealing lots of micro meteorological drivers of changing ecosystem productivity. We’ll get back to more of that in a second but first of all thanks so much for joining us and welcome to the show Cassidy.
[00:01:01] CASSIDY RANKINE: Thanks Ian and thanks for having me on.
[00:01:03] IAN: Absolutely yeah and so can you tell us a little bit. First off so what is Skymatics?
[00:01:08] CASSIDY: Right. Yes, so Skymatics is an aerial imagery service provider and we primarily use UAV or Unmanned Aerial Vehicles and we do more than just collect high resolution photos and videos. We also do a little bit of geomatics work with mapping and modeling and a lot of analysis for different industries including agriculture, construction, mining and general inspections for all the industrial companies out here in western Canada.
[00:01:40] IAN: Cool. That’s awesome, so you mentioned earlier and when we chatted that you have a sister company called Bermuda Aerial Media? How did that come about? Where does Bermuda fit into this equation?
[00:01:53] CASSIDY: Yeah good question. We get that a lot because we’re based out of Calgary, Alberta here for Skymatics but, Bermuda Aeriel Media actually started obviously in Bermuda and it was Connor Burns And E.J Burrows that started it.
[00:02:07] Back in 2011. They were the very first commercial drone service provider in the country and they did a lot of media promotion work, for videos, for all real estate and tourism industry and from there they kind of realized that as the technology of drones became increasingly more consumer friendly. You’d have to transition a bit more away from just general video and photos and wanted to concentrate on more of the technical products that UAVs can provide.
[00:02:37] So E.J. actually married a girl from Canada and packed up and moved to Calgary Alberta. And there’s a lot of oil and gas and agriculture in general. You know industrial activity in the area and founded Skymatics limited here in Calgary in 2014.
[00:02:53] IAN: Okay that’s cool. So you have a research background in satellite ground-truthing for tropical forest productivity and regeneration. So how did that really? I mean you know that’s kind of a mouthful for most people, for me a little bit. I’m kind of reading it off here of my computer screen but, maybe you can give us an idea of like you know how that led you to drones. And did you use drones in that field of study or research as well?
[00:03:21] CASSIDY” Yeah absolutely. I’ll try to keep the technical jargon down it’s years and years of remote sensory research for me trying to trying to summarize this. It’s is quite difficult to summarize because it’s a unique research program that I came out of at the University of Alberta.
[00:03:38] IAN: Ok.
[00:03:38] CASSIDY: It’s from the Center for Earth Observation sciences and what they do is a lot of carbon modeling and monitoring in Latin America, to account for deforestation, reforestation and the effects of climate change. And so we had to do a lot of carbon accounting and we used satellite, remote sensing systems and airborne Remote Sensing Systems to do all that research on large scale you know country or regional scale analyses and ended up becoming more of a calibration validation laboratory for Remote Sensing because, of the challenges with satellite based imaging and sensing.
[00:04:15] You need a lot of validation work on the ground, so my research was basically go in on the ground into the jungles and do ground base spectrometers and spectra-radiometry and actually measure light levels and micrometeorological variables on the ground to validate what we saw from space and from airborne platforms.
[00:04:35] It was kind of a natural progression into drones because we were building towers on a regular basis. We built over 30 different towers to get above the canopy of the forests that might be 40,50,60 meters tall. To actually measure incoming and reflected light continuously to validate satellite measurements. And I mean it’s quite difficult to measure things on a vegetation canopy that’s that tall and we get a limited field of view from the towers that we were building. We might be able to see an acre or or Hector at a given time.
[00:05:13] So we started the bring new tools for for getting that aerial perspective or what’s called low altitude remote sensing or near-surface remote sensing and back in late 2011 we bought our first Phantom just because it was you know a cheap technology to test out and then realize the potential of using drones for all this ground validation work for satellite imagery. At first just use it for imagery and mapping out our sites and then realized we could start to put more sophisticated sensors on them and realize just how powerful these drones became.
[00:05:47] IAN: So you guys were doing photogrammetry work with the Phantom and with the other drones in your research?
[00:05:53] CASSIDY: Well originally we weren’t, we were simply just doing you know still photographs from a perspective that would have our entire research site in one photograph and then quickly transitioned to photogrammetry and doing Point Cloud modeling that came built up in that research group led by Dr. Arturo Sanchez over the last four years.
[00:06:14] Now they work with several research groups around the world. Different Space Agency technicians from European Space Agency and DLR the German space agency, JAXA the Japanese Space Agency and NASA of course, to explore a whole bunch of different sensors strapped on the drones for this calibration validation work.
[00:06:35] IAN: That’s very cool and you mentioned the satellite imagery. So were you guys kind of combining both? I mean it is remote sensing, so whether it’s by drone or whether it’s by satellite, manned aircraft, what have you. You know it’s still remote sensing but what were some of the really the key differences that you saw? I mean we always see the drone technology being compared to satellites. So in your experience, what are the strengths and weaknesses? I guess is what I’m trying to say of drone versus satellite and maybe you could kind of tailor that to what you were doing specifically with the drones.
[00:07:13] CASSIDY: Yeah good question. So satellites have a lot of challenges in terms of sensor accuracy due to all the dense atmosphere that they have to peer through. A big challenge of why we were doing so much calibration work in the tropics is the moisture in the air and the aerosols from the heat and humidity, and generally the accuracy of the carbon modeling and forest regrowth and vegetation growth is lower accuracy in the tropics even though that’s where most of the biomass on the planet is.
[00:07:46] So we were using ground based sensor networks – wireless sensors to monitor light conditions as it moves through the canopy and hits the forest floor to figure out how much carbon was being pulled out of the air by, the amount of light that was absorbed and the real benefit. I guess the challenge of using ground sensing is you’re just taking point samples and you’re really limited spatially and how much area you’re sampling.
[00:08:12] So for a satellite that might have a single pixel of 200 meters or 100 meters all the way down to 50 meters, there’s a lot of information contained in a single pixel. So our work was to kind of tease apart all the reflectance information in a large area represented by a single pixel from satellite and having the ability to image at very fine resolution scales with drones. Let’s say you know – one, two, three or five centimeter pixels. We can actually tease apart all the information within that satellite pixel, to figure out what it actually tells us from the spectral signature we’re getting from space.
[00:08:52] IAN: Okay great. So you mentioned reflectance that’s a term that we use fairly often in the drone industry and it’s kind of you know related specifically to vegetation indices or vegetation indexes similar to well and PVI of course that everybody seems to know a lot about or likes to think that they know a lot about is a vegetation index.
[00:09:17] So can you maybe give just an explanation, I mean of how you kind of use vegetation indices today and I know you guys have an announcement that we’ll touch on that kind of relates to it. But yeah. What do you think about the whole reflectance? I mean that’s that’s probably where we should start. What is reflectance and why is it important in remote sensing?
[00:09:41] CASSIDY: Yeah you really get down to the fundamentals of remote sensing there. In my experience having worked with a lot of low altitude remote sensing systems or even you know extremely near-surface remote sensing putting sensors within five meters of a plant canopy. It’s really challenging figuring out the effects of geometry between the illumination and the target and the sensor receiving the light signal.
[00:10:07] So what we’re doing with UAV is mostly if you’re if you’re not using LIDAR, and you’re just using you know simple RGB photos or multi-spectral. It’s all passive remote sensing, so passive remote sensing just relies on the energy coming off the surface as it’s reflected from the sunlight. And in that in that signature of reflectance there’s artifacts of viewing geometry from the sun angle and how that interacts with the structure of the canopy.
[00:10:34] So there’s actually a lot of active research right now around the world trying to figure out illumination conditions and canopy structure and how that affects the strength of reflectance.Basically a spectral signature is unique not only by the material you know the vegetation and the chlorophyll and all the pigments of a plant but also the architecture of the canopy so if there’s woody material or if there’s dead vegetation in there it all mixes together in that reflected signal that we’re seeing from the UAVs.
[00:11:07] And I don’t know how many times I’ve I’ve wanted to just tear my hair out. Hearing what people say about it and NDVI from UAVs and how they’re using it to map productivity, The real key there is that there’s no there’s no standards and measuring NDVI. NDVI is just one small tool in the entire vegetation index toolset.
[00:11:33] I mean I’m happy to talk about some of those challenges going on right now and using drones for agriculture. I’ve come on board Skymatic specifically to help with all the agriculture products that we’re developing.
[00:11:45] IAN: Cool, yeah so you guys have this announcement, you guys are launching and please correct me if I’m wrong. You’re launching your Ag imagery data collection campaign and so you guys are going to use that for crop damage mapping for insurance industry. I guess specifically agricultural insurance and development of your automated image analysis software called, Skyclaim. So can you tell us a little bit about that?
[00:12:11] We’d spoken about that previously in person over here in San Francisco but tell us about Skyclaim.
[00:12:18] CASSIDY: Yeah you nailed it there. Cat’s out of the bag, hey. No it’s a good time to talk about it. We actually just in the past 10 days secured a nice government research grant for commercializing this software we’re trying to develop specifically for crop damage mapping and it’s called Skyclaim and a lot of people don’t realize the issue with the crop insurance industry is the fact that it’s government subsidized and is essentially a tax payers dollars that go in.
[00:12:51] IAN: Is it just government subsidized in Canada? Or is it also in the U.S.?
[00:12:57] CASSIDY: No it’s in the U.S. and Canada. And you know North American general has the highest rate of crop insurance and that grows every year.
[00:13:05] IAN: OK.
[00:13:05] CASSIDY: It’s a necessary measure that you need basically to protect against natural disasters for you know food security and farmer income stability. The issue with the whole system though is this, it’s not great in the sense that it’s an evolving insurance industry. It’s semi privatized. It’s a little bit different in the U.S. and Canada but in general your tax dollars go to pay a farmer when he submitted a claim when his crop gets destroyed by a storm, by hail, by wind or drought or what have you.
[00:13:36] The crop insurance – -the insurance providers don’t actually have the ability to be everywhere at once to do the on site assessments. We realize that drones are an awesome tool for farmers or insurance providers to use to quickly map out damages and support claims where the adjusters might not be able to get out there right away. And farmers don’t have to sit around and wait for adjusters to come you know 20,30,40 days after a storm hits to actually figure out what the damages are.
[00:14:09] IAN: That’s excellent. Congratulations on the launch there.
[00:14:12] CASSIDY: Yeah thanks. So we’ve launched a big data collection campaign. We realized that we also can’t be everywhere at once or right? and we want to try to tap into some crowdsourcing. So it was it was great being able to come and meet with you guys down in San Francisco a DroneDeploy there, and talk a little bit about how we want to tap into your network.
[00:14:33] I mean we’ve been out pretty pretty regularly over the last couple months up in Alberta where we’ve had lots of hailstorms and lots of super storms coming through. But what we’re trying to do is build a database that we can train Shien learning algorithms to automatically identify crop damages and specifically cator the output of what we’re mapping to the insurance industry to be able to evaluate losses very quickly on the on the matter of hours or days instead of weeks to months.
[00:15:04] Tapping into drone deploy or similar crowd based UAV data collection system is great. Now there are a lot of challenges with that and getting we can get into some of that let’s do it.
[00:15:17] IAN: Yeah kick us off man you got it.
[00:15:20] CASSIDY: So I mean I’ve come from you know 10 year research career on extensive data collection, you know being in the field. I’ve traveled all over the world worked in, four different continents, 12 different countries in the forest collecting data. And when it’s just you and a small team going to collecting data, it’s nice because you actually have control over the quality of what you’re collecting the standards there is method protocols and you know what you’re getting.
[00:15:51] Now there’s been a transition towards this citizen science crowd based data collection for research and it’s so amazingly powerful tool this crowdsource science or crowdsource anything. The challenge there though is standardization and there there’s a there’s a little bit of an issue right now in UAV data collection because there’s differences and platforms for hardware, there’s differences in cameras, there’s differences in the way that people fly them and having some way to standardize this to actually analyze that is is quite tricky.
[00:16:26] So we’re excited to potentially work with DroneDeploy, because you guys have this auto pilot application where you can fly a mission and re-fly a mission and have the same flight path and photo locations and meta data coming into it. I think we are looking forward to potentially work with you guys.
[00:16:45] IAN: Yeah, that’d be awesome. Any any way that you guys can accomplish your mission, would just be really cool for the industry as a whole. I don’t know if it is you guys who I was talking to about you know potentially, like in the future I mean we’re talking like you know quite a long time from now. But maybe it’s possible for you know a farmer or a grower to own their own drone and then actually just by having that drone and then mapping their own field after a weather event or something like that could be a way that they could maybe even get lower premiums themselves.
[00:17:23] The reason for the savings, they wouldn’t need to be a crop adjuster or you know a claims adjuster, to go out and visit the site physically and that would save a ton of money for the company.
[00:17:38] CASSIDY: Yeah absolutely. I mean we we’ve talked to farmers we’ve talked to insurance providers we’re kind of figuring out where the pain points are. We know that there’s a there’s a disconnect between what the farmer wants, what the insurance wants to do and just the lag time in getting an adjuster out. So we know you know based on projections of consumer drones, getting out there into the wild if you will that most farmers, the estimates say 4-5 farmers by 2020 or 2022. Will likely have their own little quadcopter or something similar to a quadcopter that they can easily launch out whenever they need to to get some insight on their whole field in one go. So we’re looking forward to you know having.
[00:18:24] IAN: Do you think that’s gonna happen?
[00:18:25] CASSIDY: I think that’s going to happen. I think especially with you know the FAAs most recent changes to their policies down in the U.S, Canada still kind of strict on who gets to fly these things. But regulations are getting better to allow more and more users to do this, and the cost of these drones are coming down everyday. It’s amazing what you can get for $600, $700 now, what used to cost $7,000 a couple of years ago.
[00:18:54] IAN: I totally agree with you. Speaking of Canada, what are your current thoughts on transport Canada? I mean, I would hope that there are some people listening to the show from Canada or maybe just people in the U.S. who are curious of the regulations up north. How are they? You said you said they’re pretty strict right now. But from my understanding they’ve been you know a little bit more progressive as far as you know when they started allowing commercial operations than the U.S. What’s it looking like up there in Canada these days for commercial regulations?
[00:19:26] CLASSIFY: Yes it’s a little bit cryptic right now actually there. It’s true though, I mean Canada was ahead of the game in terms of regulations compared to most countries, they were allowing commercial operations of UAV, well before the U.S. did. I don’t know if it has something to do with the fact that we just have a relatively low population density but there has been thousands and thousands of companies being granted these permits for commercial operations. Now they haven’t opened up the same way the FAA did last month. In letting anybody fly with with a simple test and you know in controlled areas.
[00:20:03] But it’s definitely getting more. There’s more knowledge being distributed from Transport Canada and the authorities on what safe to do. What’s pretty stupid to do when you’re flying a UAV and there’s, I think there’s a bit more, common sense going on than there was maybe a couple of years ago. So I’m looking forward to there being the ability for anybody to fly these things out on their field, because it’s their own property and they should be able to manage it with these tools which are which are really useful.
[00:20:32] IAN: Absolutely. I totally agree with you there. Are there any specific case studies? Maybe just one specific case study that you can talk about. Like any any problems that you guys were looking to solve on a field level basis, like you know maybe a specific field somewhere that you guys were doing some work on or maybe you could take that even another way and talk about one specific use case of the algorithms that you guys are developing for Skyclaim.
[00:21:03] CASSIDY: Yes so I mean we’re facing a major challenge in terms of trying to build a black box, that can kind of figure out what type of crop, what stage the crop is at, what kind of storm hit it? You know the intensity of the storm was there hail? The insurance policies specific to that. What kind of coverage do they have? So there’s a lot of variables coming in to be able to automate this entire process. I mean the end goal here is that a storm hits somebody goes and flies it. Be it the farmer or the insurance provider. They plug in – they upload the imagery and our system crunches everything, it spits out a pretty picture that says – here’s your damage and here’s the estimated value of the losses.
[00:21:45] We made some good progress on that. We had a wicked hailstorm come down just south of Calgary. The 29th of May and we went out the next day and the whole field was just flattened. And it was a canola crop that was pretty young stage, it might have only been 6 or 7 leaves coming off the branches, so it was maybe a foot tall but everything got flattened and we came out and map it all.
[00:22:14] Now timing is key and we’re figuring this out so we’re mapping weekly to figure out, what the exact time you have to go and basically we saw- On the first day, everything was still green right the leaves were stripped off the plants and they were just scattered all over the ground. So from the air when you’re looking at a five centimeter pixel it still looks like a healthy crop. Everything still green.
[00:22:35] We went out exactly a week later and everything turned, all the leaves turned brown that had been knocked off the plant so we could now see the signature of damage. But surprisingly because the amount of water that fell in the storm, there was a great deal of regrowth on the crop. So the insurance providers had deferred the claim, they said well we can’t pay out now we’re going to have to wait till the end of the season. So the farmer doesn’t like that because he knew that he had already had a lot of damage and the insurance provider said well it might come back and the farmers as well.
[00:23:11] If it does come back it might not be ready to harvest for you know a month or two after I normally would have. And so this is this back and forth is something we’re trying to have a more objective assessment using you Avies. So we mapped everything out and what we figured out you know based on right after the storm, versus a week after and then we just mapped it again just recently. We estimate it down to the acre level exact damages and now we’re going to wait till the end of the season and get the harvest information to figure out how that correlates to exact losses for harvest.
[00:23:48] So we can’t say for sure that we’ll be able to have this algorithm perfected for every type of crop and we’re focusing on the major crops in Alberta right now. But we’d love to expand out and get more crops that are major in the U.S. like corn. We have a lot of wheat and we have a lot of canola and barley and peas up here in Alberta. And those ones we’re getting lots of good imagery but we want to have a more universal algorithm for more crop type. So we’re hoping to get more people interested and we’ve had a lot of positive feedback so far from everyone we’ve called out to so that’s great.
[00:24:20] IAN: Cool. So how would a farmer or a grower be able to use Skyclaim? Would it would have to be offered to them through their insurer or is this something that they’ll be able to sign up for?
[00:24:33] CASSIDY: Well yeah let’s go back to you mentioning premiums and premiums are definitely increasing. I mean every year we’re seeing more intense storms and the frequency of storms increasing. This year alone we actually had more storms and rainfall. By the 15th of July in Alberta and we had seen for any July in 35 years.
[00:24:54] So there’s a little bit of worry about you know climate change and stability of food production and and farming practices having to change and premiums going up without any increase in service from insurance providers. And so we see this Skyclaim system being the middle ground where after a certain amount of time because we’ve got hard imagery and hard objective data. Insurance providers will be able to pay out an exact amount that’s been damaged, rather than estimate it to say 15% accuracy. We know we can get it down better than 15% in terms of damages. And that way their insurance, accounting and risk management – the risk basically goes down as an insurance provider. So they should based on this concrete amount of information be able to reduce their premiums for the farmers.
[00:25:45] IAN: Excellent, and so you know in that case then the farmer would kind of just like the offered Skyclaim or the insurance company actually would be very much incentivized to be able to offer their clients who are the farmers and the growers Skyclaiv then.
[00:26:02] CASSIDY: That’s right, we’re targeting the insurance providers. And the way we see it and from everyone that we’ve talked to on both sides it’s a win win. The farmers are getting a faster turnaround time. There’s no guesswork in what’s been damaged and what should be paid out. Insurance providers have concrete objective data they can use for the assessment and they know what dollar value that they owe.
[00:26:25] So in the long run the insurance providers would be taking on the imagery from the drone. It doesn’t matter who collects the drone imagery because as long as the collection is standardized then the data is accurate and everybody wins.
[00:26:39] IAN: Love it. And so if you’re a drone pilot, drone operator in Canada and the United States. Is there a way to kind of sign up to be a Skymatics or Skyclaim operator at this time or how are you guys going to be handling that? We’d love to see if we can help out through the commercial drones FM podcast.
[00:26:57] CASSIDY: Well that’s awesome yeah. So right now we’re just organically reaching out talking to farmers in person and calling them up and we’ve teamed up with a number of small hail insurance providers, Palliser insurance is one of them. The culture financial service corporation here in Alberta covers most of the province and they’re one of our research and development partners and we’re just about to launch our webpage.
[00:27:21] I wish it was ready today but it’s looking like another week or two before we launch a landing page for people to just simply submit their email and name whether they are a farm that’s interested, a UAV service provider, a precision Ag firm that deals with this stuff or an insurance provider for crop insurance. And it’s a simple sign up and we’ll keep you updated on the progress and if you want to contribute to the database. We want to incentivize that through building credit for using the software for when we release it next year.
[00:27:52] IAN: You guys have thought this out quite well. I’m very impressed. It’s really cool learning about more about what you guys are up to over there. Pivoting in just a little bit here. I love what you guys are doing. What’s your favorite thing? what do you love about the drone industry specifically? What does Cassidy love?
[00:28:13] CASSIDY: Well so I didn’t actually know if I was going to go into industry or into Academia. I’ve done a fair amount of teaching work at Universities and I love teaching. I taught a Geographic Information Systems course for environmental management, so I originally envisioned myself going into the drone technology on an academic perspective, for teaching people how it can be used for conservation and ecology, and resource management.
[00:28:40] And now I’m seeing that as things just move way faster in industry and industry research in general. So I’m pretty stoked to be on board schematics as a Technical co-founder now and pushing this stuff into actual practice. And I love the fact that not only do I get to play with cool toys like drones. but I get to do it outside.
[00:29:03] I mean I’m a naturalist. I have a biology background and I’ve done years and years of field work so just just getting out on a nice day. I mean we worked on non-rights not nice days here it’s pretty brutal in the winters we worked at 20- 30 degrees below zero operating our DJI-S1000 without too much issue.
[00:29:23] But yeah, I love the multitasking. Working a desk some days and looking at some maps that we produce and 3-D models and other days I’m out in the field.
[00:29:34] IAN: If you want to follow Skymatics on Twitter @skymatics and you can check out their website at skymatics.com. You can probably by the time this podcast is aired. Sign up to be a operator or a pilot for Skymatics and get involved in Skyclaims.
[00:29:57] Super happy to have you on the show. Thanks so much again Cassidy. If you guys want to follow the show you can check us out on Twitter @dronespodcast or Facebook.com/dronespodcast. You can go ahead and subscribe and read the podcast on iTunes or Google Play.
[00:30:14] But really appreciate everyone’s time. Thank you so much for listening. We’ll be back next time with another very exciting episode of commercial drones FM.
[00:30:23] Cassidy and I bid you farewell. We’ll go ahead and we’ll speak later on. Cheers.