In my previous post, I introduced the concept of BPUs, or Bee Power Units, as I like to call them. The basic idea is to measure the heat emitted by a bee colony to calculate the number of bees inside. I also introduced the method of using infrared imaging to measure heat, which is what I’ll be diving into in this blog.
Much like how bees’ ability to see in ultraviolet allows them to see things not visible to us humans, the other side of the spectrum (literally) allows us to view things we couldn’t otherwise see through human eyes. In case my pun wasn’t obvious enough, I’m talking about infrared light.
Bio (or, B-eye-o) hacking
For those of you who wish to one day see infrared light through your own eyes, there's hope! A laboratory at the University of Science and Technology in China has discovered a way for mammals to see in infrared. By injecting nano-particles into the eyes of mice, researchers enabled the mice to see in infrared. As cool as this sounds, I’m not quite ready to inject nano-particles into my eyes. You gotta walk before you can run. Fortunately, there exists technology that gives us humans the capability to see in Infrared with absolutely zero eyeball injections required. They call this technology "infrared cameras".
In case you haven't been following our recent activity, infrared imaging is kind of our thing. Specifically, our product Verifli analyzes infrared images of beehives in order to calculate the strength of the colony within.
Seeing things in a different light
So! Now that we've got the cool Sci-Fi stuff out of the way, let’s take a peek at some images:
Can you tell the difference? Probably! Hive B shows a little more red/orange than the other, which means there's more heat coming out of it. Naturally, since all animals (and objects in general) radiate infrared energy, the more of an animal/object you pack into a space, the more heat radiates to the space's surface area.
You've experienced this phenomenon firsthand if you've ever been in an over-crowded room that felt more and more stuffy as more people filed in. If you took an Infrared image from outside that room before and after it filled up with sweaty bodies, you'd see a difference in the temperature of the walls. Same thing happens inside a beehive: more bees, more surface heat.
How about this next picture; which of the two hives do you think has more bees?
This one's a little tougher to tell. Hive C appears to be slightly warmer in the first honey super. But the heat in Hive D's brood boxes looks a little less "splotchy" than Hive C... So it's a wash then, right? Basically the same number of bees in each hive?
Wrong! There are roughly 7.5 more frames of bees in Hive C than there are in Hive D. That's a difference of over 10,000 bees. That's nearly 3 thousand extra foraging bees out pollinating your flowers.
So what's our takeaway here? Well, by simply glancing at these pictures, there's no way the human eye could possibly tell how many bees are within the hive. That's where we come in.
But you might say “Gleb! If we can't tell the difference, doesn’t that defeat the purpose of using heat as a metric for colony strength?” To that I say, “Nope.”
The method of measuring colony size via "frame counts" has been established by the UC Extension office for quite some time now. As I discussed in my previous post, this method entails peeking between two brood boxes and estimating how many frames are covered with bees. From both personal experience and from launching Verifli this February, we found that taking frame counts is not only tedious and arduous, it tends to be extremely unreliable.
On top of that, time of day has a significant effect on the cluster size in a colony. If you take a frame count when it’s early morning and still a little chilly, the bees will be clustered tightly over fewer frames to retain heat. Later in the day, the cluster will loosen up and spread out across more frames as the weather gets warmer. But at that point in the day, thousands of foragers will have left the hive to collect nectar and pollen.
The physics are complex, but the algebra is simple: by measuring a hive's heat signature with infrared, we can calculate the number of bees inside. Notice that this equation removes the potential for human error.
Frankly, after 2 or 3 hundred frame counts, they all start to look the same. With Verifli, there's no ballpark estimates or fatigue after grading hundreds of hives in a day.
Hey readers! It’s been quite a while since my last update. Hope you’ve all had a great start to your 2019! For the growers out there, I hope your pollination season hasn’t been bogged too much by all the rain. We’ve been quite busy here at The Bee Corp with the launch of our digital hive grading product, Verifli. Speaking of hive grading, let’s talk frame counts.
How we measure colony size
For those of you not familiar, frame count is the metric used by growers and beekeepers to measure the population size of a hive. The concept is simple enough: if seven of the frames inside a hive are covered by bees, it’s a seven-frame hive. The tricky thing about this method is accuracy. A single deep frame supposedly holds about 1,500 bees. But what if the frame is only partially covered by bees? What if they’re only covering one side of the frame? What if it’s a medium or shallow frame?
Then there are other factors like weather and timing. What if it’s peak flight time and 1/3 of the bees are out foraging? What if it’s a little chilly and the bees are clustered tightly? What if I've done hundreds of frame counts today and I can't really distinguish between a 6-frame and a 7-frame hive anymore?
As a data scientist, all these variables troubled me as Wyatt and I collected more than 500 frame counts last summer.
A better measure... using SCIENCE
I hereby propose a new way to measure frame strength. Over are the days of ripping open hives, slaving over frames overflowing with bees, with the succor of smoke to yield the path! Now is the time for innovative methods to claim their rightful seat to the throne!
Aside from sounding like the ramblings of a madman, hear me out. Remember, quite a while ago, my post about how hot a bee is? That, as it turns out, is around 30-38 degrees Celsius. That got me thinking: we know there are roughly 1,500 bees on an average deep frame. We know the general heat output of an individual bee. Why not calculate colony size in terms of energy output?
How to measure bee energy
I propose a new unit of measurement for frame strength: the Bee Power Unit (BPU). Since we know how many bees we expect in an average healthy hive, and we know the average energy output of a bee, we’re able to calculate the average energy output. However, we’re not interested in some theoretical average, we want concrete details. How many bees are in that hive?
But how does one measure energy output? Which technology do we even use? If you've been following us this past year, you may already know the answer. Everything in the universe emits Infrared radiation, which is the heat energy output that can be measured with an infrared camera. Simple enough? Well, there’s a little more to it then that... Next time, I will be exploring infrared and how we can use technology to outclass the human mind.
Check back shortly for my follow-up post on this!
Imagine this: it’s a beautiful Thursday afternoon. You’re sitting on the porch, enjoying a cool breeze, rocking back and forth in a hand-crafted red cherry rocking chair. Suddenly, your serenity dissipates as your phone *dings* with a notification: “You’re running low on eggs in your fridge, would you like me to order another dozen?” This is the world we live in today.
The basics of IoT
The concept of the Internet of Things has been floating around since as early as 1982. Carnegie Mellon University pioneered a modified Coke machine, capable of reporting its inventory and the temperature of its drinks. At the time, this was thought of as groundbreaking technology, but now it’s everywhere. Your smart refrigerator tells you when to pick up groceries, your Fitbit alerts you when your heartrate reaches dangerous levels, and your driver assist warns you when there’s a car in your blind spot. All these items are part of a concept known as the Internet of Things (IoT for short).
At a basic level, IoT is a network that delivers information to decision-makers as soon as an event occurs. This information may be used to inform us when an issue exists (like when a traffic light is broken), collect data to help us understand processes (like how much of a certain input you’ve used), connect humans, monitor areas, you name it. Currently, there are over 8 billion connected devices on the planet, and this number is continuing to rise.
IoT or IoBee?
IoT is a major buzz word these days, and the limitless applications can be exciting, but likewise, such fast-paced advancement in technology can be overwhelming. You may have heard about how Amazon is working on IoT wristbands to track employees. Is this a scary “Big Brother” tactic designed to punish slackers, or a wise business strategy aimed at maximizing efficiency?
Despite the apparent privacy risks, IoT creates possibilities that can help to make our lives easier and make businesses more efficient. Industries that deal with gigantic stocks of inventory use IoT to find where things are stored, how much is there and how long it’s been in storage. Agriculture producers use IoT to monitor things like irrigation pumps and soil nutrients. This raises the question: how can beekeepers benefit from adopting IoT?
Think about what information helps you manage your bees. Wouldn’t it be useful to know when a nectar flow or a dearth has just started? How about if you got a notification when a honey-bound colony is about to swarm? What if you could track what kind of honey was being produced based on the nectar sources the bees were visiting? There are countless IoT applications that could help beekeepers better manage their hives.
Beekeepers today face many problems whose solutions may be just around the corner, in the form of IoT. In years past, issues like short battery life and shoddy communication networks (like 4G LTE and Bluetooth) made IoT applications too costly and unreliable for certain industries, but those issues are rapidly being solved.
We’re at a point where some of our crazy ideas—like, “what if my queens could tell me how many eggs they’re laying each day”—might actually be possible. This is an exciting time to start thinking about how we could use IoT to solve some of the problems beekeepers face on a daily basis.
A local Denver community marketplace near me recently hosted an introductory beekeeping class given by Colin Mann from Vine Street Farms, which provides mentoring and beekeeping consulting service in my area.
As I had been toying with the idea of becoming a backyard beekeeper, my family & I decided to make a night of it, and went to enjoy some pizza, beverage, and bees. I loved the idea of helping to provide for the needs of our honeybee neighbors and helping their population grow and thrive. And of course, having our own local honey to harvest and share with friends would be just a bonus.
It's a juggling act
I had done my research, and much of what was discussed, I felt I had a good handle on. After all, I’d read The Beekeeper’s Bible. What I hadn’t counted on was how much work was involved in keeping bees. Work AND kids AND dogs AND house AND bees...having one more job to do just wasn’t in the books for our family.
Which got me thinking...I was just worried about the work 1-2 hives would take. How much time and labor is involved in taking care of 100 hives… or 1000 or more?
What to consider
Well, according to Jamie Ellis’s article in the American Bee Journal, the answer depends on several factors: what beekeeping goals you have, what your available resources are, your local climate, how many folks are helping manage the hives, AND, most importantly, how many colonies you have.
So many jobs to do: colony size needs to be managed, hive splits need to be made, and more supers need to be added. Pests & diseases must be prevented or treated. Is the queen in good shape? Do the bees need to be fed? Have they made enough honey to last the winter?
It takes as much as 2700-3900 hours of work a year to manage 1000 hives.
Most commercial beekeeping operations’ main source of revenue nowadays is in the pollination business. With this comes an even greater amount of labor, as well as drive time (which I sure as heck would consider labor!).
So, hat’s off to the American beekeeper! They are some hardworking folks, doing their part in keeping a large part of the nation’s pollinators thriving, and trying to earn a living while doing so.
Disclaimer: No bees were hacked in the making of this event.
The Bee Corp recently teamed up with Indiana University's School of Public and Environmental Affairs (SPEA) and their Data Science course to host our first-ever Bee Data Hackathon. If you've never heard the term before, a hackathon is an event where the sponsor provides participants with a problem and the resources to solve it.
The challenge: How can beekeepers maximize honey revenue?
The resources: A giant, messy file of bee data including historical honey prices, average honey yields in different regions, costs associated with transporting bees, and much more information (some useful, some intentionally misleading).
Students were given just a few days to clean and analyze data to recommend the best locations to maximize honey revenue throughout the year. Students were scored based on how well they followed instructions, the logic behind their math, how well they cleaned the data, and how the data was displayed in a data visualization. Undergraduate and graduate students were split up to compete for jars of our honey and a feature on this blog.
The students were given raw, uncleaned data with additional errors added by The Bee Corp Data Science team. The first step was to clean the data.
Hive weight was used as a way to track honey flows, working off the assumption that a hive that weighs more is going to be filled with more honey. When moving bees to a new location, students were given labor costs of loading bees and the transportation cost per mile to determine if the move was worth it for the revenue gained from honey collection. Different premiums for different types of honey were not considered for this hack.
Surprisingly, each team had different recommendations on where to move bees based on what they found in the data! Some teams recommended certain months that were best to move bees, while others split the data into quarters. Both a graduate student and undergraduate student team were awarded with honey for having the best hacks, and our company selected the undergraduate team as the overall winner for the blog.
The team, consisting of Yin Zhan, Wanyu Wang, and Hongda Wang recommended starting the hives in Washington for the first quarter of the year, moving to Alabama for the second quarter, Texas for the third, and then moving back to Washington for the fourth quarter of the year.
Their math was highly detailed, which won them a lot of points from the judging. They calculated the difference in honey collected in different areas and subtracted the cost of moving the hives to get the net benefit of moving the hives for honey flows. Check out their visualizations above.