Rick Shory

Offering a little something you might not otherwise have

Leave a comment

GMO smokescreen

Yesterday I went to the panel discussion and movie “Food Evolution”, about Genetically Modified Organism (GMO) issues.

Some say “yes”. Some say “no”. I say “smokescreen”.

There real issue is overpopulation. The agricultural details, like GMOs, organic farming, and local production, are waves on the shore. Population is the rising tide.

They say: We will have however many billion people by the year twenty-whatever. So we need GMOs (or not) to feed them. So. Now, say, it’s twenty-whatever, and the people are sitting around, happily eating their GMO crops (or not). Game over? No. That vast population will continue to overpopulate to the year 2100, 2200, or however long they can get away with it.

One of the panelists talked about India, where farmers pirated the BT cotton, and now production is up a big chunk of percent, and pesticides are down a big chunk of percent. It sounds like the numbers, if anybody was keeping the numbers, when the potato first came to Ireland. What’s going to happen when the insects are all resistant to BT (as is already happening) and the weeds all shrug off Roundup (as is already happening)? Just waves on the shore, at an ever higher tide line.

I got a key point about population from the subtext of the movie, “Like Water for Chocolate”. Romance aside, the family dynamic was that the matriarch pressured her youngest daughter to stay home as caregiver. Third world people are not dumb. They don’t have a whole lot of kids just to be contrary. In the desperate equation of their world, a whole lot of kids makes it more likely one will be around for parents’ old age. In a culture with no pension, retirement, or health care, it’s the available option. Overall, there gets to be more and more people, and the equation grows ever more desperate. This is just one of the intractable Catch-22s of population.

GMOs are going to come because some, like the Rainbow papaya, are just too good. Estimating by how these things usually go, I bet there will be one or two “gotcha”s, where some dire thing like mad cow disease will emerge, due to one particular GMO crop. And this will prove to the opponents it’s Armageddon. Until the general populace forgets about it, and life goes on.

Anyone considering getting into the GMO debate, on either side, should just skip it. Work on population issues instead. GMOs are fun because there are fingers to point, and laws to pass, and lots of theater. Overpopulation is no fun because, in the words of Pogo, “We have met the enemy, and he is us.”

close up of Greenlogger, in case

1 Comment


From the early 20teens, till now, I have been working on a project I call “Greenlogger”.

It started from work I did with Dr. Heidi Steltzer at Colorado State University. She had the idea to log “greenness”, over the growing season, from a natural ecosystem such as a patch of prairie or tundra.

This is important in climate research. If you set up a number of such monitors, you can do experiments to simulate climate change. For example, in dry prairie, you can artificially water some sites and exclude rainfall from others. In tundra, you can melt the snow away early from some areas, using black ground cloth. You can hold the summer temperature a few degrees warmer by placing small tentlike structures. Also, long term records across many years will be able to show actual climate change.

tentlike structures on tundra

Structures used to warm tundra sites, for climate change experiments

You monitor “greenness” spectrally, that is by looking at what light is reflected from plants and whatever else is covering the ground. You want this to be automatic, that is by electronic instrumentation. For climate change research, you want remote undisturbed sites, typically miles out on the high plains, in alpine meadows, or north of the Arctic circle. Visiting such places is expensive. You can’t afford to pay for people to go there, day after day, for manual observation. The other advantage of instruments is that they have no bias or opinion, and they don’t get bored.

Spectral data is available from satellites, but only to a certain resolution, typically with pixels 30 meters on a side. If your experimental sites are only the size of a card table, the satellites will see nothing. Some day drones may be feasible, but they presently lack the reliability and repeatability, not to mention the flying range. There are various legal and jurisdiction issues with drones too.

So, Dr. Steltzer had got her research proposal funded to use individual ground-based monitors, one at each site. I came onboard to make this happen.

I inherited a good part of the design.

You might think the spectral way to monitor greenness would be to take pictures, and analyze them for the color green. This is fraught with all sorts of complications, such as dark shadows in the frame, changing light conditions, and the fact that plants are such different colors green. To keep consistent with decades of scientific work, such as from satellite imagery, the greenness factor we used was “Normalized Difference Vegetation Index”, abbreviated “NDVI”.

Here’s an oversimplified version, but it will explain the equipment design. Plants absorb visible light to use for photosynthesis, but they reflect infrared because it is no use to them. So, basically, the higher the infrared reflectance relative to visible, the more “greenness” down there. At large scale, you can make greenness maps of whole continents from satellite imagery. At small scale, there are instruments to clip onto individual leaves. We were working at an intermediate scale, looking at reflectance from a living ecosystem, such as meadow or tundra, in chunks about one meter size.

The design for the monitoring devices had developed from two directions. One, obviously, was to detect the infrared and visible spectral bands. The other was weatherproofing. Satellites are far above the atmosphere, and you can take the clip-on devices home. However, ground based instruments that will run unattended must handle all the vicissitudes the environment can throw at them. The evident choice, at least for off-the-shelf, was weather station parts.

To make all this happen, I had to use quite a bit of patching and overdesign. Some of the spectral detectors were photodiodes. These had amplifier circuits. Which needed batteries. Which required cases for those batteries. The batteries had to be kind of big, to assure they would run the whole time. And so the cases were big. And needed supports. Kind of big supports. By the time I came along, the design had solidified into the “mantis”.

site showing manits, with person for scale

“Mantis” greenness monitor, deployed in Wyoming

Can you see the resemblance? The frame of steel bars looks rather like a giant insect. The head is looking out, pensive and intent. The body is slung behind.

It wasn’t long before the mantis began to evolve. I had to adapt it for a project in Alaska. There would be more extensive data collection, so there would be additional sensors. These would require a more complex weather station box. That meant “bigger”. Each mantis would have its own solar panel. Even as a mock-up, the insect is metamorphosing, sprouting new appendages.

mantis mock-up, frame and instrument boxes

Mock-up of mantis, adapted for tundra project

The sensor cables had to be protected from gnawing varmints out on the tundra, so they all were sheathed in metal conduit. The Alaska design looked less like an insect than an octopus.

tundra version of mantis, top view

“Octopus” mantis

As in all science, you need lots of “replicates”. You can’t just have one experiment and one control, because any two sites will naturally be different. For the Alaska project we needed about two dozen of these mantises.

I had to finalize the design, and scale up for the total number of parts. I cleaned out three local Home Depot stores, to procure some parts! I had to figure things out, down to the last nut and bolt, and get it all shipped to Alaska. There would be no neighborhood hardware stores out on the North Slope tundra.

Things went well. At the research station, we spent some days in the lab trailers assembling all the mantises.

people assembling mantis parts

Mantis assembly

When they were ready, we took them out to the tundra and got them going.

person carrying completed mantis on his back

Mantis on the way to tundra site

In all, I worked on this project for three years, setting up the mantises each spring, and bringing them in at the end of summer. Each one weighed thirty-five pounds. Each one had to be taken a quarter mile out via a boardwalk, so as to keep the tundra pristine. Sometimes people helped me with them. Sometimes, in the spring, we could use snowmobiles. But still, there was a lot of lugging. I couldn’t help thinking about what all the thirty-five pounds was doing. I knew the design.

The heavy steel frames were to support the boxes, which were to anchor the sheathing, which was to protect the cables, which were to reach the sensors. But the active guts down in the sensors was — tiny. At the other end, the frame needed extra iron to support a sizable solar panel, and a big battery, to power the weather station, which had to run all the time because it was general purpose. But the actual data chip down in the recorder was — tiny.

man carrying mantis on back across tundra, Brooks Range in the background

Packing thirty-five pounds of iron

What if I could put a tiny sensor right with a tiny data chip? Suddenly, all the boxes, sheathing and cables disappear.

The spectral readings are only in the daytime, none at night. So that’s the only time you need solar power. Could the instrument “sleep” at night, and get by with a tiny solar panel, and a tiny battery, just enough to wake it up each morning?

A good bit of the mantis design was how to get the data out. A weather station case needs robust hinges and a latch, to stay weatherproof. You open it, and plug in a cable. The other end goes to your laptop, which you have to lug out to the site. You have to make sure your laptop stays charged, and try to keep it from getting rained on too much. You have to be sure to bring the right connector cable! Also, while you have the weather station case open, it can catch rain, hail, and snow. So you put in desiccant packs to dry it out. And indicator cards to monitor that the desiccant is still working. And more desiccant packs when the first ones quit.

What if you never had to open the instrument case? What if the system transmitted it’s data wirelessly, such as by Bluetooth? No need to bring a USB cable, or worry if you brought the one with the right style end.  What if, instead of a laptop, you could pull the data in on your smartphone, which you could just keep tucked inside your jacket pocket?

I started working on this. I had not done much electronics since grad school, so I had to get back up to speed. I could not use the popular Maker platforms, like Arduino and Raspberry Pi, because they need too much power. My thing had to run on the trickle of energy available from a few solar cells. I could not depend on a wall plug nearby.

So, I had to get down to the raw microcontroller level. A microcontroller is like a one-chip computer. (The word is abbreviated “uC”, as the lowercase “u” is easier to type than the Greek letter “mu” (“µ”), for “micro”.) Many uCs have low-power “sleep” modes, but you need to program them on a chip level for this.

I found that uCs had advanced quite a lot since I’d used them in my Masters project. Overall, much easier to program. Interfaces had been standardized, so it took fewer pins to connect to other chips. I knew I needed at least light sensors, and a micro-SD card for data storage.

I got used to surface-mount components. Before this, I had always worked with through-hole parts. Through-hole electronic parts have wire leads that you put through holes on the circuit board. Then, you melt solder into the hole. This both makes the electrical connection and holds the wire in place. With surface-mount, however, the component has only metal patches for leads, or short pins. These connect to flat metal pads on the circuit board. The solder acts as both an electrical bridge, and “glue” to hold the part on the board.

through hole and surface mount light emitting diodes

Through-hole compared to surface-mount (SM) LEDs. The SM LEDs are the three pale patches in the carrier strip.

With no wire leads, surface mount parts can be much smaller. At first it was mind-bending to work with an electronic chip no bigger than a grain of aquarium gravel. Steady hand, and don’t sneeze. Pretty soon, though, I was thinking, “This one sure is wasting a lot of board space. Can’t I find a smaller version?”

greenlogger prototypes in glass jars

Prototypes in Mason jars

Some of my first prototypes were in Mason jars. I learned that you can mollify the TSA by simply putting a nice note, with your phone number, in your checked baggage. Say something like, “This is a vegetation data recorder, for environmental research.” No need to say, “Not a bomb!”

Once in a real case, I hoped it looked less like a bomb.

greenlogger prototype boards in clear plastic case

Prototype in weatherproof case

I called my device “Greenlogger”. I was doing this on my own, not part of any job. Of course I thought about eventually making some money off it, but I wanted it reliable first. So, instead of trying to sell them at this point, I offered to loan them out for testing.

There is no substitute for real-world testing. I did not know if it would work to run the instrument in a totally sealed case. Maybe the electronics would get too hot, or there would be some other problem. But I decided to try. I rigged up some basic stands from PCV pipe.

greenlogger mounted on stand made of PVC pipe

Simple stand

In field tests, the instruments worked, but I learned other things too! At one site, where researchers set Greenloggers out on Colorado’s Mt. Evans, at 14,130 ft elevation, animals tore the heads off. What else would leave teeth marks? So I had to re-design the mounts.

In a few years, my Greenloggers were standing in the field next to mantises. The solar power was keeping them charged, so they could run indefinitely.

field site in Wyoming sagebrush, both mantises and greenloggers

Greenloggers with mantises

The scheme I came up with to get the data by Bluetooth was tap-to-wake. Most of the time, the Bluetooth is shut off, to save power. When you want to communicate, you rouse the logger with a sharp tap. My design contains an “accelerometer”, which measures all forces of acceleration. A tap is a rapid acceleration, and so it’s the signal to wake up and connect.

Overall, things were working pretty well. The light sensors were getting readings that spanned about 6 orders of magnitude, from moonlight to full noon sun. I put a temperature sensor on the circuit board. This would not tell much about the environment during the day, when the case bakes in the sun. However, it might give a clue if something failed. If an instrument died, and the temperature record leading up to that was climbing and climbing; well we need to figure out how to keep things cool. At night, though, the instrument temperature would drop to ambient, and the record would correspond to local weather.

Greenloggers mounted above head-high vegetation, on long-legs PVC stands

Long-legs Greenlogger stands

It’s important for an instrument to know what time it is. Each mantis was, essentially, a semi-mobile weather station. Weather stations need to timestamp their data. If, say, the temperature is recorded, but not when it was that temperature, well, that’s not much use.

Commercial weather station instruments incorporate a real-time clock (RTC). This is like an embedded wristwatch. Modern electronics can keep pretty accurate time, to about a minute per month. For the mantis weather stations, the RTC would be set during the initialization process, while connected to a laptop. After that, it’s understood that a free-running RTC can “drift”, that is, run a little fast or slow.  To keep it accurate, you need to periodically correct any drift, and that means a field visit.

I built an RTC chip into the Greenlogger. You can set the RTC by Bluetooth, so you do not need to open the instrument case. My prototypes kept pretty good electronic time, but of course there was the inevitable drift. This was not going to be good enough for months, or years, of unattended operation.

Early on, I considered a doing it like radio clocks, which set themselves by the US standard time signal transmitted from Colorado, or perhaps use one of the European services. But my instruments might be deployed in far remote locations, out of range. I needed it to work anywhere in the world. I thought GPS would be the way, but it took a while to figure out.

GPS works by triangulating on three satellites. The GPS receiver knows the distance to the satellites by very accurate time signals, so timing in inherent in the technology. GPS output contains this time signal, along with the location.

A big part of the solution was simply the physical technology. A “GPS receiver” basically consists of a chip and an antenna. The antenna has to be good enough to pick up the faint signals from distant satellites. The chip (or chip set) handles all the complex math of extracting those satellite signals into simple usable data. Both the antenna and the chip posed serious conundrums in terms of size, cost, and power management in my design.

The good news is that GPS is becoming so universal that the technology is advancing rapidly, and things are being mass produced. I could get the chips for under $10, in quantity. The antenna, however was another matter.

Discreet antennas are expensive, and take special connectors so as not to degrade the signal. Also, they are quite “big” as electronics goes. An integrated circuit chip can be shrunk to the size of a rice grain, because it does everything by microscopic transistors. However, an antenna has to be a certain minimum size to match the wavelengths it deals with. The GPS in your smartphone actually uses part of the internal metal casing for its antenna, but this is serious woo woo design, like doing acupuncture on a cricket.

Fortunately, modules were becoming available that integrated the GPS chip right with an antenna, as well as all the onboard electronics. I designed in one of these modules, but then the company went out of business. This was right when I was putting some of my prototypes out for long term testing. So they had a “hole” in the board where the GPS was supposed to be.

Other products came available, but they were bigger, and harder to interface. “Big” may not seem like much of a complaint when the thing is half the size of your thumb, but board real estate is precious. I had finalized my design to fit in a certain small plastic case. If I had to rework that, it would be a big step backwards.

GPS is a classic example of “asynchronous”. For an electronic system to read, say, a memory chip, or a sensor, it just, well, reads it. This happens in a nanosecond, or at worst a few milliseconds. On the other hand, a GPS subsystem doesn’t just “have” the data you want. It has to go get it from the satellites. This can take a few minutes, or at worst half an hour! For my Greenlogger, this ought to be OK, because it just needs to correct for RTC drift maybe a couple times a month. But how to run that?

I had found a new GPS module that would work, but controlling it was looking complicated. My main uC would be pretty busy: First, it would try to wake up the GPS. Then, check if the GPS actually did wake up. Next, see if the GPS is transmitting anything. If so, see if what the GPS is transmitting makes any sense. If it does, winnow through the firehose-spray of information the GPS is emitting, to see if we have got a time fix yet. If good, snip out this tidbit, and set the system time. Keep track of how long all this is taking. It could be, we are stashed in a metal file cabinet somewhere, and the GPS is never going to get a reading. If it takes too long, forget it. Gracefully shut down the GPS, whether we got a fix of not. Make sure the GPS is correctly shut down, so it isn’t leaking precious system power. If we never got a fix, peek out every now and again to see whether the project scientist has finally put this device outdoors under the open sky, where we can breathe!

This would have taken about a quarter of the total computing power of the main uC, juggled in along with all the normal data logging. It would have taken a bunch of board space, and a rat’s nest of signal traces. I started toying with the idea to put all this on a separate board, with it’s own auxiliary microcontroller. The GPS module itself was already on a separate board, in order to fit everything inside the instrument case.

This turned out to be the solution. There is kind of a joke in microcontroller design, about sleep modes. Modern uCs feature a wonderful array of sleep modes. The uC can shut down functions to save power. But if a uC goes too deep asleep, so it isn’t doing anything any more, how can it ever wake up again? Kind of like the joke about write-only memory. But in this odd case, it was just what I wanted.

The main uC chip sends one time-request pulse to the GPS uC, and then forgets about it. The GPS uC takes it from there. It handles all the waking up of the GPS module, babysitting it while it watches the sky for satellites, and patiently listens to it babble about what it’s seeing. If the GPS module takes too long, its uC puts it back to bed. But if all goes well, the GPS uC finally gets a valid time from the GPS, and sends it as a set-time signal back to the main system. This is the same signal you can enter, say from your smartphone, to manually set the time. The main system has only the relatively simple housekeeping, to keep track of how many days since it last got a time update, and periodically ask for a new one. Meanwhile, every time the GPS uC runs, it finishes by swallowing a whole bottle of sleeping pills. So it then uses no more system power. This is OK, because each time the main system wants a time signal, it brute-force resets the GPS uC, to raise it from the dead.

After a number of design iterations, I finally had it so the Greenlogger could set its own time anywhere in the world. The GPS module added a somewhat uncomfortable $30 to the parts cost, but, well, how much would it be send a technician to, say, Greenland once a month for the sole purpose of updating the instrument’s clock?

Greenlogger, on camouflage colored stand, to blend in with sagebrush terrain

Camo Greenlogger

In spring of 2014, I had the opportunity to set out three Greenloggers for long-term testing, at remote sites in western Wyoming. In autumn of 2016, I went back to look for them. One had been stolen, but the other two had kept on running through two winters, recording temperatures down to -24 degrees Centigrade.

In the winter weather, one of the instruments held a record of temperature staying at exactly freezing for about three days, with muffled light levels, indicting it was buried in snow. Then, as its battery ran low, it went into hibernation and stopped recording. It remained running in ultra-low power mode. About ten days later, when it got more solar power, it woke up and started recording again.

Overall, I was satisfied how robust they were, but this was in the gap when I did not have GPS working. After two and half years, both devices had serious clock drifts, of 4 and 5 months! So now, with GPS installed for automatic time setting, I have prototypes out for winter testing in Greenland and Alaska.

I thought these loggers could be repurposed. For example, they already serve as solar site evaluators. They record just how much sunshine reaches a spot, day after day, through all weather.

They also serve as trackers. One prototype I loaned, was shipped back to me, broken. From the log, I could see what had happened. I had packed it up to ship at 6:30 PM on June 25, 2017. After that, it recorded darkness. About 5 PM on July 11, it started detecting light again. It got a GPS fix, and corrected its clock by 31 seconds. It also recorded that it was now in Alaska, on a tongue of land in a small lake in the middle of Yukon Delta National Wildlife Refuge.

Research collaboration plans change. A few days later, it recorded that it was at the airport in Durango, Colorado, evidently en route on a transfer to Greenland. On July 20, it recorded a location on Cape Cod, Massachusetts. That evening, it was dropped hard enough to reset. It lost its location, though the clock kept running. The same thing happened a couple more times over the next few days. On July 22, it was recording temperatures below freezing. Freezing in late July? Greenland! The light level traces showed the never-quite-dark summer cycle of the midnight sun.

Then, on July 24, the record abruptly stopped. I received the device back August 9, shipped from an address in Maine. The battery clip was broken off the circuit board, evidently from impact. Repaired, in Portland, Oregon, it recorded the solar eclipse of August 21, 2017.

Every electronic system needs a Reset button. To avoid having to open the instrument case, I had equipped the circuit with a magnetic reed switch. This works similar to how a strong magnet picks up a chain of paper clips. The magnetism travels through the iron, and makes normally un-magnetic pieces stick together. In the Greenlogger, you can bring a strong magnet close to a certain corner, and two tiny contacts, normally separate, will touch. Thus you can “press” the reset button from outside the case.

It turns out other things can make the contacts touch, such as a hard slam. I thought about leaving this in the design, to determine if the scientists were playing softball with my instruments. But, no; if I want that I should program the accelerometer. Instead, I plan to replace the reed switch with a magnetoresistive sensor, which responds only to magnetism, not physical shock.

close up of Greenlogger, in case

Latest design

I know of one more issue I need to fix. I call it the climb-out-of-reset hang. The commercial weather station boxes we used have this same problem. It is classic in energy harvesting designs.

If ever the battery discharges too low, so the system completely stops, the instrument can never start up again, even with sunshine blazing on the solar panel. It seems there would be plenty of power available, but this is what goes wrong:

Say the battery drains down, and the system totally dies. When there is solar power again, the battery starts to recharge. There is an exact point of voltage when the system electronics are just barely able to start. They start, and try to run through their initialization routines. However, this small sip of power is enough to drop the battery back down below the critical threshold, and the system dies again. The battery recharges. The system starts — and the cycle repeats, forever.

If you connect the system to a well charged battery, the voltage droops a little during initialization. But right after a successful startup, the system can go into battery management mode, and keep itself on a strict energy diet. However, on a slowly-charging battery, what the system really needs to do is hold off on powering up, at all, until the charge is well above the start threshold. But the system had no way to know how to do this — because it’s still dead.

Since the system cannot rely on the intelligence of the uC, it needs some sort of hardwired holdoff circuit right up next to the battery. This is tricky because these “dumb” electronics have to work, and make critical decisions, based on fractions of a volt. Not much electronics works reliably on fractions of a volt!

Everything else in this project was impossible, but now it’s working. It’s going to be really cool when the holdoff circuit is too!

Documents for using Greenloggers: https://github.com/rickshory/Greenlogger-Docs

The code that runs the Greenlogger microcontrollers:

Main board uC code: https://github.com/rickshory/AVRGreenlogger

GPS board uC code: https://github.com/rickshory/GPS_time_841

The printed circuit boards: https://github.com/rickshory/Greenlogger-PCBs

The mantis was developed under government research funding, and so is in the public domain. I wrote sections of the documentation, which includes instructions for data processing. We developed two tools for processing data from the mantises, Greenloggers, and other recorders such as iButtons.

The documentation: https://github.com/rickshory/mantis-docs

The two data tools:

One using, Microsoft Access: https://github.com/rickshory/mantis-Access

A cross-platform one using Python: https://github.com/rickshory/NDVI-modules


man wearing eclipse glasses

1 Comment

Eclipse, high desert

The eclipse was an exercise creative anxiety. Not the eclipse itself, but planning to go there, and not knowing what to expect. Gas shortages. An estimated million extra people visiting the state. Projections of the “biggest traffic event in Oregon history”. Not to mention, the weather.

Months before the eclipse, I looked at maps. The eclipse would be in the morning. West of the Cascades, August is usually good weather, but there can be morning fog. So I looked to the east side.

All across the interior West, the path of totality crossed remote roads, built for a handful of vehicles per minute. In normal circumstances, you would have no trouble going there, pulling off on the shoulder, and watching the show in the sky. But with a zillion other people who have the same idea, it could be a disaster. One breakdown, and everybody’s plans are for nothing. Even finding shoulder space could be a jumble. Every summer a flash mob congeals to watch Fourth-of-July fireworks over remote Lake Crowley, Hwy 395 north of Bishop, California. This was going to be a lot bigger than Fourth-of-July.

Most of the central Oregon high desert is rugged and mountainous, with only a spiderweb of roads There was one place, though, that stood out. North of one little town, the path of totality rolled across a V-shaped triangle of flat agricultural land, gridded with farm roads. Surely in there would be lots of space, and road shoulders. The little town, at the bottom point of the “V”, was named “Madras”.

So, it would be a matter of being there on that Monday morning. Maybe you could just get up early, and drive across the mountains. But again, a zillion other people with the same idea. Better to be there ahead of time.

Motel rooms in Madras sold out two years in advance. But Madras did a logical thing. They set up a campground a couple miles north, at the intersection of Highway 26, which forms the west side of the agricultural triangle, and a road going east into the triangle named “Dogwood Lane”. They called it “Solartown”. A bunch of us went in on five of these campsites. So that’s where I would be for the eclipse. On Google Maps, Solartown showed as an alfalfa field.

I tried to plan ahead. I reserved my rental car months in advance. When I picked it up, I stocked it with twelve gallons of water and ten gallons of gas. I colluded with Chris to carry food we could eat without cooking. I pictured the whole thing as a dry run for a natural disaster. No way to get anywhere. No way to get anything. No way to find out what’s going on.

Chris works nights, so with one thing and another we could not leave till at least Saturday. I kept checking the traffic. There were reports of jams in central Oregon. But, miraculously, the route to there seemed clear. I pictured everything becoming bumper-to-bumper on a moment’s notice. Again, whatever I might think of doing, everybody else might also. I considered driving over during the night, but the hundred miles over the mountain passes are nice scenery, not to be missed. We took a chance and slept the night, home in Portland. We got going about 7am on Sunday.

Heading out out of town, there was no significant traffic. Less even than a ski weekend! About halfway, on Mt. Hood, we relaxed and took a break at Timberline Lodge. Then on down the east side.

At Warm Springs, the air was full of smoke. Would it obscure the sun? But we continued on mile after mile. The smoke cleared out. The road stayed open. Only as we came to the agricultural triangle did the traffic get thick.

Right away, it was evident that, at least if money were no object, you could find a place to stay. Campgrounds and festivals were springing up all over. We plowed ahead down Highway 26, towards the intersection of Dogwood Lane.

The acres of tents and RVs came in view. But when we got there, the turn in to Dogwood Lane was blocked off, closed. Seriously closed. Barricades. National Guard. We rolled on south to the outskirts of Madras.

In hindsight it was obvious. Of course it would not make sense to have everybody turn in directly off the highway. Half the people turning would be making left turns. Instead, you’d come in from the east, through the grid of farm roads.

I won’t go into all the details, but it was a zoo. Cars were lined up stopped on the asphalt. Inching ahead a little every ten minutes. We got to know individual juniper trees and tumbleweeds. A guy driving from the other way slowed down to say you could get in quicker on this other route. Driving miles around, to try the other road. Getting lost. Getting directions from people in orange vests at the intersections. Finding out their directions were wrong. Maybe even they didn’t have current information.

Eventually, we were in a line of stopped traffic again, but at least the Solartown entrance was in the distance. We sat with the engine turned off most of the time. The sun was getting hot. People walked like refugees along the road shoulder. There were lots of bicycles. Bicycles would have been a good idea.

We inched along. We made the last turn. With stops and starts, we were getting close. I got out and walked ahead. The directions emailed from our co-campers a few days before didn’t seem to fit. Things change fast in Solartown. Sometimes I was able to reach people by phone. But the connection was bad, and kept dropping. I walked ahead.

A drone hovered over the creeping line of cars. A guy was talking into a microphone. “Are you the news?” I asked. They were doing a story on this, the traffic, the scene. That’s what it’s all about, I realized, the scene. No matter if we never get to the campsite, the eclipse will happen, and we’ll see it. But equally memorable is the scene.

Now we were in the thick of Solartown, not knowing quite where to go. Trucks rumbled past, sprinkling water to keep the dust down. Trucks rumbled past to pump out the porta-potties. There was the shower trailer, as per the directions, but now where do we go? I got through to someone by phone. Oh, we should have turned right. We convinced the cars behind us to back up till we could wedge through the intersection. Down this lane. Now go in here.

Finally, we were to the campsites. There was a parking space for us. Slip the car in. Turn off the engine. A lot of the guys there, I had known from California years ago. We had really only been stuck in traffic about three hours. Nineteen hours till the eclipse. I lay down under the shade canopy and caught up on sleep.

There were various things to do the rest of the day. One RV had a wall with displays and information about eclipses. People had telescopes with solar filters. Across the fence was the the alfalfa field from the aerial photo.

tent city across the alfalfa field

Tent city

Being as I am a behind-the-scenes kind of guy, I was impressed by some of the behind-the-scenes planning, namely the land prep. Someone had thought ahead. They used the agricultural technology as for planting wheat, and had drill-seeded the site with a turfgrass called “tall fescue”. How do I know what it was? I’m a grass botanist. In the supermarket you see strawberries. You know they’re strawberries, not raspberries, because you’ve worked with strawberries. It was a good choice. Tall fescue is rather coarse, but very tough.

There were mysterious bare-dirt swatches a couple feet wide curving through. On investigation, these suggested wheel tracks of center-pivot irrigation. This was desert land. The grass seed would not have grown without added water. The dirt tracks continued on into the carrot-seed field fenced off adjacent to the campground. There was the center-pivot pipe, parked out of the way.

To make it easy for people to see the campsite boundaries, they were tidily gridded off by foot-wide brown stripes in the grass. Similar with the driving lanes. What had browned the grass? At first I thought maybe heat-kill, but how would you do that? I talked to a turfgrass guy I met, and he pointed out some of the overspray stipple. They had used standard weed-spray technology, glyphosate, to “paint” the brown.

The “lawn” was a bit bumpy for bicycling, but fine for walking barefoot. Surely, most campers never gave it a thought. It was just there. But they would certainly have noticed if it were still an alfalfa field, or grain stubble. Every step would have been torture.

I was impressed by the farming artistry. If you want to grow wheat, you do this. If you want to grow carrots, you do that. If you want to grow a campground, you do like so. Tall fescue can be tricky to get a good stand of, but Solartown was fine. Probably two weeks after the eclipse, they were going to plow it all up and plant something else.

grid pattern of sun crescents cast by the holes in a colander

Sun crescents through a colander

Night came and we bedded down. Morning came, and the eclipse began. There was the first bite out of the corner of the sun. Guys played with the crescent-shaped sun images coming through tiny holes. The daylight didn’t seem a lot less bright, but the temperature started to feel cold. Chris asked for a sweater. By then we were about a quarter mile away in a farm field. “Sorry, you’re going to have to tough it out.”

people all facing the same angle, viewing the sun with the first bite out of the sun's disc

First bite out of the sun

people taking pictures of a man with the pattern of sun crescents across his face

Funny portraiture

people ranged across an agricultural field, waiting for totality

Sun worshippers

The darkness came on, and then there was totality. The land was dark, like night. But all around the horizon was a distant glow, like sunrise or sunset. If the usual figure of the sun is a shining face, totality is the face of the dark sleeping moon, ringed round with wild white Einstein hair of the solar corona.

man wearing eclipse glasses

Wear those glasses

As soon as light returned, “Bye, I’ve got to get Chris back for work tonight.” The campground was gridlock, but we made all the right choices. We cut through the farm roads to the east edge of the triangle, Highway 97. On the way, we could see Highway 26 a mile or so over to the west, a line of vehicles glittering in the returned sun, scarcely moving. An ambulance wailed as it worked up along side them. Oh, God forbid!

On the farm roads, scattered cars were parked along the shoulders. Sure enough, it looked like people had done exactly what I’d first thought of, drive in early that morning. Now, those day trippers were sitting in lawn chairs reading books, waiting for the tens of thousands of visitors ahead of them to get out of the way.

On our route, we were maybe two hours inching along the highway, then things opened out and it was fine. We drove over the scenic high plateaus east of the Deschutes River, got to Interstate 84 at The Dalles, and were home in Portland by 4 o’clock.

In the pre-eclipse anticipation, I recognized a long-forgotten feeling. It felt like High School days, when the world is full of lots of thing where you don’t know what to expect. Someone scoffed at High School being “the best years of our lives”. Yes, a lot’s awful, but it fills your sails. I’m going to put myself in more situations where I don’t know what to expect