pre-GOSH: Open Soil Research


#1

Hi
I will try to arrive at Shenzhen the 7 and want to do some soil work 8 and/or 9.

@gbathree I don’t see how can I use your data flow right now but I’m interested on the Handheld Reflectometer, where are you with this?

Another interesting subject would be soil respiration/CO2 chambers. I know @dusjagr and others are advancing with that. I already proposed a Workshop inside GOSH but maybe is better to do something in the pre?

I have been colecting some resources about soil health (from gosh and non) and I’have tried to map them here. It’s kind of incomplete and messy but you can filter things. Is a very first approach. I wanted to visualize gosh tools/resources/complements/possibilities. I’m still not sure if it is very usefull, What do you think? Do you know a better tool to map or maybe can suggest a better method?

We are just beginning to work on the soil of a grape farm so I’m kind of toomuchmotivated… The only test we have done so far is soil chromas. They don’t look so pretty but we are learning very fast about bokashi and other bio-supplyes to improve them.

saludos


pre-GOSH activities in Shenzhen
#2

Yip, we have loads of ongoing discussions and activity in our current HUMUS.Sapiens | Open Soil Research initiative. See more info on the hackteria forum, specifically on the chambers i documented more on thewiki..

Let’s start a dedicated threat / topic on pre-GOSH soil geeking!

m


#3

Hey Marc! I have been following Hackteria since GOSH 2017 after being introduced by @ShamsJaber. We built the DIY microscope along with the kids! I am interested in working and learning more about the ‘Plastic Fantastic’ project in Hackteria forum. Will there be any activity around that? I am currently working on mechanical recycling machines by Precious plastic to tackle plastic pollution. Let me know.
Cheers!


#4

Sorry late replying.

Yes @nanocastro , we have built a small lab to test soils, and are building calibration database using the reflectometer to soil carbon. That works quite well - with good metadata (like soil type), you can get r2 in the .7 - .8 range for soil carbon prediction. Calcium carbonate (which is a white form of soil carbon) screws up the model, but it is fairly unusual and is sometimes easily visually identified in soils.

In total, between our work with the bionutrient food association and @dan.kane at Yale, there will be several thousand soil samples all with metadata (soil type, location, etc.), soil carbon data (we’re using the LOI method (burn carbon in an oven, weigh before weigh after) and Dan K is using a gas chromatograph), soil respiration (measure CO2 produced by soil over 24 hour period), and minerals using XRF (from Na --> U).

We’ll be keeping fairly large amounts of extra of all of the samples, and I would LOVE! to collaborate with anyone interested in using them, measuring additional parameters (like soil chromatography), etc etc. Best would be to come here and stay for a few weeks if you wanted to test stuff - I can cover your stay and dinners while your here!

I’m arriving in Shenzhen on the 9th, and staying an extra few days and could bring reflectometers and CO2 sensors similar to Marc.

We just got an order for 300 reflectometers from the bionutrient food association, so we’ll be making a bunch, probably doing PCB assembly in china with Elecrow. If I can sneak some in, perhaps I could get a few to you all rather than shipping.


#5

Hi Greg
Thanks for the answer and your generous proposal, I’m confident that something to collaborate is going to arise at GOSH.
It would be great to have in hand one sneaked reflectometer …
Saludos


#6

I really also want to see your method for soil chromatography, and it’d be super fun to use machine learning to see what we can correlate to those images. Could you do a session on soil chromatography and what you’ve learned, or something like that? Don’t know if others are interested, but I definitely am.


#7

hi Greg
Sure, we can do a circular chromatography session. I will have a few meeting with the guys that are working at the university to get the test protocol right. I’m not sure If I can transport the reagents with me but maybe we can get them at Shenzhen (nothing too strange, just silver nitrate and NaOH (analitical grade).
saludos


#8

Hi Greg!
@gbathree
If you have the opportunity to do device (reflectometers) yet for us it would be great. Thank you.


#9

@gaudi you wanna add some plastikologi to pre-GOSH activities? maybe you can ask your colleagues from LU to support?

m


#10

Hello everyone
I will arrive to Shenzhen on the 6th
Thanks to @leandro.mastrantonio I already have most of the material to do some soil chromatography there. I hope everything makes it through customs so I can show you @gbathree. @Rachel and everyone else of course you are more than welcome to join in.
Looking forward to meet you all for some chinese soil geeking
Nos vemos pronto


#11

let me know how plans develop! I arrive on the 4th and can meet up the 6th, but for the 7th-9th I planned a side trip to Guilin (since it is my first time in China!) - will be back in Shenzhen the afternoon of the 9th…


#12

Awesome!!! I’m super excited. If you have any background research, or work showing connections between soil chromatography (your own or others) that you wanted to share or present on I’d love to hear it also.

I’ll be there on the 9th, though I’ll be doing a lot of organizing stuff, I think I could take some time for this.

Glad this came together!


#13

Hey! What is the final plan coming out of the thread? I am evaluating what to do on the 7th, 8th and 9th and this thread caught my eye.

Thanks!

JP


#14

I think I already shared this one page summary from my interesting experience doing the circular chromatographs… but here it is for the thread!
Again some space and dark spaces are needed (for the silver nitrate) and a set up to do timelapses of the development can be nice!
They are really art - like irises of eyes! We made duplicates so could see different samples are quite reproducible. Does anyone have the book mentioned? The author really compared factory farming vs. bio setups (also when the same site is turned to bio…) the different colors tell about how much organic matter and various minerals are in each sample - do one of you have the ‘code’??


#15

Hey is @spMohanty coming to GOSH? This is a perfect application for visual machine learning. You should come to the soil research pre-event Mohanty if you’re coming.

Also, we run a small soil testing lab, if this is fairly inexpensive and easy, we could run hundreds of samples alongside what we already do to get some data to correlate to. Could be so cool!

This reminds me of the pHlip test in biodiesel (http://www.phliptest.com/index.html) - complex (nuanced) visual indicator of quality.


#16

Sorry, two posts in a row, but wanted to throw this in.

One more soil related item. So there’s a lot of research showing the importance of compaction in both agriculture and arboriculture, and the tool to measure compaction is a penetrometer (example - https://www.amazon.com/Dickey-John-15585-Soil-Compaction-Tester/dp/B004NELR6K).

Problem is, they are hard to interpret, as you have to correlate pressure with distance under ground, and take several samples around the tree. It’s a classic “the information is there but just too hard for most people to interpret” problem (probably similar to soil chromatography).

Anyway, we’re working on integrating into Our Sci, to use the phone camera to ‘see’ the pressure gauge in real time, so you can graph the pressure. We would need to identify some other tech (maybe low-cost time of flight sensors like those on adafruit… or somethign else?) to get real-time depth readings.

But we also really want to integrate a simple (impedance) soil moisture sensor on the end of the penetrometer. If you did, you would not only graph soil compaction by depth, but also soil moisture by depth. This would provide a pretty complete understanding of where compaction is happening, and where it’s impacting water infiltration, which is a big deal.

I’m thinking about bringing stuff to play with this or try to make something… may be bigger than I can do there in a day, but if folks find it interesting I could bring stuff.


#17

This might be something we could also image (and analyze) with a FlyPi when we integrate some CV there! It seems like CV might be sth many people would be interested in…


#18

just a quick point, with those 20cm circles, we had duplicates for about 10 samples, I think. For 100 samples, we would really need a lot of space…
Not just computer space. (samples are first treated in flasks with a very basic solution, then silver nitrate is run into the circles in the dark, and then the soil extract is run, with time lapse imaging…)
Btw. I love automatic imaging analyses, and have been getting some students to look into automatic micronuclei detection recently - any chance you might have input for us on that??


#19

@gbathree: I am indeed coming to GOSH (assuming my VISA works out smoothly !), but my current plan is to reach Shenzhen on 9th October (Tuesday) around noon, and then I fly out on Sunday, 14th Ocrober.

And regarding the automated testing approach, its definitely worth a try.
Now, if we are using the phone camera to “see” the pressure gauge, then we are just making the problem un-necessarily hard; coming up with a neater solution which can be hooked onto any smartphone and stream the data would be the idea case.
And even in that case, we will still run into some of the same issues that we discussed last year at GOSH, with time series data like in this use case, its expensive to collect large amounts of data (unless we can crowdsource the data collectiton efficiently), and without large amounts of data, the data we will be playing with will not be representative of the real world data we wish to deploy the final model on; and hence the model will give flowery results for a good publication, but in all probability would be over fitting to the collected dataset.

But if we can put together the soil moisture sensor, depth sensors, pressure gauges and can collect some data around it; then atleast we can see the feasibility of coming up. with automated approaches. And once thats reasonably demonstrated, we will have to focus on figuring out how to collect vasts amounts of diverse data (which is pretty much what we machine learning researchers are doing these days -_- )

looking forward to this !!


#20

This is a bit off topic : but I would definitely want to have a workshop on automated analysis of signals (I carefully use the word signal, as I am talking about everything from noisy time series from cheap sensors to HD video feeds, all of which we are more and more treating in a unified way over the last couple of years) in context of Open Hardware.
The bottleneck in almost all the cases is far from the actual machine learning model or experiments, but really about figuring out how to collect large amounts of good data (They even call it Software Engineering 2.0 these days). The crowd at GOSH could really contribute a lot to this discussion; especially when meaningfully done in context. of concrete projects that some of the attendees care about !