Open Source Hardware for Agriculture
Lead by Greg, Nano, Mark, Ananda
Summary
Soil is one of the most biodiverse places on Earth. Soil analysis has primarily been focused on helping farmers manage soil for food production. There are a large range of tests that can be performed to measure soil health (soil chromatography, spectral analysis to determine total organic carbon, monitoring soil respiration with a CO2 sensor, etc.) with different costs and required inputs, but ultimately it depends on what resources are readily available and how to make those results actionable – usually for farmers.
There is a growing toolkit and javascript library for the Our Sci platform and more discussion on http://forum.goatech.org/
Notes
(Documentarian missed some of the intros, so please add your name)
Elizabeth – wanted to learn
Nano- working on cooperative farming, and make food, wine beer, we want to learn from our processes
Greg – his company works in ag, and has tools to share
Ananda - industrial designer and works with food banks, small scale farmers, etc.
Soil chromatography
People who want to talk about ag tech and how to scale it up OR
geek out on some tools// show and tell of tools
What tools do you have? What kind of tools do you need?
Urban food, tools for urban agriculture, food production in compact spaces
Soil chromatography – like taking a picture of soil
paper is prepared with silver nitrate, a few more steps, Then put out to dry
Are you trying to understand what is in the soil?
It is not a qualitative test
You can read some of the patterns, you can see if the soil is aerated, or more compacted, some that measure enzymes, you can see if there are relationships between the zones in the print and if there are spikes in between the zones
Nano has an interesting book
Soil that has been affected by heavy machinery, pesticides (this causes the microbiology of the soil to die, it is less complex, it is kind of a dead soil) vs. organic soil – you can see that is more complex
The soil chromatography – it is a snapshot
Soil health is a really live ecosystem that leads to healthy fruit, what is the outcome of the soil, from not healthy soil, you have less nutritious food
(Tool demonstrated by Mark)
A CO2 sensor, that you can buy for air quality monitoring, $30 or maybe up to $60, its a household sensor
reductionist approach, only one parameter
We should use each other’s platforms, instead of reinventing things
If we want to use something for the field, it has to be hardy and more robust, it will be thrown around and experience extreme conditions
The Our Sci platform, we use all parts that we can find readily available
But a cover on the soil, and the microbes slowly release C02, its called soil respiration
There are three prototypes, a button is nice, and its nice to have screen to receive results without having to transfer test results to a computer to read
The measurement takes an hour
Always make sure there is a jack, so we can sonify your data
It also allows for an alarm
You can test the CO2 monitor by breathing on it
But it still has to be calibrated
You could have the chamber installed for a longer term and to take measurements over time (though you need to lift up the chamber occasionally to allow the soil to recover.)
Greg has another tool to measure soil respiration,
We use a flow sensor so we can blow some air over the sensor to know what it is, uses a cheaper syringe that facilitates taking mass sampling
Colorimeter, shoots out a light and measures how much light comes back
It has a bunch of LEDS, shoots the lights out, it hits something, for example a nitrate test, or say soil or leaf
To do color-imagery, or you can measure soil carbon
You can also measure NVDI (i.e. greenness)
How do you take the samples?
Greg’s database is built around air-dried soils
Our Sci platform --flexible system for collecting and representing data
You can also call an API-- say soil classification data, weather data for the last 10 days and use it for predictive analysis
Data visualization, Our Sci, you just need to code a bit java script
Plotly, R Studio , Chartess,
Small scale farmers don’t want to spend alot of time looking at screens or data, for the results/ findings to be more instructive
Total Organic Carbon //
Total Organic Carbon – it retains rain water, microbial activity is higher
Humus is one of the fractions of carbon
What can the model tell farmers? Better prediction of yield
Nitrate, phosphates, and potassium (NPK) – you are worried about this if you are farming in an industrial way, and you put a lot of money spent on fertilizer
We need new measure of the health of the soil, we were stuck in this NPK thing
Microscopy of soil is useful – learn who are the good and bad guys
Microbiome maps
A set up from TOC is Total DNA
Then Differentiating general classes
Soil is the most biodiverse place on Earth
Others will put soil in box, heat it up and all the critters climb up the side, capture them, put them in water and take a look
You have to think about what is of utility to the farmer
Industrial farmer may need one type of data, household farmer may need another type of data
Yield is one thing, but quality of food is another factor
We are starting to redefine what is a good quality product
Colors are one of the keys to knowing the quality of the food
Water is major challenge, soil moisture sensors
Soil moisture sensors don’t have a long life
Could get the continuous level of soil moisture
Are there better ways to measure soil moisture?
Detecting microplastics in soils, and things like heavy metals,
Our SciKit – // Photo sync
We are going to train an AI to be able to detect different parameters from chromatographs
Minerals – // heavy metals – very expensive to measure
Repeat the protocol in your field and visualize changes over time
Greg uses spectral data, but we have talked about capturing images
But we need a predictive model to be able to have more conclusive results from a single image
You would need to send a whole bunch of samples to a lab, and tie to the images
Nano was saying even if you have a small correlation between some other data point
What are weak correlative data that could help us improve models
Tip and Que systems with satellites – if the satellite picks something up, you can queue a sensor
A lot of tools for small farmers
Question about openly accessible satellite data that is of the quality that people need
What do we want to measure?
Nano -has a map of things you can map, and how you can map them
You have to prepare the filers, the samples/ afternoon session will address the process
Other simple tests – ribbon test?
All the tests take practice, you need to pay attention to the soil, you need to wet the soil and see how it behaves
In Soil Chrome – when the soil is more grey – it is more dead
If you want resources – gathering of ag tech
We have some java script libraries for Our Sci but could use help building up the library.
