Hacker News new | past | comments | ask | show | jobs | submit login
Open Soil Spectral Library (soilspectroscopy.github.io)
100 points by protontypes on Aug 1, 2022 | hide | past | favorite | 16 comments



https://www.thermofisher.com/blog/mining/can-you-name-all-17...

Some of the elements that fisher might use like gypsum, not sure if any of it is safe either though. Soil needs more research, is what most people think but growing things is getting harder not sure if the opensoil project can cover enough of the problems yet.

A more safer alternative might be using transition metals and post transition metals in plant soil. Though using transition minerals usually results in having to be treated with something that can cover coding soil too.


Soil needs microbiology, gas exchange and hydrology this is what makes elements available. Nothing needs to be added and all the elements are plentiful. We killed the microbiology, the research is already done. Keep many kinds of plants in the soil year round, stop adding chems, this increases yield 3x and water retention enough to reduce irrigation demand up to 70%. Strange thing is, this is free and removes a bunch of business models + gov subsidies from industrial ag and the universities that support them through "research". Just finished a year talking to farmers and growers of all kinds across the US.


> We killed the microbiology, the research is already done. Keep many kinds of plants in the soil year round, stop adding chems, this increases yield 3x and water retention enough to reduce irrigation demand up to 70%.

Permaculture, no-till, ... Not sure there are definitive research conclusions on this, especially "3x yield".

IME, the conclusions are not as straight cut, nor as general and imply profound changes, such as mixing crops, cultivating more land, reducing yield, never leaving the soil exposed. Many things that have to be done together, with local optimizations, to get significant improvement.

Just like "electric cars" don't directly solve the "many privately owned petrol cars" problem. A solution being the "electric car tech" + another ownership / sharing model + urbanism changes.


"Applications to a sandy soil of the partially purified surface-active compounds improved soil water retention up to 314.3% compared to untreated soil. Similarly, after 36 h of incubation, the humidity uptake rate of treated sandy soil was up to 607.7% higher than untreated controls." ... "Overall, results revealed that polyextremotolerant bioemulsifiers of bacteria from arid and desert soils represent potential sources to develop new natural soil-wetting agents for improving water retention in arid soils." https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5984429/


Have you gone out and done some experiments for yourself? or is this from reading about it?


Water retention is something but food that isn't as good or yeild that results in not as good food is a even worse problem too though, the specifics about gypsum should be more researched. I suggested transition metals because it's different and it may be safer but even the alternatives associated with fisher are just as scary for food growth too.


Wow, can you point me to research and books on this? I happen to have a bachelor degree in the field and am always eager to find non traditional ways to grow especially pasture for dairy cows, but anything harvestable will be interesting just for the soil science alone.


Hi Tim, can you provide some resources + links?


Wow what a coincidence! I was just looking for open source soil spectral datasets earlier today. Thank you for posting.


What came up as the first result when you searched for open source soil spectral dataset if not this collection?


From the docs, Database: read only details on this page https://soilspectroscopy.github.io/ossl-manual/index.html#os...

1.4 OSSL mongoDB MongoDB is an Open Source noSQL DB hence fast and fully scalable and extendable (affordable costs for cloud solutions such as MongoDB Atlas and similar). TensorFlow and other cutting-edge ML algorithms can be easily integrated and served through a GUI.

To access OSSL DB best use the mongoDB either through a graphical user interface using Robo 3T, or by using the mongodb via R. The following parameters (database credentials) allow ready only access to DB:

Funding from: NIFA Invests Over $7 Million in Big Data, Artificial Intelligence, and Other Cyberinformatics Research


> MongoDB is an Open Source noSQL DB hence fast and fully scalable and extendable

'hence'? Does not follow. Sure, this particular implementation may be considered fast and scalable. The wording seems to indicate it is 'fast' and 'scalable' just because it is NoSQL.


This is from the docs, not me. I was providing the details to quickly get others connected. I could have made that more clear.


Anybody have a good resource for building hardware for this sort of thing on a hacker budget? I’d love to build a Tricorder for my garden.


If you want to get some serious machine I would look for NirScan Nano. It is not cheap (~1,5k) but the performance is solid. Another possible route is getting mems based spectrometry sensor from Hamamatsu (~600) but building well performing acquisition board is not that simple and evaluation board is quite expensive.





Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: