Modeling Soil Health Indicators at Scale for Agriculture and Climate

Compnay Name

Temporai

Service Details

Our long-term project is to build ecosystem modeling tools to measure soil health indicators globally and at scale. This idea proposal is the first phase of the project: finding data sources of subsoil organic carbon (SOC) lab measurements. SOC measurements will become the labeled data or target variable of this future model, which can be incorporated into a carbon-credit trading marketplace.

Problem Description

Our climate is changing rapidly due to the continuous emission of greenhouse gasses into the atmosphere, whether done by humans, nature, or both. Soil and biomass can sequester substantial amounts of these emissions. This process helps alleviate the emissions problem as well as provides nutritional benefits to soil health.

What is needed is an economically feasible, efficient, scalable, autonomous, and non-invasive (i.e., not digging pits) method to accurately measure the amount of carbon stored both in above-ground biomass, but also sub-soil carbon. While estimating above-ground biomass carbon sequestration is relatively easy, obtaining accurate sub-soil estimates has proven to be challenging. Our proposed model aims to solve this challenge.

Solution Description

We will apply ensembles of advanced AI (including generative methods) and other statistical models as the basis of the ecosystem modeling tools. These tools will then be used to estimate sub-soil SOC contents. In order to do this, we need to first collect accurate SOC source data from lab-analyzed sub-soil samples, along with their geo-tagged locations. These SOC source data will then be used to train future ecosystem modeling tools.

The purpose of this ideation proposal is to collect these SOC source data. If the collection is successful, then ecosystem modeling tools as a standalone AI service will be a future proposal in a subsequent Deep Funding round.

Milestone & Budget

Deliverables

Deliverable 1: List and description of the collected SOC data, as well as an explanation of how it is a suitable labeled data or target variable of this future model

Deliverable 2: Price and source of the collected SOC data

Deliverable 3: List of experts or other contributors who took part in the SOC data collection, maybe pseudonymized

Deliverable 4: A report on the SOC data collection results

Revenue Sharing

Since we are collecting the SOC data and not building out the service in this ideation stage, there will be no revenues accrued. However, if the future ecosystem modeling tool is funded by a subsequent Deep Funding round, then yes that AI service will be revenue-shared. It will adhere to the API Calls “user-friendly” template for revenue sharing. That means if this service crosses the threshold of $1,000 in monthly revenue, then 10% of the additional revenue over $1,000 will be fed back into the SNET and Deep Funding wallets.

Marketing & Competition

We will further substantiate our marketing strategy once we are at the stage of proposing and specifying our ecosystem modeling tools. For this ideation stage, we are focused on collecting SOC data sources to be used in the ecosystem modeling tools in the future.

Nevertheless, we can say that potential clients for the ecosystem modeling tools include the agricultural sector (i.e., forming groups) and climate/sustainability agencies. For example, the tools can feed these climate and ecosystem data to carbon credit marketplaces. If farmers sequester the carbon, they then receive carbon credits and sell sell to counterparties that need to fulfill their carbon quota. Ultimately, this incentivizes the farmers to utilize more efficient, non-evasive, and sustainable agricultural techniques.

Needed Resources

Here is the external service on SNET’s marketplace that can be utilized for this project upon deployment:

• SYBIL: Use this service to forecast time-series and sequential data. This service can provide general forecasting services to future ecosystem modeling tools. See our separate SYBIL proposal
  here

  , which is awarded in DF2 - Pool A [New projects].

Long Description

Company Name

Temporai

Summary

Our long-term project is to build ecosystem modeling tools to measure soil health indicators globally and at scale. This idea proposal is the first phase of the project: finding data sources of subsoil organic carbon (SOC) lab measurements. SOC measurements will become the labeled data or target variable of this future model, which can be incorporated into a carbon-credit trading marketplace.

Funding Amount

$5,000

The Problem to be Solved

Our climate is changing rapidly due to the continuous emission of greenhouse gasses into the atmosphere, whether done by humans, nature, or both. Soil and biomass can sequester substantial amounts of these emissions. This process helps alleviate the emissions problem as well as provides nutritional benefits to soil health.

What is needed is an economically feasible, efficient, scalable, autonomous, and non-invasive (i.e., not digging pits) method to accurately measure the amount of carbon stored both in above-ground biomass, but also sub-soil carbon. While estimating above-ground biomass carbon sequestration is relatively easy, obtaining accurate sub-soil estimates has proven to be challenging. Our proposed model aims to solve this challenge.

Our Solution

We will apply ensembles of advanced AI (including generative methods) and other statistical models as the basis of the ecosystem modeling tools. These tools will then be used to estimate sub-soil SOC contents. In order to do this, we need to first collect accurate SOC source data from lab-analyzed sub-soil samples, along with their geo-tagged locations. These SOC source data will then be used to train future ecosystem modeling tools.

The purpose of this ideation proposal is to collect these SOC source data. If the collection is successful, then ecosystem modeling tools as a standalone AI service will be a future proposal in a subsequent Deep Funding round.

External services

Here is the external service on SNET’s marketplace that can be utilized for this project upon deployment:

• SYBIL: Use this service to forecast time-series and sequential data. This service can provide general forecasting services to future ecosystem modeling tools. See our separate SYBIL proposal
  here

  , which is awarded in DF2 - Pool A [New projects].

Marketing Strategy

We will further substantiate our marketing strategy once we are at the stage of proposing and specifying our ecosystem modeling tools. For this ideation stage, we are focused on collecting SOC data sources to be used in the ecosystem modeling tools in the future.

Nevertheless, we can say that potential clients for the ecosystem modeling tools include the agricultural sector (i.e., forming groups) and climate/sustainability agencies. For example, the tools can feed these climate and ecosystem data to carbon credit marketplaces. If farmers sequester the carbon, they then receive carbon credits and sell sell to counterparties that need to fulfill their carbon quota. Ultimately, this incentivizes the farmers to utilize more efficient, non-evasive, and sustainable agricultural techniques.

Our Project Milestones and Cost Breakdown

Deliverable description

1. Data sourcing: either the raw SOC data tabular format (if it can be made publically available) or a description of the said data
2. Data payment: list of SOC data sources, including name, institution, price (if paid), and other metadata
3. Recruiting experts: list of experts, their background, and their institutions (if their names can be made publically available or pseudonymized)
4. Data report: report in a Word doc or PDF

Notes: the milestone # does not necessarily correspond to the ordering that we will complete. We will finalize the ordering upon signing the contract, but the ordering may be the following: # 3, 2, 1, 4. Also, any residual funding from milestones #2 and #3 will go to milestone #1, or efforts in data sourcing.

Risk and Mitigation

The primary risk is the ability to obtain quality SOC lab data. The factors of this risk include:

1. The time, effort, and expertise to source this data
2. The monetary cost of acquiring the SOC data, which may require data purchases or subscription services
3. Spurious or incomplete data

We allocated a portion of our budget (milestones #2 and #3) to mitigate factors 1. and 2. respectively. Factor 3. is the reason why we started with this ideation proposal instead of a full ecosystem modeling tool proposal. If we cannot obtain quality SOC lab data, then we cannot move forward with the ecosystem modeling tool proposal, which will command a higher budget than $5,000. Therefore, this ideation proposal can also be considered as a feasibility study of a core component of the ecosystem modeling tools: the SOC labeled data for it to train on.

Voluntary Revenue

Since we are collecting the SOC data and not building out the service in this ideation stage, there will be no revenues accrued. However, if the future ecosystem modeling tool is funded by a subsequent Deep Funding round, then yes that AI service will be revenue-shared. It will adhere to the API Calls “user-friendly” template for revenue sharing. That means if this service crosses the threshold of $1,000 in monthly revenue, then 10% of the additional revenue over $1,000 will be fed back into the SNET and Deep Funding wallets.

Open Source

We will try to make the SOC data publicly available if possible. As for the future ecosystem modeling tool, if it is funded by a subsequent Deep Funding round, then yes it will be open-sourced. It will be under the GNU General Public License (GPL) v3.0 License.

Our Team

Team

Kevin R.C. - Ideation Lead

• Senior Data Scientist / AI Researcher
• DF2 2x Awardee (
  SYBIL

  and

  Onboard NeuralProphet

  )

• Former Lead Core Developer and Maintainer of NeuralProphet
• 8+ years in the finance and crypto domains
• 3+ years of Python PyPI open-source experience (including
  dnntime

  ,

  nsc

  , and

  neuralprophet

  )

• LinkedIn
  link
• GitHub
  link

Mahdi T.R. - Senior Data Scientist

• Data/Computational scientist, engineer, mentor, and YouTube content creator (
  MLBoost

  )

• 10+ years of experience in developing mathematical and machine-learning models for modeling complex physical phenomena
• Developed a code simulating microgravity solidification experiments for the NASA-ESA sponsored CETSOL project [
  Link

  ]

• Author of papers in top Physics journals (
  PRL

  ,

  Nature

  ) and ML conferences (

  NeurIPS

  )

• Ph.D. in Mechanical & Industrial Engineering @ University of Iowa
• LinkedIn
  link
• GitHub
  link
• YouTube
  link
• Bio
  link

Advisor

Matt I. - AI and Climate Advisor

• Chief Science Officer (CSO) at SingularityNET
• LinkedIn
  link

Related Links

None