"Data Wallet" Background doc

Context

High level requirements for a repository where farm details + practice records captured as part of the Bihar MRV project can be stored to support the “data wallet” whereby farmers grant consent to share that data for value add services.

Prepared on Aug 20 for FarmOS to inform Aug 25 demo and help us take a call on whether to use FarmOS as a repository.

Overview

1. We need a repository or "data wallet" where we can store a bunch of data about farms and practices followed on those farms. 

   1. This season we are recording data on 250 farmers, want to cover 1,600 farmers next season (Rabi or winter season; November 2021 to March 2022) and 25,000 farmers during the 2023 Kharif or Monsoon season (July to October) 

   2. Key roles:

      1. Data Collector: JEEViKA’s VRPs capture and and verify data on farmer practices. They happen to be using a data collection form developed by DG to do this.

      2. Data Steward: DG plays this role as its helps farmers get visibility to what data is being captured and grant consent to share the data.

2. The individual farms are small, 0.5 to one hectare and for each season, we are capturing ~20 data datapoints related to practices followed which are collected via ~5 forms deployed over the course of the season. The data is captured via Kobo Collect by frontline extension workers (also called Village Resource Persons or VRPs), validated and then recorded. Over time, we may move to a model where farmers report data directly and VRPs validate it. See workflow below: 

Open

3. An initial use case for this data is to it to generate emissions estimates; to do so, need to re-format the data to meet the input requirements of various models (CFT, DNDC, CENTURY). Some of the data required for this may come from third party sources like ISRIC Soilgrids and providers of weather data. In this case, I expect we would query those sources based on farmer lat/long and the farmer specific data could be added to their FarmOS repository.

4. We want to store derived values from these models like GHG emissions per hectare and yield scaled emissions in each farmers repository/wallet and share it back with the farmers and VRPs via some interface (TBD) and also publish this data to a dashboard

5. Looking ahead, want to build a workflow where farmers can grant consent to share certain data fields with various stakeholders. This could be facilitated via Farmstack in two patterns. We are open to whatever makes most sense as a generalizable approach:

• Consent is collected through FarmOS who then shares data and push notifications to the farmers

• Data consumer gets consent from the farmer and presents it to farmOS to initiate data transfer

Questions for FarmOS 

1. Which hosting options makes the most sense for this use case and what does it require from DG?

2. Does FarmOS or Surveystack have a mechanism for capturing polygons that represent farm boundaries?

3. Are the individual farmer records separated in some way? How do we manage who has access to the records such that farmers can access their own information but not others? How is this related to the “single instance, multiple user” work?

4. How do we access this data once its recorded in FarmOS?

5. Whats the status of the integration between FarmOS and CoolFarmTool and DNDC?

6. How is FarmOS differentiated from some of the other open source farm management tools like Litefarm, Agrosense, Tania, etc.)

Additional stuff

• Sample data file

• Project overview doc