6.7 Data privacy and security

Only the specific theme related content has been highlighted here. To a get feel for the scenario, read here.

1. Data onboarding

The number of stakeholders in project SIS will require a keen eye over privacy and security of data. For onboarding, SoilScience researchers will have the responsibility of bringing in TPP data, but on-the-ground data collection is spread between the GSS teams and the project partners. Security of data was a concern raised during the initial proposal of the project, given the vulnerable status of farmers in Dataland. Malicious actors that came across their interviews or soil sample test results could potentially take advantage of the farmers with fraudulent schemes or predatory competition. To avoid this problem, only PPs and authorized staff at the MOA will be able to onboard (raw) data collected to the project’s centralized repository, with further responsibility for ensuring deletion of any data stored in unsecured, local devices (like those used to record interviews) and keeping oversight on GSS and farmers.

This system will be governed by a simple login system. As per the CDM, the centralized repository data will be onboarded to is a private Github repository, with access only provided to SoilScience, PPs, and authorized staff at the MOA. SoilScience will audit the work of all parties in data collection to ensure privacy and security.

2. Data processing

All PII will be removed from the data received from Dataland, as recommended in the data privacy and security workbook. This includes deletion of names, contact details, and specific location (although latitude and longitude will be reserved for dataset linkage in the next stage before deletion). For the sake of version control, this processing is not done by overwriting the raw data sources, but by making a ‘clean’ copy of each one. However, after the processing is complete, the raw data sources will be deleted from the repository.

3. Data enrichment

After the datasets are linked, the latitude and longitude columns will have served their purpose. Although it would be nice to keep their measurements for each soil sample and its corresponding meteorological data points for reproducibility, releasing latitude and longitude in the final SIS dataset would put farmers at risk. However, to retain a level of geographic granularity, latitude and longitude will be used to categorize entries in the dataset by sub-regions in the Dataland Highlands. We will likely use sub-regions based on the data points’ relation to landmark geographic features—for example, some data will be categorized as ‘West side of Mt. Data’. Once these categorizations are made, latitude, longitude, and linking fields can be removed from the dataset. It is likely that we will delete these closer to time of product release so that we can validate our work and link further datasets if required.

4. Data analysis

More high-level analysis will be run on the data, as well as the entire repository, to ensure it is clean and, importantly, that farmers’ details have been deleted properly. Only authorized researchers at SoilScience will be allowed to do so, although Dataland’s MOA will be consulted on findings afterward.

Only the specific theme related content has been highlighted here. To get a feel for the scenario, read here.

1. Data onboarding

Our main concern with regard to privacy and security is ensuring that the data is only accessible to core people in the project. Although the data is not necessarily sensitive, we would like to keep things organized and safe, especially if required by the satellite data licenses.

To do so, we are designing a login system with four levels, as suggested in the instructions.

• Uploading: WRO researchers can upload their collected ground field data, and edit it if required. Similarly, Project Partners (PPs) can upload transcripts from the interviews with farmers.
• Uploading and analyzing: SoilScience researchers can upload satellite data and edit it. Meanwhile, they can view and edit other data in the repository and then use all of the datasets to build the topographical map and empirical analysis.
• Content management: The PO at SoilScience will be the only person with authorization to delete files in the repo and change the overall folder structure.
• Viewing: Consumers of the final outputs (i.e., users of the topographical map) will only have permission to view the map.

2. Data enrichment

Building the points from latitude and longitude could potentially create a privacy risk. Although interview data will likely not be sensitive in nature, it might still put farmers in vulnerable positions. Further PII, like the farmers’ names could have similar effects, and must therefore be identified. Via a contract, we will have consent from interviewed farmers to publish this information, but we believe education is far more important: farmers should know what their interviews are contributing to and, moreover, how they might be at risk if their data is online. Our project partners in Waterways are experienced in providing this sort of education.

3. Data analysis

The only people authorized to perform the data analysis will be researchers at SoilScience. This will once again be governed by our login system.