Soil and data scientists unite!

Finding data can be a real pain. Environmental modellers need observational data to see how accurate a model is, is it a good representation of the real world? We also need data as an input to models, and as a starting point for models. Without data it’s pretty limiting what you can do with a model; you can’t be confident that it works well, and maybe you can’t even run it. How can we make data more connected and search-able?

In the Soil Value project we’re developing a model that simulates nutrient cycling (carbon, nitrogen and phosphorus) in agricultural landscapes. The model has been tested before in natural landscapes, but agricultural systems are heavily modified with practices such as fertiliser additions, cropping, tillage changing the soils. To test the model we require data from a range of agricultural sites that include information on how the land has been managed, and how soil chemistry has changed. To add further complication we need this data over long time scales (100 years plus is good) as soil change can be a slow process.  These kind of data sets do exist, but they’re often very hard to find.

We’ve been working with the Ensemble project and talking to data scientists in Lancaster Data Science Institute to find a solution to these problems. One idea is using the sematic web. The semantic web provides support for an interconnected web of data. Data is linked with descriptive data called metadata, and this metadata can be searched easily. The semantic web provides a suitable way to make data and models more discoverable, and allow us to reason across different datasets.

Having data we can search would make finding suitable data a lot easier, but there’s a few things that need to happen before semantic web technologies can be applied. The metatdata that describes the data needs to exist, and as a community we need to develop vocabularies (ontologies) for our metadata. We also need to put data in repositories that are findable. All of this isn’t impossible and it’s something we hope to look at further as the project unfolds. Hopefully we can develop a shift in the way we store and search data used in environmental science.

We presented these ideas at the BONARES Conference 2018: Soil as a sustainable resource. To view our conference poster check out our publications page.