Some of the best sources of big data are from those things that are furthest away, satellites. What data are satellites good at providing and what can’t they provide? What are different tools for collecting data that are necessary and compliment them?
Indeed satellite-based remote sensing data are becoming very detailed, in terms of spatial and temporal resolutions, and imagery analysis is already a powerful tool supporting decision making at multiple levels. For example, Digital Globe now has a historical archive of sub-meter resolution imagery for more than 15 years across the globe. This is a tremendous resource that we can use to see the changes in plant growth and stresses at plot level and deforestation and land use at landscape scale. The ultimate holy grail of using remote sensing in agriculture is the accurate prediction of crop yields especially in smallholders’ farms. However, we are not quite there yet. There are many methods being researched very actively, yet most of the methods are developed in rather homogeneous, large-scale farms in U.S. and Canada, not so common in developing regions like sub-Saharan Africa, where 70% of farms are intercropped and weeding is not done well. For these reasons, groundtruthing is still very much needed. We need a good collection of in-field crop cut and household survey data to help understand how farmers manage the fields and how much they produce and, relate this data with how they look from the space. There are many projects doing this, but there is no coordinated effort to share those groundtruthing data yet. This will be a real low hanging fruit, to develop a way to pool all the groundtruthing data from small farmers’ fields and make a big dataset to fine-tune the yield prediction models.
We talked about big data approaches and how they affect agricultural production at farms. What about governments? Do you see the role of big data in the policy making processes?
Yes! First, policy makers need lots of data beyond the farm-level. The boundary of agriculture expands to a broad spectrum of sectors much beyond the traditional definition of production-focused agriculture – such as logistics, infrastructure, and energy. To know where to invest on what, what are the bottlenecks in the value-chain, it’d be essential to have access to these variety of data accessible and interconnected. If you can monitor consumers’ demand change from the food markets’ point of sale over time, for example consumers use mango pulps more than legumes in dai, you can perhaps inform farmers and help the value-chain to adjust supply and demand in advance and avoid any dramatic price shocks. When data analytics tools are provided together with the large datasets from multiple sources across sectors, government and policy makers will be able to make much more targeted investment and planning decisions. In India, for example, last year Ministry of Finance announced an ambitious plan to double farmers’ income by 2022. Our research showed that this goal can not be achieved by only focusing on staple crops, such as rice, wheat, or maize. Policy should be made to encourage farmers to grow more vegetables and fruits and diversify for additional income generation. Big data and analytics can help the Ministry to develop a set of detailed scenarios as blueprints, analyze trade-offs, and monitor the progresses and also provide course-adjustments through real-time data. There are lots of opportunities, and we will see these examples more and more, likely starting from here in Telangana.
CGIAR-CSI 