Monitoring Our Agri-food Systems at Higher Resolutions Using Non-Conventional Data
Some time ago, we published a blog, on the need for non-conventional data and data analytics to help fill data gaps, and contribute to the understanding of our complex agri-food system. When we wrote that blogpost though, we were uncertain on exactly which aspects of the agri-food system we might better understand through the use of non-conventional data. Pulse Lab Jakarta together with CGIAR Platform for Big Data in Agriculture over the past few months, has been exploring various use cases, benefits and pitfalls from the use of non-conventional data, such as pseudonymised mobile network data, internet traffic and social media data in the context of filling data gaps, and enabling high resolution monitoring of agri-food systems. In the exploratory journey so far, it is becoming clear that non-conventional data sources are helping us to better understand the behavioural aspect of the individuals (stakeholders) linked to the agri-food system, for instance behavioural aspects of farmers and consumers, and variations within and between them across space and time.
In one of our previous case studies, we showed that it is possible to use Twitter (and tweets) to nowcast prices of key agricultural commodities, in near-real-time here in Indonesia. This study was a result of consumers responding to price variability and shocks, and their response logged in on social media. This exemplifies how tweets reflected the changing behaviour of consumers, and were used to build our price nowcasting model. The ability to capture, model and even predict behavioural aspects of stakeholders linked to the agri-food system at large spatial scales, and high temporal frequencies, is helping us shed more light, on a highly dynamic facet of the agri-food system.
While non-conventional data sources often only capture digital crumbs (and don’t capture people’s characteristics in-depth), they are produced by everyone who is “digitally-connected”. This means that it is possible to obtain insights from stakeholders across the different parts of the value chain, i.e. from farmers, input suppliers, consumers, etc., hence adding to our current knowledge of complex interrelationships that inherently exist within the agri-food system.
Digital connectivity is out-competing physical connectivity. In some countries, while road access between villages may be lacking, these villages can still be connected to each other through mobile phone communication. Yet, mobile phones need not be the only means to obtain digital trails of human behaviour. For instance, access to micro-finance products and services for farmers has vastly expanded over time across rural areas in the developing world. Often micro-finance institutions maintain and monitor data about their customers (e.g. farmers), and this data is digital. In fact, this data can provide proxies on various aspects of farmer characteristics.
A good example is alternative digital credit scoring, that is used to measure and monitor the resilience of farmers to economic and climatic shocks. While most of the production end of the agri-food system is centered in rural and peri-urban areas, most of the demand driving this production comes from urban areas. A key data gap on the consumption end is the lack of a higher frequency spatio-temporal data on people’s diets, which again could be potentially obtained from digital services such as online food delivery platforms, photo-sharing platforms, etc. Digital connectedness among and between stakeholders, both in rural, peri-urban, and urban areas in the agri-food system is a reality, and there is a significant opportunity to further leverage these digital crumbs to fill data gaps, and help monitor agri-food systems at high resolutions.
Global Conference Smart Agriculture Conference
We showcased this concept of using non-conventional data sources in combination with available remote sensing/GIS and survey-based data to fill data gaps, and enable improved monitoring of agri-food systems at the recent Global Climate Smart Agriculture conference. Attended by more than 400 participants with expertise in the areas of agriculture, food security and climate change, the Conference was an opportunity to present some of our past and ongoing research related to the use of non-conventional data in the context of agri-food systems to experts, and in return we obtained sector-specific knowledge, context-dependent justification of the preliminary results.
Systemic analysis and monitoring of the agri-food systems (at any scale — local, national and international) has been challenging due to lack of actionable data. However, in our interactions with experts during the conference, it became evident that non-conventional data is starting to lead to a data-deluge, which means, there is more data (i.e. digital crumbs) available than ever before. This would also mean that the traditional approach to agri-food systems research will have to adapt in this context, in a manner where new research questions need to be built based on the available data; this is unlike the traditional approach of first developing a research question and then collecting data. Our presentation and interactions in the conference were particularly useful, because it helped to outline new research questions based on our preliminary data analysis across different projects, and forge new partnerships for collaborative projects.
Specifically for climate smart agriculture, which is essential for a sustainable agri-food system, we recognised that non-conventional data sources, in combination with various other data sources, have the potential to provide new insights in relation to adoption and scaling up sustainable agricultural production practices. This is because the adoption of a particular practice or intervention relates to behavioural aspects of farmers, which is aptly suited for study using non-conventional data sources. The experts who were part of the Synthesis Panel at the conference, while summarising the outcomes of various sessions of the conference, emphasised that digital and smartphones are the way forward.
This is a valuable piece of insight because mobile phones are ideal data collection points that enable capturing of digital crumbs, which can be further used to study farmer behaviour. In addition, the panellists also stressed the need to better understand the customer (e.g. smallholder farmers), to provide customised solutions and to further facilitate adoption of sustainable agricultural production by smallholder farmers. One approach to understand the customer is through their social networks, and we are currently exploring key characteristics and differences in social networks of farming communities, and changes across time and space, using call detail records (CDRs). We’ll be sharing more on this exploration as we move forward, stay tuned.
Dharani Dhar Burra is a visiting scientist from CGIAR Platform for Big Data in Agriculture. He has been with us at Pulse Lab Jakarta on secondment and is working on projects that will use non-conventional data sources to better understand, and derive new insights about the complex relationships between agriculture production, their impacts on the environment, farmer incomes and food consumption.
Pulse Lab Jakarta is grateful for the generous support from the Government of Australia
