New Approaches to Capture High Frequency Agricultural Dynamics in Africa through Mobile Phones

Abstract:

Crop failure early warning systems relying on remote sensing constitute a new critical resource to assess areas where food shortages may arise, but there is a disconnect between the patterns of crop production on the ground and the environmental and decision-making dynamics that led to a particular crop production outcome. In Africa many governments use mid-growing season household surveys to get an on-the-ground assessment of current agricultural conditions. But these efforts are cost prohibitive over large scales and only offer a one-time snapshot at a particular time point. They also rely on farmers to recall past decisions and farmer recall may be imperfect when answering retrospectively on a decision made several months back (e.g. quantity of seed planted).

We introduce a novel mobile-phone based approach to acquire information from farmers over large spatial extents, at high frequency at relatively low-cost compared to household survey approaches. This system makes compromises in number of questions which can feasibly be asked of a respondent (compared to household interviews), but the benefit of capturing weekly data from farmers is very exciting. We present data gathered from farmers in Kenya and Zambia to understand key dimensions of agricultural decision making such as choice of seed variety/planting date, frequency and timing of weeding/fertilizing and coping strategies such as pursuing off-farm labor. A particularly novel aspect of this work is reporting from farmers of what their expectation of end-season harvest will be on a week-by-week basis. Farmer’s themselves can serve as sentinels of crop failure in this system. And farmers responses to drought are as much driven by their expectations of looming crop failure that may be different from that gleaned from remote sensing based assessment. This work is one piece of a larger design to link farmers to high-density meteorological data in Africa as an additional tool to improve crop failure early warning systems and understand adaptation to climate variability.