Photo credit: Neil Palmer/CIAT

As the open data movement began to pick up momentum almost a decade ago, and we were just beginning to populate the CELL5M database, HarvestChoice was a mere neophyte compared to the likes of FAO and World Bank. Our data paradigm, indeed our contribution to the movement, would center on a particular sweet spot in spatial analyses: the mesoscale.

The enormous mix of farming systems throughout Africa South of the Sahara (SSA) is underpinned by a suite of local conditions ranging from the agro-ecological to the socio-economical. Too coarse-resolution data disregards this spatial variability. Moreover, policymakers need disaggregated data for on-target decision making within relevant domains, such as by agro-ecological zone, development corridor, subnational administrative unit, or watershed, to name a few.

To be sure, national statistics are critical for fingering the pulse of human welfare at country and global levels. Nevertheless, like an enlarged snapshot that looks absurdly distorted, statistics also rely on resolution. The higher the resolution, the better the picture.

Enter the pixel and the HarvestChoice CELL5M database. HarvestChoice’s grid system includes over 350,000 pixels blanketing the SSA region. At 5 arc-minute resolution, each pixel is ~10 kilometers x 10 kilometers and stacked with over 750 geospatial data layers (and counting). Layers include data on agricultural production and farming systems, agroecology, market access, health and nutritional outcomes, and income and poverty. Freely available literally at your fingertips, stakeholders and researchers can tap into the CELL5M, aggregate and overlay cross-cutting indicators, create data visuals including maps, download shape and CSV files, or plug into an open API for more advanced analyses.

Once you set data free into the wild of the public realm, however, it can be difficult to monitor their impact. What indicators are appropriate for reporting on the use of … well, other indicators? Obviously, applications like Google Analytics are useful for monitoring traffic and downloads, but in terms of where open data eventually entangle themselves, like a tree branch floating downriver, is often anybody’s guess. (Therein lies the challenge when seeking continued funding for open data products!)

Since 2010, we have documented over 120 publications on Google Scholar with permanent DOIs that have cited HarvestChoice datasets in their analysis (click here for bibliography). HarvestChoice data products are reaching users outside institutional walls and underpinning global challenges such as climate change, food security, poverty, and nutrition. The most popular products are agricultural data generated from the Spatial Production Allocation Model (SPAM); at least 71 papers have cited SPAM. Agroecology data were also frequently referenced, accounting for over 1/3 of citations, followed by data on market access, demographics, and administrative boundaries.

HarvestChoice has come of age during a data revolution when products are beginning to flood the market. Technological innovations in remote-sensing satellite imagery, drones, econometrics, modelling, and crowd-sourcing, as well as micro-level data collected from expansive household surveys (e.g., the Demographic and Health Surveys (DHS) program), are further advancing the data frontier. And with the Sustainable Development Goals (SDGs) already a year old, policymakers and the analysts who support them critically need spatially-explicit, open source data products perhaps more than ever before.

Visit to access data, maps, tools, and publications.


HarvestChoice, 2016. "Meandering Downstream: Tracking HarvestChoice Data ." International Food Policy Research Institute, Washington, DC., and University of Minnesota, St. Paul, MN. Available online at

Aug 11, 2016