Ninety days ahead of hunger.

Weekly forecasts of food crisis risk, 90 days ahead, for 43 countries.

Sentinel-2 cloudless 2016 by EOX IT Services GmbH, contains modified Copernicus Sentinel data 2016, CC BY 4.0

The challenge

By the time a food crisis is declared, the harvest that caused it failed months earlier.

Rainfall, vegetation, food prices and conflict all move before people go hungry. The signal is there, but it reaches decision makers late.

CERES reads those signals every week and turns them into a forecast for the next 90 days. Every forecast is logged before the outcome is known and scored after.

What we deliver.

  • Forecast the risk of a food crisis 90 days ahead, for 43 countries, every week.
  • Read satellite rainfall and vegetation together with food prices and conflict.
  • Log every forecast before the outcome and score it against what happened.
  • Record the exact version of the code behind every forecast.
  • Publish the forecasts as open data and the method on arXiv.

How it works.

Read
Rainfall from CHIRPS, vegetation from MODIS, and prices and conflict from open sources.
Forecast
A probabilistic model gives the risk for each region 90 days ahead.
Score
Every forecast is held to the same targets as the simplest alternative.