Ground State
Ground State helps a national meteorological service decide when to act on a drought forecast. More than eighty countries deal with chronic drought, and seasonal forecasts already exist for most of them. But a forecast is not a decision. Deciding when to act on one is hard, especially for a slow-onset disaster like drought, where resources have to be committed before there is anything to see. That decision depends on a threshold, called a trigger, agreed in advance: if the forecast crosses it, funds are released. Calibrating a trigger takes several organisations, decades of historical data, and a judgment about how many false alarms a country will accept in exchange for catching real droughts earlier.
Doing that well asks for climate science, statistics, food security expertise and local operational knowledge in the same room, which is a rare combination to assemble and a harder one to keep available. Most of the work still sits in spreadsheets and analysis scripts. We built Ground State with the World Food Programme's Geospatial Intelligence Unit and Mozambique's National Institute of Meteorology, and designed it to be released as a digital public good.
What the software does
A district can have hundreds of thousands of possible triggers. The meteorologist narrows them by saying what a usable one has to do (which forecast window it reads, a minimum hit rate, a maximum false alarm rate) and the ranked candidates update as the sliders move. Sometimes nothing clears the thresholds, so they come down until every window and every drought severity in that district has a trigger. Then the same pass runs for the next district, and the next.
In Maputo
We ran working sessions at INAM in Maputo, watching meteorologists use the tool on their own machines, with their own data, in the middle of their own week. Those sessions settled things we would have got wrong from a distance: which steps happen in one sitting and which get picked up days later, what a forecaster already knows without being told, where a screen assumed a resource nobody had. We work this way on everything, as close to the people doing the job as we can get and as early as we can get there.
We are actively seeking funding for Ground State. If you're interested in using it, or in supporting the work, get in touch.