Our work runs along three threads. Each one starts with a physical process and ends with a decision somebody has to make.
Season-ahead forecasting of streamflow and hydroclimatic extremes, satellite observation of surface water, and what that information is worth once a water system has to act on it.
This thread runs from global attribution of peak-flow variability to ENSO, through season-ahead forecasts for hydropower and flood preparedness, to the reservoir-operation question underneath all of it: at what lead time does a forecast stop being interesting and start being useful? Current work brings SWOT satellite observations of surface water into that chain for Manitoba’s hydropower system, and extends the same question to other extremes across the Canadian Prairies.
HarvestStat — harmonized subnational crop statistics — and Earth-observation prediction of crop yields and food security.
The data problem comes before the modelling problem. Crop statistics across Sub-Saharan Africa sit in dozens of national agencies with incompatible boundaries and no common format. We lead HarvestStat Africa, which harmonizes them into a single open dataset, so that satellite-based yield estimates can be validated against something consistent instead of against whatever was at hand. The lab also contributes to community benchmarking efforts for subnational yield forecasting.
Who is exposed, who can adapt, and who gets displaced — linking physical hazard to social and health vulnerability, and to the adaptation policy that follows.
Hazard alone does not determine harm. This thread links physical exposure to social and health vulnerability — flood risk in Bangladesh, geographic isolation in Peru, heat and green infrastructure in the U.S. Great Lakes — and asks which adaptation policies reach the people carrying the risk.