This peer learning session featured ignitia and Precision Development (PxD), spotlighting cutting-edge approaches to weather forecasting for smallholder farmers. The session explored forecast accuracy, user behavior change, and the economics of delivery channels like USSD and IVR.
Andrew Lala, CEO of Ignitia, shared how their weather platform, designed for fast-evolving tropical weather systems, delivers hyper-local rainfall and seasonal forecasts at 1–9 km resolution. Built using physics-informed machine learning models trained on 15+ years of reforecast and ground-truth data, Ignitia’s system enables high-accuracy nowcasting, 14-day forecasts, and six-month rainfall anomaly predictions. Forecasts are delivered via SMS, WhatsApp, IVR, and dashboards, with accuracy metrics surpassing traditional global models.
Madhav Vaidyanathan, Senior Project Manager, and Anjaney Singh, Research Manager at PxD, presented PxD’s collaboration with the University of Chicago’s Human-Centered Weather Forecasts Initiative and the Government of India. The project used ensemble forecasting and statistical post-processing to generate actionable probabilistic rainfall predictions. PxD delivered these forecasts via localized voice and SMS messages in 11 Indian languages. Their approach emphasized behavioral design to build trust and comprehension, resulting in adoption and impact on farm decision-making.
Resources
Supporting Readings:
- PxD initiative page on weather forecasts
- U Chicago – Human-Centered Weather Forecasts Initiative post
- Shocked podcast series Episode 6: Farmer’s A.I.manac
- The Economist article AI models ace their predictions of India’s monsoon rains
- Bloomberg article featuring ignitia and detailing the state of ai weather forecasting and ongoing challenges globally
- UN Early Warnings for All feature of ignitia among other top innovators for Ai forecasting
- 60 decibels impact studies on ignitia’s impact on smallholder farmers showing improved yields, climate resilience, and other assessed changes including within bundled services to farmers
- Benin published study relative to control which “showed that ignitia farmers recorded higher yields than non-Ignitia farmers, with a difference of 30% for peppers, 22% for maize, 20% for soybeans, 17% for peanuts and 13% for cassava”