Real-time Analytics
The Israel Meteorological Service needed to produce faster and more localized weather forecasts for floods, fire seasons, and other severe conditions, but conventional numerical weather prediction required large CPU clusters and hours of runtime that limited operational responsiveness.
Reported impact
IMS cut compute costs by 90 percent, delivered high-resolution forecasts in minutes rather than hours, achieved 2.5-kilometer hyperlocal resolution, and helped the Tel Aviv-Yafo Municipality save 20-30 percent of its preparedness budget while giving agencies faster flood-response guidance.
IMS built a sovereign AI forecasting stack around NVIDIA Earth-2 to create F3, a high-resolution precipitation-prediction system tailored to Israel. The service uses NVIDIA Earth-2 Corrector Diffusion (CorrDiff), a neural network model trained to emulate IMS's regional high-resolution ICON forecasts while taking lower-resolution input from ECMWF IFS. NVIDIA PhysicsNeMo supports the Earth-2 modeling workflow, and the combined software stack runs on GPUs rather than large CPU clusters, allowing IMS to issue probabilistic, 2.5-kilometer forecasts four times a day. Because the system is operated as a full in-house hardware-and-software capability, IMS can keep the forecasting environment sovereign while delivering fast guidance to municipalities and emergency agencies that need localized flood and storm information within minutes instead of hours.
Low-resolution ECMWF IFS forecasts and local weather context feed CorrDiff inside the Earth-2 stack, the model generates high-resolution probabilistic weather fields, and IMS distributes 2.5-kilometer forecasts to municipal and emergency agencies for planning and flood response.
IMS adopted NVIDIA Earth-2 to develop a high-resolution weather forecasting system specifically designed for precipitation prediction, known as F3.
IMS specifically utilized NVIDIA Earth-2 Corrector Diffusion (CorrDiff), a neural network model that improves the resolution of weather forecasts without sacrificing accuracy.
It has 90 percent lower computational costs compared to non-AI systems while delivering answers in minutes instead of hours.
The team trained CorrDiff to emulate their regional high-resolution ICON model with low-resolution input from the European Centre for Medium-Range Weather Forecasts' Integrated Forecasting System (ECMWF IFS).
They’re now able to issue hyperlocal 2.5-kilometer resolution forecasts.
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
NVIDIA Sample Oct 9, 2026 California Department of Transportation (Caltrans) Caltrans needed to inspect and monitor roughly 1,200 active transportation projects more frequently and safely, but manual site visits were slow, inconsistent, and risky for staff working around traffic, bridges, and… Remote operations up to 100 miles Esri ArcGIS · NVIDIA CUDA · NVIDIA Jetson Orin NVIDIA Sample Jun 29, 2026 Kaohsiung City Government Kaohsiung needed citywide situational awareness across transportation, flooding, disaster response, and public-safety workflows, but departmental silos and traditional computer vision left agencies unable to interpret… 30,000 diverse smart city camera streams integrated NVIDIA Cosmos · NVIDIA DGX · NVIDIA Metropolis Blueprint for Video Search and Summarization NVIDIA Sample Archived · Jan 28, 2026 City of Raleigh Raleigh needed to improve congestion management, mobility, and street safety as the city grew, but manual traffic counts and disconnected sensor repositories made it hard to visualize conditions, respond to incidents… $9.7 million annually in time and fuel costs Esri ArcGIS · Microsoft Azure · NVIDIA Blueprint for video search and summarization (VSS) 