At FROGCAST, our weather forecasts leverage cutting-edge technology to deliver a wide range of atmospheric parameters, global coverage, optimal accuracy, and dependable service. Designed today to anticipate tomorrow, FROGCAST empowers you to make informed decisions by covering all your weather-related needs.
FROGCAST aggregates data from over 20 Numerical Weather Prediction (NWP) models provided by national meteorological agencies worldwide.
This comprehensive approach allows us to provide precise and reliable weather forecasts for any location, up to 15 days into the future, by capturing the specific weather characteristics of each region.
In areas that are complex to model and where high-resolution models are unavailable, we also leverage the high-resolution WRF (Weather Research and Forecasting) model, which can analyze local phenomena with increased accuracy (resolution up to 1 km).
All models are carefully pre-processed and then rigorously weighted according to their specificities (spatial and temporal resolutions, time horizons, etc.) and performance.
This multi-model approach allows us to compensate for the individual errors of different meteorological agencies, enabling a better quantification of uncertainty and, consequently, a continuous improvement in the performance of our forecasts.
Our machine learning algorithms continuously update the weighting between models at any point on the globe to track their evolution and ensure the best possible Frogcast forecast.
We continuously collect and process in-situ measurements from various observation sites around the world.
These data include precise readings such as temperature, precipitation, and wind speed, taken directly on-site where weather phenomena occur.
By comparing these real-world measurements with the forecasts generated by our various weather models, we continuously evaluate their accuracy. This evaluation allows us to identify, for each point on the globe and each meteorological variable, the best-performing model.
This analysis is crucial for adjusting the weight / importance we assign to each model in our calculations.
For example, if a model is more reliable in predicting temperatures in a given region, we will give it more weight for this type of forecast.
At FROGCAST, we’re pushing the boundaries of weather intelligence to deliver forecasts that redefine accuracy and relevance. Our unique combination of cutting-edge technology, advanced scientific methods, and proven expertise enables us to provide reliable, actionable, and tailored to the most demanding requirements.
Our forecasts are powered by cutting-edge Big Data technologies, allowing us to collect, store, access, process, and visualize vast amounts of meteorological data.
FROGCAST leverages a wide array of geospatial data from over 20 world-renowned weather providers such as ECMWF and NOAA, combined with numerical weather prediction (NWP) models, satellite observations, and ground-based sensor readings.
Our team, backed by Big Data technologies, harmonizes these extensive datasets to generate consistent and reliable forecasts. Additionally, we archive years of all data produced by FROGCAST, allowing us to perform detailed and robust evaluations to ensure the continuous improvement of our models.
For more specific needs, FROGCAST uses a regional model called WRF (Weather Research and Forecasting), capable of achieving resolutions as fine as 1 km.
This model is optimized for complex areas (terrain, coastal zones, aerosols), delivering remarkably accurate local data.
It allows us to analyze local phenomena in detail, significantly improving the precision of our forecasts.
Uncertainty is inherent characteristic of weather forecasting. Our solution addresses this limitation by providing probabilistic forecasts, offering a more nuanced and detailed view of future conditions.
For each meteorological parameters, FROGCAST includes a set of 11 quantiles in addition to the average forecast. The level of dispersion between these quantiles highlights differences between weather scenarios provided by NWP (Numerical Weather Prediction) models. By estimating the probability of an event occurring, Frogcast helps you make better-informed decisions and manage weather-related risks more effectively.
The growing availability of high-quality atmospheric data enables FROGCAST to integrate machine learning techniques into its algorithms, particularly for:
no credit card required | no commitment
FROGCAST provides a comprehensive range of meteorological weather parameters, tailored to meet the specific requirements of various industries. Whether you operate in energy, agriculture, logistics, or infrastructure management, you will find the data that suits your needs.
Before
Generic and/or imprecise forecasts, often unsuitable for your operations.
With FROGCAST
Optimal combination of some twenty numerical weather prediction models. Forecasts up to 2.6 times more accurate than individual models from national meteorological agencies.
Before
Frequent discrepancies between forecasts and actual on-the-ground conditions.
With FROGCAST
Data from reputable and verified sources. Solution used at over 14,000 sites worldwide.
Before
No consideration of uncertainty: static forecasts, often inaccurate in critical situations.
With FROGCAST
Integrate probabilistic forecasts: confidence intervals built from 11 quantiles per weather variable help you objectively quantify uncertainty and make more informed decisions.
Before
Laborious API integration and often non-standardized data formats.
With FROGCAST
FROGCAST API requires no development. It follows REST standards, uses the JSON format to encode objects, and relies on HTTPS for security. Integration in just a few clicks.
Before
Approximations due to low granularity of topographic features.
With FROGCAST
Post-processing of altitude-related parameters using 90m topographic data for optimal precision in capturing local variations.
Before
Incomplete and low-resolution forecasts made precise decision-making difficult.
With FROGCAST
Access global forecasts up to 15 days ahead, updated 8 times daily, with high-resolution data down to 1 km using the WRF model.
English English 
French
Get quarterly insights into the energy sector and our latest products.
