ABOUT
For the next hours, for intraday forecasting need
SteadySat provides advanced solar irradiance and production forecasts up to 6 hours ahead. This product combines real-time satellite imagery with in-situ observations processing, weather forecast, physical models and artificial intelligence.
Cloud properties and evolution are monitored and predicted in real-time, improving the accuracy of solar forecasts for the next few hours and anticipating the risks of variability.
Intraday power forecast for a 12MWp PV plant (Germany)
Intraday power forecast for a 17MWp PV plant (Australia)
Satellite-based GHI forecast above Western Europe
6 hours
Max time horizon
5 - 15 minutes
Update frequency
1 min
Forecast time-step
Power, GHI, DNI, DHI, GTI, Temperature, Wind
Available parameters
Site, Portfolio, City, Region or Country
Coverage
PV, Trackers, Bifacial, CSP
Technology
API, SFTP, etc.
Data delivery
P10, P20,… , P80, P90
Confidence levels
KEY BENEFITS
WORLDWIDE COVERAGE
Real-time process of images coming from five geostationary meteorological satellites.
ACCURACY IMPROVEMENT
Frequently updated information of current and upcoming weather situations. Local weather effects are more accurately predicted at specific locations with nowcasting. SteadySat perfectly complements SteadyMet with precise intraday forecasts thanks to high-resolution remote sensing observations.
TAILORED OUTPUTS
In terms of parameters, update frequency, granularity and delivery settings.
CUSTOMERS NEEDS
This product is tailored to meet your solar business needs. It is particularly suitable for the optimization of storage dispatch and intraday trading markets.
CUSTOMERS NEEDS
This product is tailored to meet your solar business needs. It is particularly suitable for the optimization of storage dispatch and intraday trading markets.
METHODOLOGY
SteadyMet is based on an optimal combination of several Numerical Weather Prediction (NWP) models. These models simulate the evolution of the atmosphere (step 2) from initial atmospheric conditions estimated by assimilation of meteorological observation data over a global or local granularity (step 1). For this purpose, the area of calculation is divided into a three-dimensional grid with more or less large mesh.
Weather model’s outputs of interest (solar radiation, temperature, wind, etc.) are then optimized (step 3), by using production/irradiance observations, to take into account local phenomena, and by leveraging artificial intelligence technologies.
Ready-to-use forecasts, in a customized format, are then disseminated (step 4) through our proprietary web interface, csv files delivered via (S)FTP platforms or API.
Step 1
DATA ACQUISITION
From 5 geostationary satellites.
Different channels providing information on the atmosphere, clouds and aerosols.
Every 5 to 15 minutes.
Step 2
MODELING
Cloud detection, classification and motion prediction.
Accurate estimation of clear sky conditions using real-time aerosols prediction.
Calculation direct and diffuse irradiance components in cloudy conditions.
PV modeling based on physical models and plant features.
High-resolution topographical corrections (down to 90m).
Probabilistic forecasting using physical and statistical approaches.
Step 3
OPTIMIZATION
Based on historical and/or real-time on-site measurements.
Continuous accuracy improvements using state-of-the-art machine learning techniques.
To take into account local weather phenomena and power plants’ behavior.
Step 4
DELIVERY
Flexible delivery (API, SFTP, etc..).
Customized format (csv, txt, etc.).
Dedicated and secured web interfaces (visualization, data analytics and warnings).
Forecast performance monitoring.
METHODOLOGY
SteadySat compiles live images from 5 geostationary satellites (step 1) and processes them (estimation of motion vector and application on the last available image) to estimate the clouds position for the next few hours. This cloud position is used to forecast the irradiance (step 2). The satellite imagery also allows us to estimate the global horizontal irradiance at acquisition time (10 to 15 minutes).
Forecasts are then optimized (step 3), by using production/irradiance observations, to take into account local phenomena, and by leveraging artificial intelligence technologies.
Ready-to-use forecasts, in a customized format, are then disseminated (step 4) through our proprietary web interface, csv files delivered via (S)FTP platforms or API.
DATA ACQUISITION
MODELING
OPTIMIZATION
DELIVERY
From 5 geostationary satellites.
Different channels providing information on the atmosphere, clouds and aerosols.
Every 5 to 15 minutes.
Cloud detection, classification and motion prediction.
Accurate estimation of clear sky conditions using real-time aerosols prediction.
Calculation direct and diffuse irradiance components in cloudy conditions.
PV modeling based on physical models and plant features.
High-resolution topographical corrections (down to 90m).
Probabilistic forecasting using physical and statistical approaches.
Based on historical and/or real-time on-site measurements.
Continuous accuracy improvements using state-of-the-art machine learning techniques.
To take into account local weather phenomena and power plants’ behavior.
Flexible delivery (API, SFTP, etc..).
Customized format (csv, txt, etc.).
Dedicated and secured web interfaces (visualization, data analytics and warnings).
Forecast performance monitoring.
