- CalibSun | Scientific publications
Scientific publications
CalibSun is a spin-off of Solaïs, a solar engineering consulting firm with over 18 years of expertise in the photovoltaic sector, which allows us to have invaluable practical knowledge of PV financing, development and operations.
- Forecasting
- 2025
On the optimal selection of meteorological variables as inputs to machine learning models for solar irradiance forecasting
Variables output from numerical weather models are essential and widely used as inputs to machine learning models for forecasting the solar irradiance available at ground level. This abundance of data necessitates an effective optimization strategy to identify the most relevant features and eliminate those with limited impact in order to reduce computational resources.
Filter by
- Forecasting
- 2022
Cleaning Operations Planning Induced by Monitoring and Modelling of Soiling on Photovoltaic Modules
Dust and soiling degrade the yields of desert-based or industrial solar plants. The authors propose a method to calibrate DustIQ optical sensor measurements across multiple sites. They introduce a decision-making algorithm that uses weather forecasts to schedule module cleanings when they are most cost-effective.
- Forecasting
- 2022
Towards a New Approach for Practical Economic Performance Evaluation of a Non-Grid Connected Hybrid Energy System Without Storage with Integrated Solar Forecasting
For remote, off-grid systems (such as mine sites), using energy storage systems (BESS) increases capital costs. This work presents a Power Management System (PMS) that leverages highly precise, short-term solar forecasts to operate PV/diesel hybrid configurations without storage. It introduces an operational cost-based metric that replaces traditional metrics like RMSE or MAE.
- Probabilistic
- 2021
A New Approach for Probabilistic Solar Forecasting Using Satellite-Based Cloud Motion Vectors
This paper introduces a method combining physical and statistical techniques. It uses clear-sky index data and Cloud Motion Vectors (CMV) from satellite feeds (traditionally used only for deterministic models) to generate precise probabilistic forecasts. Tested on ground-measured GHI data, the model reduces the Continuous Ranked Probability Score (CRPS) by 37% to 62% against standard baselines.
- Solar resource assessment
- 2021
Characterization of Convergence and Robustness of the Kernel Density Mapping Method for Site Adaptation of Global Horizontal Irradiation in Western Europe
Focuses on adjusting historical satellite databases (Solargis, HelioClim-3, CAMS Radiation) with on-site pyranometric measurements. By using a Kernel Density Mapping (KDM) algorithm, the systematic biases of satellite databases are reduced from ±3% to 0.5% across the Western European sites studied, increasing financing accuracy.
- Probabilistic
- 2020
A Novel Approach for Seamless Probabilistic Photovoltaic Power Forecasting Covering Multiple Time Frames
This research adapts the Analog Ensemble (AnEn) model to generate continuous, transparent probabilistic forecasts across multiple time horizons. The algorithm can start at any point of the day and adaptively selects the most relevant inputs for each horizon, making it ideal for continuous intraday trading.
- BESS
- 2020
Combined Operation Strategies of PV – Storage Systems Integrated into Electricity Markets
Building on previous work, this paper details control strategies for PV/BESS installations in electricity markets. It introduces a Model Predictive Control (MPC) model that tracks battery degradation costs at both the day-ahead and real-time dispatch stages to prevent premature battery wear.
- Probabilistic
- 2020
Towards a Simple and Robust Probabilistic Analysis of Solar Variability Based on the Probability of Transition Between Sky State Variability Classes
Solar variability is a key driver of grid instability in isolated or hybrid (PV/Diesel) power plants. This study models local solar variability by analyzing the mathematical transition probability between sky states, helping developers assess the feasibility and predictability of new hybrid installations.
- Forecasting
- 2019
An Integrated Approach for Value-Driven Energy Forecasting and Data-Informed Decision-Making: Application to Renewable Energy Trading
Rather than using a two-step approach (forecasting first, then optimizing dispatch), this paper unifies the forecasting and decision-making steps into a single function. The entire predictive engine is optimized to maximize market revenue.
Scientific partners
Our authors
CalibSun was created in 2023 by Nicolas Thévenin and Sébastien Pitaval, the founders of Solaïs, a company that has been an expert in photovoltaics for more than 18 years and a world leader in glare studies.
- CalibSun’s resources
Our resources
Solar & data expertise
Deep-dive our specific know-hows on photovoltaic, solar resource and data science and how they enhance our services.
