Meteorological model data

What meteorological model data Solargis provides

Solargis sources air temperature, wind, humidity, precipitation, snow, atmospheric pressure and UV radiation from global meteorological models, then post-processes them to resolutions as fine as 1 km for use in PV simulation, monitoring and forecasting. Two model families are used: operational numerical weather prediction (NWP) models for forecasts covering the next hours to days, and reanalysis models for consistent historical data spanning several decades. This article describes both families, the models Solargis ingests from each, and the postprocessing applied before the data reaches Solargis products.

Global meteorological models simulate the atmosphere's behavior at the global level using mathematical modelling. They are the only practical way to characterize the meteorological and environmental conditions required for solar energy applications at any project site, over both short- and long-term horizons.

Operational NWP models and reanalysis models compared

Solargis uses operational NWP models for the future and reanalysis models for the past.

For short-term forecasting, numerical weather prediction (NWP) models generate weather predictions for the next few hours and days. Solargis Forecast uses these predictions to support real-time operations and immediate decision-making, enabling operators to plan and schedule maintenance activities, manage energy storage, and efficiently balance supply and demand. In this context, these models are typically referred to as operational NWP models.

For long-term historical analysis, different meteorological models are used to provide reanalysis data: a comprehensive dataset that combines historical observational data into global model outputs. Solargis Prospect, Solargis Evaluate and Solargis Monitor use this reanalysis data to offer historical and recent data services, giving analysts accurate and detailed information about past weather conditions and more reliable assessments for solar projects.

Figure 1: Map representation of precipitation data from global reanalysis models

Meteorological parameters affecting PV assets

Solargis supplies meteorological and environmental parameters for three distinct purposes: PV performance simulation, PV risk and degradation analysis, and as inputs to other Solargis models. Photovoltaic (PV) assets are influenced by changing site conditions, which is why each of these uses depends on a different set of parameters.

Parameters used in PV performance simulation

Solargis meteorological data is critical for simulating the expected performance of photovoltaic systems. Air temperature (TEMP) and wind speed (WS) are key factors in determining the PV cell’s temperature, which directly affects its conversion efficiency. Relative humidity (RH) and precipitable water (PWAT) are necessary for evaluating the spectral response of the cells, while precipitation (PREC) and snow depth water equivalent (SDWE) are used to estimate energy losses caused by environmental factors – see Snow loss model.

Parameters used in PV risk and degradation analysis

Solargis meteorological parameters are also used for risk analyses of energy assets. Ultraviolet radiation (UVA, UVB) and thermal cycling driven by extreme air temperatures (TEMP) are important for assessing the risk of accelerated degradation in PV modules. The physical integrity of solar installations is further threatened in areas prone to severe weather, which requires characterization using parameters such as wind gusts (WG) and other relevant data.

Parameters used as inputs to other Solargis models

In addition to performance and risk considerations, certain meteorological and environmental parameters serve as inputs to other Solargis models. Information on ozone content, water vapor (WV), and aerosol optical depth (AOD) is necessary for running semi-empirical solar irradiance models, while NWP-based solar irradiance is crucial for solar power forecasting. Other parameters feed the Solargis ground albedo model (see Satellite-based albedo data) and the Solargis soiling loss estimates (see Soiling loss model), both of which depend on accurate and comprehensive meteorological data.

Operational NWP data in Solargis Forecast

Solargis Forecast is driven by operational Numerical Weather Prediction (NWP) forecasts, which provide real-time information about atmospheric conditions from hours to days ahead. This data is used in Solargis Forecast services, and to a small extent in Solargis Monitor.

How operational NWP forecasts are produced

Operational NWP forecasts used by Solargis are produced in two stages:

• Real-time atmospheric data is collected from a wide variety of sources including weather stations, satellites, aircraft, buoys, and radiosondes (weather balloons).

• This raw data is then assimilated into a coherent format suitable for numerical models. The assimilated data is used to initialize the NWP model. Using mathematical equations based on the laws of physics (such as fluid dynamics and thermodynamics), the model simulates the future state of the atmosphere. The Earth’s atmosphere is divided into a three-dimensional grid, and calculations are made for each grid point. The results are produced at regular intervals (e.g., every 6 hours) and for various lead times (e.g., 12 hours, 24 hours, 7 days ahead).

Operational NWP models used by Solargis Forecast

Table legend:

• D+0 means the same day when forecasts are delivered.

• D+n represents predictions for n days after delivery.

Reanalysis meteorological data in Solargis Prospect, Evaluate and Monitor

Meteorological model reanalysis recreates past atmospheric conditions over a long historical period by integrating past observational data with a fixed version of the meteorological model. Solargis Prospect, Solargis Evaluate and Solargis Monitor all use reanalysis data.

How meteorological reanalysis is produced

Reanalysis datasets used by Solargis are produced in two stages:

• Historical weather data is collected from multiple sources, including surface observations, satellite records, and archived meteorological data. Since historical data can have gaps and inconsistencies, it undergoes rigorous quality control and preprocessing to correct errors and standardize the data. Similar to operational forecasts, this historical data is assimilated into the global model. The key difference here is that the same version of the model and assimilation system is used throughout the entire reanalysis period to ensure consistency.

• The meteorological model is run retrospectively using the assimilated historical data, producing a comprehensive dataset that represents the state of the atmosphere over time. This step uses a fixed set of model physics and parameters to avoid inconsistencies. The reanalysis generates long-term datasets that provide continuous, gridded representations of various atmospheric parameters over several decades.

These datasets include key parameters for PV solar energy applications like temperature, wind (speed and direction), relative humidity, precipitation, precipitable water, water equivalent of accumulated snow depth, ultraviolet radiation, and snow density.

Reanalysis models can also supply historical solar irradiance data, but their results have lower resolution and accuracy than satellite-based models. Solargis therefore derives historical solar irradiance from satellites, not from reanalysis – see Solargis satellite-based model.

Reanalysis and NWP models used by Solargis Evaluate and Monitor

Table legend:

• D-n represents days before the present time of delivery.

Postprocessing of Solargis meteorological model data

Solargis post-processes raw meteorological model output to refine its spatial resolution before the data is used in solar energy applications. The original resolution of meteorological models, typically 0.1° to 0.25°, represents broad geographic regions rather than specific sites.

Spatial downscaling of air temperature with lapse rate and SRTM-3

Solargis downscales air temperature to approximately 1 km resolution by applying a calculated lapse rate together with the SRTM-3 Digital Elevation Model. This improves accuracy and consistency for long-term analyses, particularly in areas of complex terrain.

Weather models provide not only surface-level temperature data (e.g., at 2 meters above ground) but also vertical temperature profiles across multiple atmospheric layers, from the surface to the top of the atmosphere. By analyzing these profiles, a simplified parameterization of temperature variation with altitude is developed. Generally, temperature decreases with increasing altitude, but temperature inversions, where temperature rises with altitude, are common near the surface, particularly over cold surfaces.

To account for vertical temperature changes, the lapse rate is calculated, representing the rate of temperature change with altitude. This lapse rate varies across time and locations, influenced by weather patterns, local microclimates, and topographic features. By applying the lapse rate, temperature data are spatially downscaled to a finer resolution, ensuring greater accuracy for regions with complex terrain. Using the SRTM-3 Digital Elevation Model, the spatial resolution of Solargis air temperature data is refined to approximately 1 km, significantly improving its precision for localized applications. For how Solargis uses digital elevation models more broadly, see Terrain models.

Secondary parameters derived from model outputs

Some Solargis meteorological parameters are not taken directly from the models but derived from them. Derived parameters include UVA and UVB radiation, dew point temperature (TD), wet bulb temperature (WBT), true accumulated snow depth (TSD), and cooling and heating degree days (CDD, HDD). These additional parameters are critical for specialized solar energy analyses and applications.

Meteorological parameters and resolutions after postprocessing

The table below lists each meteorological parameter Solargis delivers, together with its unit, its native model resolution, its resolution after Solargis postprocessing, and the models it is derived from.

How recent Solargis meteorological data is updated (DAY-1 to DAY-3)

Solargis meteorological data for DAY-3 and earlier can be considered definitive. DAY-1 and DAY-2 values are taken from NWP-based forecasts so that recent data stays up to date, and are later replaced with values from the re-analyzed archive.

Note: Because DAY-1 and DAY-2 Solargis meteorological values come from forecasts rather than reanalysis, they are provisional and can change once the archive is updated. For the full delivery model behind Solargis Monitor and Solargis Forecast, see Origin of the data for Monitor and Forecast.

Accuracy of Solargis meteorological data

Solargis validates its meteorological model data against high-quality ground measurements at 12,000 to 14,000 sites worldwide. Air temperature (TEMP), wind speed (WS) and relative humidity (RH) are each compared against ground stations across all climate zones, with accuracy reported as bias, standard deviation and RMSE at hourly, daily and monthly resolution.

Across all three parameters, RMSE decreases consistently with temporal aggregation, and accuracy during solar generation hours is higher than at night. Deviations are larger in mountainous and coastal terrain and in regions with sparse input data for NWP models, where large-scale models cannot capture local microclimates. For the full statistics per parameter, see Validation of meteorological parameters.