PVsyst 8.1 Meteorological Data Masterclass: Every Source Compared and Tested (2026)

Introduction

Here is one of the most common questions solar engineers ask: “I ran the same system design in PVsyst with two different meteorological sources and got E_Grid results that differ by 9%. Which one is right?”

This question gets to the heart of solar energy simulation. The meteorological data fed into PVsyst — primarily Global Horizontal Irradiance (GHI), Diffuse Horizontal Irradiance (DHI), ambient temperature, and wind speed — directly determines the accuracy of every number in your simulation report. If your input GHI is 5% higher than reality, your E_Grid will be approximately 5% higher than reality. On a 50 MW project, a 5% GHI overestimate translates to several million dollars of error in projected annual revenue.

This tutorial covers every meteorological data source available in PVsyst 8.1 in complete technical detail: what each source is, how it works, how to import it, where it performs well and where it struggles, and — crucially — how to make sense of it when sources disagree. We also cover advanced topics including inter-annual variability assessment, building a custom TMY from measured data, and the P50/P90 uncertainty framework that underpins bankable energy yield reports.

💡 Prerequisite: This is Tutorial 2 in our PVsyst 8.1 series. If you haven’t yet run your first simulation, start with PVsyst 8.1 Grid-Connected Simulation: The Complete Engineer’s Walkthrough first.

---

Why Meteorological Data Quality Matters So Much

Before comparing sources, it’s worth understanding exactly how meteorological data flows through a PVsyst simulation — and therefore how input errors propagate.

The PVsyst Calculation Chain

At every hourly time step, PVsyst executes the following sequence:

1. Read GHI and DHI from the meteorological file
2. Apply a decomposition model (Erbs, Reindl, or Perez) to estimate DNI from GHI and DHI
3. Apply a transposition model (Perez, Hay-Davies, or isotropic) to convert horizontal irradiance to plane-of-array (POA) irradiance on the tilted module surface
4. Apply the horizon shading factor for the current sun position
5. Apply near shading, soiling, and IAM losses
6. Calculate cell temperature using the thermal model (ambient temperature + irradiance + wind speed)
7. Calculate DC power from the one-diode model at actual cell temperature and irradiance
8. Calculate AC output through the inverter efficiency model
9. Record E_Grid for this time step

The meteorological data feeds steps 1, 6, and (via the sun position algorithm) step 4. An error in GHI cascades through decomposition, transposition, thermal model, and power model — it is never corrected downstream. This is why GHI accuracy is the single largest determinant of simulation accuracy.

Three Fundamental Sources of Error in Climate Data

Ground Station Measurement Error: Pyranometers at well-maintained meteorological stations are the most accurate source of irradiance data, with typical measurement uncertainty of 2–3% for Class A instruments. However, ground station networks have limited spatial coverage — for most of the world’s land area, the nearest high-quality pyranometer station is tens to hundreds of kilometers away.

Interpolation Error: Sources like Meteonorm interpolate from nearby ground stations to produce data for locations between stations. The further your site is from reference stations, the larger the interpolation uncertainty. In densely-measured regions (Western Europe, Japan), Meteonorm interpolation errors are typically small. In sparse-measurement regions (parts of Africa, Central Asia, remote deserts), they can be significant.

Satellite Retrieval Error: Satellite-based sources (PVGIS, NASA NSRDB, Solcast) use imagery from meteorological satellites to estimate surface irradiance. The retrieval accuracy depends on the satellite coverage of your region, the cloud detection algorithm, atmospheric correction quality, and the spatial and temporal resolution of the satellite data. For clear-sky arid sites, satellite-derived GHI is typically accurate to within 3–5%. For frequently overcast or hazy sites (humid tropics, industrial regions), uncertainties are higher.

---

All Meteorological Sources in PVsyst 8.1 — Complete Reference

PVsyst 8.1 natively supports eight meteorological data pathways. Here is a complete technical analysis of each.

---

1. Meteonorm 8.2

What It Is

Meteonorm is a global climate database developed by Meteotest, a Swiss company. Version 8.2, bundled with PVsyst 8.1, draws on data from more than 8,000 ground-based meteorological stations worldwide, supplemented by satellite data for regions with sparse station networks.

How It Works

Meteonorm operates in two stages. First, it interpolates monthly mean climate statistics (GHI, temperature, wind, humidity) from nearby reference stations to your target location using a combination of Kriging interpolation and physical correction factors (elevation, terrain roughness, distance from coast). Second, it uses a stochastic hourly generator to synthesize an 8,760-hour TMY file that statistically reproduces the interpolated monthly statistics while maintaining physically realistic hour-to-hour variability.

The reference period for Meteonorm 8.2 is 1991–2020 — a 30-year climatological normal period aligned with the WMO standard.

Strengths:

• Fully offline — data is bundled with the PVsyst installation, no internet required
• Always includes ambient temperature and wind speed — essential for the thermal model
• Industry standard — accepted without question by independent engineers and lenders globally
• Good coverage in well-measured regions — Western Europe, North America, East Asia, urban centers worldwide

Weaknesses:

• Interpolation errors increase with distance from reference stations
• The stochastic hourly generator introduces some randomness — two runs with the same Meteonorm source for the same location can give slightly different hourly sequences (though annual totals are stable)
• May not fully reflect recent climate trends if your region has experienced significant irradiance change since 1991

Best Use Cases: Primary source for sites in well-measured regions. Always include as one of the two sources in a dual-source simulation due to its industry-standard status.

Step-by-Step Import:

1. Open your project and click Site in the left navigation panel
2. Set your location using the interactive map or enter GPS coordinates directly
3. Enter altitude above sea level — do not leave this at 0
4. In the Meteo database dropdown, select Meteonorm 8.2
5. Click Import — data loads in under 5 seconds with no internet required
6. Review the monthly GHI bar chart that appears. Verify the seasonal pattern and annual total look reasonable for your site

---

2. PVGIS-TMY 5.2

What It Is

PVGIS (Photovoltaic Geographical Information System) is a project of the European Commission’s Joint Research Centre (JRC) in Ispra, Italy. Version 5.2, supported in PVsyst 8.1, uses SARAH-3 satellite data from the Meteosat satellites (covering Europe, Africa, the Middle East, and parts of Asia) and ERA5 reanalysis data for the rest of the world to generate TMY files at hourly resolution.

How It Works

PVGIS 5.2 retrieves satellite imagery from Meteosat and applies a physical irradiance retrieval algorithm (Heliosat-4 for Meteosat-covered regions) to estimate surface GHI and DNI at each satellite pixel. These satellite-derived irradiance values are then used to construct a TMY by selecting representative months from a multi-year archive (2005–2020 reference period for PVGIS 5.2).

Strengths:

• Free — no subscription required
• Excellent coverage in Europe, Middle East, and Africa — Meteosat satellites provide high-quality coverage for approximately 60% of global landmass, and these are precisely the regions with the best solar resources
• Regularly updated — JRC releases new versions incorporating extended data archives
• Higher spatial resolution than Meteonorm in regions far from ground stations — satellite pixels are 3–5 km across, while Meteonorm interpolation can span 50+ km in sparse regions
• Reference period 2005–2020 — more recent than Meteonorm, better captures recent climate patterns

Weaknesses:

• Requires internet connection for PVsyst to download data
• Coverage quality degrades at high latitudes (above 65°N or below 65°S) where Meteosat viewing angle degrades
• ERA5-based retrieval used for the Americas and parts of Asia is less accurate than the Meteosat-Heliosat retrieval

Best Use Cases: Primary or co-primary source for sites in Europe, Africa, Middle East, and South/Southeast Asia. Essential as second-source cross-check everywhere. For sites in Meteosat coverage area, PVGIS often outperforms Meteonorm in absolute accuracy.

Step-by-Step Import:

1. In the Site window, select PVGIS-TMY 5.2 from the Meteo database dropdown
2. Ensure your computer has an active internet connection
3. Click Import — PVsyst sends a request to the JRC PVGIS API and downloads the TMY file (typically 10–30 seconds depending on server load)
4. Review the monthly GHI chart and compare with your Meteonorm result

Manual Download Alternative: If internet access is unavailable from within PVsyst, you can download PVGIS data manually from pvgis.ec.europa.eu, selecting TMY output in CSV format, then import via Tools → Meteo files → Import.

---

3. NASA NSRDB v4

What It Is

The National Solar Radiation Database (NSRDB) is maintained by the National Renewable Energy Laboratory (NREL) and uses data from NOAA’s GOES satellites. Version 4, integrated in PVsyst 8.1, uses the Physical Solar Model v4 (PSM v4) for irradiance retrieval — an improvement from the PSM v3 used in earlier PVsyst versions.

How It Works

PSM v4 applies a physics-based atmospheric radiative transfer model to GOES satellite imagery to estimate surface irradiance at 2 km spatial resolution and 30-minute temporal resolution (degraded to hourly for PVsyst import). The model explicitly treats clouds, aerosols, water vapor, and ozone.

Strengths:

• Highest accuracy in North and South America — GOES satellite coverage is optimized for these regions
• Free
• 2 km spatial resolution for the Americas — the best available spatial resolution from any free source
• Explicitly models aerosol optical depth — important for sites in polluted regions or areas affected by wildfire smoke

Weaknesses:

• Coverage degrades significantly outside the Americas — uses lower-quality data sources for Europe, Asia, and Africa
• Requires internet connection
• 30-minute native resolution may miss short-duration cloud transients important for battery storage simulation

Best Use Cases: Primary source for US, Canadian, and Latin American projects. Secondary source everywhere else. For global projects, do not use as the primary source outside the Americas.

Step-by-Step Import:

1. Select NASA-SSE from the Meteo database dropdown
2. Ensure internet is connected
3. Click Import (30–60 seconds typical download time)

---

4. Solcast TMY

What It Is

Solcast is an Australian company that combines high-resolution satellite imagery from multiple geostationary satellites with machine learning post-processing to produce irradiance estimates that are continuously validated against ground measurements.

How It Works

Solcast uses data from Meteosat (Europe/Africa/Middle East), Himawari (Asia-Pacific), GOES-East and GOES-West (Americas) to achieve near-global coverage. Their algorithms are continuously trained and validated against a network of ground pyranometers, and they publish routine accuracy statistics.

Strengths:

• Highest absolute accuracy of any commercially available source — typical GHI bias <1–2% vs ground measurements globally
• 2 km spatial resolution globally — the best spatial resolution available for a global dataset
• Continuously validated — Solcast publishes benchmark accuracy metrics updated regularly
• Historical data available back to 2007 — sufficient for robust inter-annual variability analysis
• Sub-hourly resolution available — 5-minute data for sub-hourly PVsyst simulation

Weaknesses:

• Paid subscription required — pricing is based on data usage; contact Solcast for current pricing
• Overkill for small residential systems where the cost exceeds the value of the precision

Best Use Cases: Recommended for utility-scale projects (typically 10 MW+) where improved accuracy pays back in reduced financing cost, better P90 estimates, and higher investor confidence. Also valuable for projects in regions with complex aerosol conditions (dusty deserts, industrial zones, high-smoke environments) where free sources are less accurate.

Import Method: Access via API key from within PVsyst 8.1, or download as CSV from the Solcast Toolkit web interface and import manually. API access requires your Solcast subscription key to be entered in PVsyst settings.

---

5. ERA5 (ECMWF Reanalysis Version 5)

What It Is

ERA5 is the fifth generation reanalysis dataset produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). Reanalysis means it combines a numerical weather prediction model with historical observations using data assimilation — producing a physically consistent, globally complete dataset that covers 1940 to near-present.

How It Works

ERA5 is not primarily a solar irradiance database — it is a full atmospheric reanalysis that includes surface radiation as one of many output variables. Solar radiation in ERA5 is derived from the IFS (Integrated Forecast System) radiation scheme, not from direct satellite retrieval. This means ERA5 solar radiation accuracy is limited by the model’s cloud simulation quality, which is generally inferior to dedicated satellite-retrieval products for solar energy applications.

Strengths:

• Long historical record — 1940 to present, enabling robust inter-annual variability analysis with 80+ years of data
• Free — available via the Copernicus Climate Data Store (CDS)
• Globally consistent — no gaps, no inhomogeneities from changing satellite instruments
• Includes full atmospheric state — wind, temperature, humidity at multiple pressure levels
• Hourly temporal resolution, 31 km spatial resolution

Weaknesses:

• GHI accuracy is inferior to satellite-retrieval sources — ERA5 typically shows biases of 5–10% in arid regions and larger biases in regions with frequent complex cloud cover
• Not available directly from the PVsyst interface — requires manual download and format conversion
• 31 km spatial resolution misses mesoscale effects (local terrain, sea breeze, orographic clouds)

Best Use Cases: ERA5 is most valuable for inter-annual variability analysis, not as a primary simulation source. Use ERA5 historical data to estimate the standard deviation of annual GHI over multiple decades — this sigma value feeds directly into your P90 calculation. For the P50 simulation itself, use Meteonorm or PVGIS.

Step-by-Step Manual Import:

1. Create a free account at cds.climate.copernicus.eu
2. Navigate to ERA5 → Download data → ERA5 hourly data on single levels
3. Select variables: Surface solar radiation downwards (ssrd) and 2m temperature
4. Select your target year range and geographic bounding box
5. Download as NetCDF format
6. Convert to PVsyst CSV format using Python:

import xarray as xr
import pandas as pd

ds = xr.open_dataset('era5_download.nc')
df = pd.DataFrame({
    'Month': ds.time.dt.month.values,
    'Day': ds.time.dt.day.values,
    'Hour': ds.time.dt.hour.values,
    # ERA5 ssrd is in J/m², convert to W/m² (divide by 3600)
    'GHI': ds['ssrd'].sel(latitude=lat, longitude=lon, method='nearest').values / 3600,
    'Ta': ds['t2m'].sel(latitude=lat, longitude=lon, method='nearest').values - 273.15
})
df.to_csv('era5_pvsyst.csv', index=False)

7. In PVsyst: Tools → Meteo files → Import from file → select your CSV

---

6. Solar Anywhere TGY

What It Is

Solar Anywhere is a US-based solar irradiance data service from Clean Power Research. It provides Typical Gross Year (TGY) data primarily for the North American market.

Coverage and Best Use: Optimized for the US market. For non-US projects, PVGIS or Meteonorm is a better choice. For US utility-scale projects, Solar Anywhere competes with NSRDB and Solcast in terms of accuracy, with particular strength in the western US where their validation network is densest.

---

7. Meteonorm 9 Sub-Hourly Data

What It Is

With PVsyst 8.1’s sub-hourly simulation capability activated, you can import Meteonorm 9 data at minute-level resolution (10-minute or 1-minute time steps).

When to Use It: Sub-hourly data becomes important when your simulation involves:

• Grid export limits — clipping behavior is significantly underestimated with hourly data if the export limit is close to the system’s peak power
• Battery storage dispatch — charging and discharging decisions happen within hours, and ramp rates matter
• High-PV-penetration grids — grid stability analysis requires sub-hourly generation data

For most standard grid-connected simulation without storage, hourly TMY data is sufficient and sub-hourly data does not meaningfully change the annual E_Grid result (typically <0.5% difference).

How to Import: In the Meteo database dropdown, look for Meteonorm 9 (sub-hourly) options. Note that sub-hourly data availability depends on your site’s position within Meteonorm’s coverage grid.

---

8. Custom Measured Data (On-Site Pyranometer)

What It Is

If you have actual measured irradiance data from a pyranometer installed at or near your site — from a resource assessment campaign, from an operating station, or from a national meteorological network — you can import it directly into PVsyst.

Why It Matters

On-site measured data, when properly collected and quality-controlled, is always more accurate than any modeled or satellite-derived source. For large utility projects where financing is at stake, a 1–3 year on-site measurement campaign is a worthwhile investment. Independent engineers reviewing bankable reports give significantly more weight to simulations backed by on-site measurement.

Data Requirements:

• Minimum 1 year, ideally 3+ years for a reliable long-term representative dataset
• Measurement interval: hourly or sub-hourly (15-minute or 1-minute preferred)
• Instrument class: ISO 9060 Class A pyranometer (e.g., Kipp & Zonen CMP11, Hukseflux SR20) for bankable-grade data
• Regular calibration (annual for Class A instruments)
• Data quality control: flag and correct for instrument soiling, shadowing, and sensor drift

PVsyst CSV Format Requirements:

The PVsyst meteo CSV format requires these columns at minimum:

Month  Day  Hour  GHI  DHI  Ta  Ws
1      1    1     0    0    5   2.3
1      1    2     0    0    4   1.8
...

Where:

• GHI = Global Horizontal Irradiance (W/m²)
• DHI = Diffuse Horizontal Irradiance (W/m²) — if not measured, PVsyst can estimate it from GHI using a decomposition model, but measured DHI is preferred
• Ta = Ambient temperature (°C)
• Ws = Wind speed (m/s) — optional but improves thermal model accuracy

Import Method: Use Tools → Meteo files → Import from file to load your CSV. PVsyst will validate the file and report any format errors.

---

Comprehensive Source Comparison

Master Comparison Table

Source	Best Region	GHI Accuracy	Reference Period	Cost	Internet Required	Offline
Meteonorm 8.2	Well-measured regions globally	★★★★	1991–2020	Included with license	No	✅
PVGIS-TMY 5.2	Europe, Middle East, Africa	★★★★★	2005–2020	Free	Yes	❌
NASA NSRDB v4	North & South America	★★★★★	1998–present	Free	Yes	❌
Solcast TMY	Global	★★★★★★	2007–present	Subscription	Yes	❌
ERA5	Global (inter-annual analysis)	★★★	1940–present	Free	Yes	❌
Solar Anywhere	USA	★★★★★	Variable	Subscription	Yes	❌
On-site measured	Your exact site	★★★★★★	Your campaign	Equipment cost	No	✅

Regional Guidance

Europe: Use PVGIS-TMY 5.2 as the primary source — Meteosat satellite coverage for Europe is the best in the world. Meteonorm as second source. Solcast for large utility projects.

Middle East (Saudi Arabia, UAE, Jordan, Iraq): PVGIS 5.2 primary (excellent Meteosat coverage). Meteonorm secondary. For projects above 50 MW, Solcast is strongly recommended — aerosol variability in the Gulf region can significantly affect GHI and a higher-accuracy source pays for itself.

Africa: PVGIS 5.2 primary (Meteosat coverage excellent across all of Africa). Meteonorm secondary. In regions with frequent harmattan dust or industrial aerosols, consider Solcast.

North America (USA, Canada, Mexico): NASA NSRDB v4 primary. Meteonorm secondary. Solar Anywhere or Solcast for utility projects requiring the highest accuracy.

South and Central America: NASA NSRDB v4 primary (GOES coverage good for the entire continent). Meteonorm secondary.

South and Southeast Asia: Solcast or PVGIS primary (Himawari satellite coverage for SE Asia via Solcast; Meteosat-extended coverage via PVGIS for South Asia). Meteonorm secondary. NASA for cross-check.

Central Asia (Kazakhstan, Uzbekistan, Tajikistan): PVGIS 5.2 primary (Meteosat coverage extends to Central Asia). Meteonorm secondary. Sparse ground station coverage means satellite sources often outperform Meteonorm in this region.

Australia and Oceania: Solcast primary (excellent Himawari satellite coverage). Meteonorm secondary.

---

Step-by-Step: Running a Dual-Source Simulation in PVsyst 8.1

This is the workflow you should follow for every professional simulation. The goal is to quantify the meteorological uncertainty in your yield estimate.

Step 1 — Complete Your Base Simulation with Source A

Run your full simulation (system definition, losses, shading — everything) using Meteonorm 8.2 as the meteorological source. Record these results:

• Annual GHI (kWh/m²)
• E_Grid (MWh/year)
• Performance Ratio (%)
• Specific production (kWh/kWp)

Step 2 — Duplicate the Variant

In PVsyst, right-click your variant and select Duplicate. This creates an identical copy of the entire system definition. Give it a name that identifies the meteorological source (e.g., “Variant B — PVGIS 5.2”).

Step 3 — Change Only the Meteorological Source

In the duplicated variant, go to Site and change the Meteo database selection to PVGIS-TMY 5.2. Click Import. Everything else about the system remains identical.

Step 4 — Run the Simulation Again

Click Run Simulation for Variant B. Record the same output metrics.

Step 5 — Compare and Interpret

Metric	Variant A (Meteonorm)	Variant B (PVGIS)	Delta (%)
Annual GHI (kWh/m²)	___	___	___
E_Grid (MWh/year)	___	___	___
PR (%)	___	___	___
Specific Production	___	___	___

Interpreting the GHI Delta:

The GHI delta tells you how much the two sources disagree on the fundamental irradiance at your site:

• <3%: Excellent agreement. Both sources are likely reliable for your site. Use Meteonorm as P50 (industry standard) and note in your report that PVGIS confirms within 3%.
• 3–5%: Good agreement within expected uncertainty. Report both numbers and document which source you’re using for the P50 estimate and why.
• 5–8%: Notable disagreement. Add a third source (Solcast, or on-site measured data if available). Investigate whether one source has known limitations for your specific region.
• >8%: Significant disagreement. Do not finalize the report until the discrepancy is explained. Common causes: site in complex microclimate, one source lacks coverage for the region, or site coordinates are incorrect.

Why PR Can Move in the Opposite Direction from E_Grid: If PVGIS reports higher GHI than Meteonorm, you will typically see higher E_Grid but lower PR from the PVGIS simulation. This is not a contradiction — PR is normalized by irradiance. Higher irradiance means higher cell temperatures, which increases thermal losses. The net result is higher absolute energy yield (more irradiance input) but lower PR (larger fraction lost to heat).

---

Understanding Why Sources Disagree: A Technical Deep-Dive

When sources give different answers, it is almost always traceable to one of these four root causes.

Root Cause 1 — Different Measurement Methodologies

Meteonorm interpolates from ground stations; PVGIS retrieves from satellite imagery. These are fundamentally different measurement approaches with different error characteristics. Ground station interpolation is highly accurate near stations but degrades with distance. Satellite retrieval provides spatially continuous coverage but can be biased by undetected thin cirrus clouds, aerosol loading, or land surface reflectance effects.

In practice: if your site is within 20 km of a high-quality WMO-standard meteorological station, Meteonorm may outperform PVGIS. If your site is in a region with sparse ground coverage and clear Meteosat visibility, PVGIS is likely more accurate.

Root Cause 2 — Different Reference Periods

The reference period determines what “typical” means for the TMY construction:

Source	Reference Period	Length
Meteonorm 8.2	1991–2020	30 years
PVGIS-TMY 5.2	2005–2020	15 years
NASA NSRDB v4	1998–present	~26 years
Solcast TMY	2007–present	~17 years

If global brightening (the observed increase in surface solar radiation since the 1980s, due to reduced aerosol loading in many regions) has affected your site, the more recent reference period of PVGIS may give slightly higher GHI than Meteonorm’s 30-year average. Conversely, if recent years have been cloudier at your specific site, Meteonorm’s longer average may produce more conservative (lower) estimates.

Root Cause 3 — Decomposition Model Differences

Both sources provide GHI, but PVsyst needs DNI (direct normal irradiance) to calculate plane-of-array irradiance accurately. The conversion from GHI to DNI uses a decomposition model. PVsyst applies the same decomposition model (Erbs or Perez, configurable in settings) to all imported hourly TMY files, so this is generally not a source of between-source differences. However, if one source provides actual measured DNI alongside GHI (as Solcast and NSRDB can), using that directly is more accurate than decomposing GHI.

Root Cause 4 — Spatial Resolution and Microclimate Effects

PVGIS 5.2 has 3–5 km spatial resolution. Meteonorm interpolation spacing depends on station density — it can be 10–100 km in sparse regions. For sites with strong microclimate effects (coastal sites with sea breeze clouds, valley fog, urban heat islands, orographic precipitation), neither 5 km satellite pixels nor 50 km interpolation captures the site-specific effect. Only on-site measurement resolves this.

---

Advanced Topics

Building a Custom TMY from Multi-Year Measured Data

If you have multiple years of on-site pyranometer measurements, you can build a site-specific TMY that is more representative than any modeled source. The standard ISO 15927-4 method works as follows:

1. For each calendar month across all years of data, calculate the monthly GHI total
2. Rank the monthly GHI totals for each month
3. Select the month from the available years whose GHI total is closest to the long-term median for that month
4. Assemble the 12 selected months into a complete 8,760-hour TMY year
5. Smooth the boundaries between selected months to avoid discontinuities

PVsyst does not automate this process, but you can implement it in Python using the pvlib library and then import the result as a custom CSV.

Inter-Annual Variability Assessment and P50/P90 Estimation

Every bankable energy yield report must include a P90 estimate — the energy yield that will be exceeded in 9 out of 10 years. The P90 is lower than the P50 (median yield) because irradiance varies from year to year, and the P90 accounts for the risk of below-average irradiance years.

The standard methodology:

Step 1 — Establish the long-term mean GHI (P50 irradiance) This is the GHI in your TMY file (Meteonorm or PVGIS). The TMY is designed to represent the long-term average, so your TMY simulation gives you the P50 yield estimate.

Step 2 — Estimate inter-annual GHI variability (σ_GHI) Download ERA5 hourly data for your site covering 20–30 years. For each year, calculate the annual GHI total. The standard deviation of these annual GHI values is your inter-annual variability estimate σ_GHI.

For most mid-latitude sites, σ_GHI is typically 2–5% of the mean GHI. Desert sites tend toward the lower end (stable, cloud-free climate). Temperate maritime sites (UK, Pacific Northwest) tend toward the higher end.

Step 3 — Calculate the P90 yield

P90_GHI = P50_GHI × (1 - 1.28 × σ_GHI_fraction)
P90_E_Grid ≈ P50_E_Grid × (P90_GHI / P50_GHI)

The 1.28 factor comes from the one-tailed normal distribution at the 10th percentile.

Example:

• P50 GHI from Meteonorm: 2,050 kWh/m²/year
• σ_GHI from ERA5 30-year analysis: 3.5%
• P90 GHI = 2,050 × (1 – 1.28 × 0.035) = 2,050 × 0.955 = 1,958 kWh/m²
• P50 E_Grid = 42.0 GWh/year
• P90 E_Grid ≈ 42.0 × (1,958 / 2,050) = 40.1 GWh/year

This means the project will produce at least 40.1 GWh in 9 out of 10 years, based on inter-annual irradiance variability alone. (A complete P90 analysis also includes technology uncertainty, modeling uncertainty, and degradation — each adding their own sigma in quadrature.)

Validating Your Source Against Satellite-Derived GHI Maps

Before finalizing your source selection, do a quick sanity check using PVGIS’s online map interface (pvgis.ec.europa.eu → Interactive maps → Global horizontal irradiation). Look up your site’s GHI on the map and compare it with what your imported Meteonorm data says. If they disagree by more than 5–7%, investigate before proceeding.

The Clearness Index as a Data Quality Check

The clearness index Kt is the ratio of surface GHI to extraterrestrial horizontal irradiance. For a given location and time of year, physically plausible Kt values range from about 0.2 (heavily overcast) to about 0.85 (exceptionally clear). If your imported hourly TMY shows Kt values consistently above 0.85 or below 0.15, there is likely a data quality issue — either the import coordinates are wrong, the decomposition model is misbehaving, or the source file is corrupted.

In PVsyst, you can inspect the hourly meteo data via Tools → Meteo files → View/Edit to check for anomalies.

---

Troubleshooting Common Import Problems

Problem 1 — PVGIS or NASA Download Times Out

Symptom: PVsyst shows a download progress bar that stalls or a timeout error. Causes: Slow internet, JRC/NREL server downtime (rare), or firewall blocking the API request. Solutions:

• Try again after a few minutes — server load fluctuates
• Check JRC PVGIS server status at pvgis.ec.europa.eu
• Use the manual download route: get the TMY CSV directly from the PVGIS web interface using your browser, then import via Tools → Meteo files → Import
• If all online sources are blocked (corporate network, restrictive firewall), use Meteonorm which is fully offline

Problem 2 — Two Sources Differ by More Than 8%

Symptom: Meteonorm and PVGIS give E_Grid results more than 8% apart. Diagnostic steps:

1. Check the coordinates — are they pointing to the right location? Verify on PVsyst’s built-in map
2. Check the altitude — is it correct? A wrong altitude doesn’t cause 8% GHI discrepancy but confirms care was taken
3. Load a third source (Solcast or NASA) and see which of the first two it’s closer to
4. Check whether your site has known microclimate characteristics (coastal, valley, industrial) that neither source models well
5. If possible, contact the local meteorological service for any available station data near your site

Problem 3 — Custom CSV Import Fails

Symptom: PVsyst rejects your CSV file with a format error. Common causes and fixes:

• Wrong decimal separator: PVsyst expects decimal points (3.14), not commas (3,14). Fix your CSV locale settings.
• Wrong column order or missing columns: PVsyst expects a specific column structure. Open a valid Meteonorm-exported CSV via Tools → Meteo files → View and match your file’s column layout to it.
• Wrong time convention: PVsyst uses solar time, not local time. If your station data uses UTC or local civil time, you need to shift the hours accordingly.
• Wrong units: GHI must be in W/m² (instantaneous), not Wh/m² (hourly integral). If your data uses Wh/m² (common for some station data formats), divide all radiation values by 1 before importing — they are numerically equal for hourly data, but make sure.

Problem 4 — Monthly GHI Chart Looks Inverted or Wrong

Symptom: After import, the monthly GHI bars show peak radiation in December-January rather than June-July (for a Northern Hemisphere site), or values that are wildly out of range. Causes:

• Latitude sign error (entered negative latitude for a Northern Hemisphere site, or vice versa)
• Longitude error placing the site in the wrong hemisphere
• Elevation dramatically wrong (e.g., entered 10000 m by accident) Fix: Go back to Site → Location and verify every coordinate field. Click the map to visually confirm the pin is in the right location.

---

Frequently Asked Questions

What is the difference between GHI, DNI, and DHI — and which does PVsyst actually use?

GHI (Global Horizontal Irradiance) is total solar irradiance on a horizontal surface — direct beam + diffuse sky + reflected ground. DNI (Direct Normal Irradiance) is the beam component measured perpendicular to the sun’s rays. DHI (Diffuse Horizontal Irradiance) is the sky scatter component on a horizontal surface, excluding direct beam.

PVsyst reads GHI and DHI from the TMY file (or GHI alone if DHI is unavailable). It derives DNI using a decomposition model. It then uses DNI, DHI, and ground-reflected radiation as inputs to the transposition model to calculate plane-of-array (POA) irradiance. If the TMY file includes measured DNI directly (as Solcast and NSRDB files can), PVsyst will use that instead of decomposing GHI — which improves accuracy.

What is a TMY and why is it better than single-year real data for simulation?

A Typical Meteorological Year (TMY) is a synthetic 8,760-hour dataset constructed by selecting representative months from a multi-year archive. The goal is that each month represents the statistical median for that calendar month, so the full year represents long-term average conditions rather than any single anomalous year. For long-term yield prediction (25-year financial models), TMY is the correct input. For comparing actual measured production against the simulation baseline, use a real measured year.

Why does PR decrease when I switch from Meteonorm to PVGIS, even though E_Grid increases?

This is physically correct and expected. Performance Ratio (PR) is normalized by incident irradiance — it measures what fraction of the input energy is successfully converted to grid energy. If PVGIS reports higher GHI, the modules receive more irradiance, which raises cell temperature, which increases thermal losses. More energy is lost proportionally, so PR falls. E_Grid still increases because the absolute input is larger, but the conversion efficiency (PR) is slightly worse.

Can I use multiple meteorological files in the same PVsyst project?

Yes — each Variant in a PVsyst project can have its own meteorological file assigned. Create one variant per source and run simulations for each. The Compare Variants function lets you see results side by side.

How many years of on-site data do I need before it’s worth using instead of Meteonorm?

Statistically, a single year of measured data has high uncertainty because it may be an anomalous year. Three years is the practical minimum for a representative dataset. Five years gives reasonable confidence. For very large projects where financing is substantial, a 1-year measurement campaign followed by correlation with long-term satellite data (MCP — Measure-Correlate-Predict method) is the industry standard.

Does Meteonorm 8.2 include data for temperature and wind speed?

Yes — Meteonorm includes monthly statistics for ambient temperature (Ta) and wind speed (Ws), which the stochastic hourly generator translates into hourly values. These are essential for PVsyst’s thermal model. PVGIS also provides Ta, but if a source does not include temperature data, PVsyst falls back on a default temperature model which slightly reduces simulation accuracy.

What is the impact of choosing the wrong meteorological source on my final E_Grid number?

In well-measured regions with multiple reliable sources agreeing to within 3%, the choice of source affects E_Grid by at most 2–4%. In poorly-measured regions where sources disagree by 8–12%, the choice of source can shift your E_Grid by 6–10%. This translates to meaningful financial model differences — on a 100 MWp project generating ~160 GWh/year, a 5% E_Grid error means ~8 GWh/year difference, worth millions of dollars in annual revenue projection.

---

Conclusion and Practical Decision Framework

Choosing the right meteorological source is not a single universal answer — it depends on your site’s location, available sources, project size, and accuracy requirements. Here is a practical decision framework:

For any project, always:

1. Run simulations with at least two independent sources
2. Compare the GHI delta between sources — anything above 5% requires investigation
3. Document which source you used as P50 and why
4. Include the meteorological uncertainty explicitly in your P90 analysis

Source selection by region:

• Europe, Middle East, Africa → PVGIS 5.2 (primary) + Meteonorm 8.2 (secondary)
• North & South America → NASA NSRDB v4 (primary) + Meteonorm 8.2 (secondary)
• Central and South Asia → PVGIS 5.2 (primary) + Meteonorm 8.2 (secondary)
• Global utility-scale (>20 MW) → Solcast (primary) + PVGIS or NASA (secondary)
• Site with measured data (>3 years) → Custom TMY (primary) + Meteonorm (secondary)

For P90 uncertainty analysis: Use ERA5 30-year historical data to quantify inter-annual GHI variability (σ), then apply P90 = P50 × (1 – 1.28σ).

No meteorological source is perfect. The goal of using two independent sources is not to find the “true” answer — it is to quantify the range of uncertainty inherent in any irradiance estimate, and to communicate that uncertainty honestly in your report. That transparency is the foundation of a genuinely bankable energy yield assessment.

💡 Run your simulations with full access to all PVsyst 8.1 features including Meteonorm 8.2. Download the complete licensed version at Docrack.me — PVsyst 8.1.

---