EPW Future Weather Generator
What is this tool?
The EPW Future Weather Generator V2.0 creates climate-adjusted weather files for building energy simulation. It takes existing EPW (EnergyPlus Weather) files and morphs them to represent future climate conditions based on IPCC climate scenarios and global climate model projections. This is for energy modelers who need to assess how buildings will perform under climate change, not for meteorological research.
The tool implements the Future Weather Generator (FWG) methodology documented by Rodrigues et al. (2023), including Kusuda-Achenbach ground temperature calculations and multiple diffuse radiation separation models. Advanced users can import custom climate deltas from official FWG outputs or other sources for location-specific accuracy. If you're doing climate-responsive design, long-term resilience planning, or future-proofing buildings against warming temperatures, this generates the weather files your simulation software needs.
Official FWG Tool: future-weather-generator.adai.pt
Download the tool here: EPW Future Weather Generator V1.0
System Requirements
Browser: Any modern browser with JavaScript enabled. Chrome, Firefox, Edge, Safari all work fine.
Network: Required only for the logo image. All calculations run entirely in your browser.
Files: Standalone HTML file. No installation, no dependencies, no server needed. The Christchurch TMYx weather file is embedded as a default example.
Quick Start
Generate Future Weather Files (5 minutes)
The core workflow. You have a present-day EPW file, you need future climate versions.
Step 1: Load your EPW file
Drop an EPW file onto the upload area or click to browse. The Christchurch file loads automatically as a demo. You can load multiple files and process them all at once.
Step 2: Select climate scenarios
Check the SSP scenarios you want:
SSP1-2.6: Low emissions, aggressive mitigation
SSP2-4.5: Middle of the road
SSP3-7.0: Regional rivalry, high emissions
SSP5-8.5: Fossil-fueled development, worst case
Most compliance work uses SSP2-4.5 and SSP5-8.5 to bracket likely outcomes.
Step 3: Select time periods
Check the future years: 2030, 2050, 2080, 2090. Near-term (2030, 2050) for operational planning, far-term (2080, 2090) for building lifespan analysis.
Step 4: Choose GCM model
Select a Global Climate Model or use the ensemble average. Different models have different climate sensitivities:
Ensemble Average: Balanced projection across multiple models
CanESM5: High sensitivity (warmer projections)
EC-Earth3: Moderate sensitivity
ACCESS-CM2: Low sensitivity (cooler projections)
For most work, use ensemble average. For sensitivity analysis, run multiple GCMs.
Step 5: Click Generate Future Weather
Progress bar shows completion. Each combination of scenario and year produces one output file. Three scenarios times four years equals twelve files per input EPW.
Step 6: Download results
Generated files appear in the results panel. Click the download button on each file, or load them directly into EnergyPlus, IES VE, or other simulation software that accepts EPW format.
Compare Weather Files (3 minutes)
Use the built-in viewer to compare original and morphed files.
Step 1: Select files for comparison
Click the file badges above the charts. Selected files highlight in blue. Maximum five files can display simultaneously.
Step 2: Choose time filter
Year: Full 8760 hours
Summer: December-February (Southern Hemisphere) or June-August (Northern Hemisphere)
Winter: June-August (Southern Hemisphere) or December-February (Northern Hemisphere)
Peak Summer Day: Hottest 24-hour period by average temperature
Peak Summer Week: Hottest 168-hour period
Peak Winter Day: Coldest 24-hour period
Peak Winter Week: Coldest 168-hour period
Peak periods are calculated from the first selected file and applied to all displayed files for direct comparison.
Step 3: Toggle statistics
Click "Show Statistics" to display min, mean, max, P5, and P95 values for temperature, solar radiation, and humidity. Statistics update when you change the time filter.
Step 4: Export diagnostics
Click "Export Diagnostics" to download a comprehensive text report for verifying morphing quality. The report includes:
Header information (location, coordinates, elevation)
Ground temperatures at all depths
Annual statistics for all key variables (N6-N21)
Monthly temperature and solar summaries
Peak period identification
Data quality checks (missing values, physical consistency)
Sample hourly data for each season
File comparison (temperature, humidity, radiation deltas)
Use this to verify that morphed files are physically reasonable before running simulations.
Understanding the Inputs
Climate Scenarios (SSP)
Shared Socioeconomic Pathways describe different futures based on policy choices, technology development, and societal trends. They replace the older RCP scenarios.
SSP1-2.6 assumes rapid decarbonization, global cooperation, sustainable development. Temperature increase limited to approximately 1.8°C by 2100. Use this for optimistic policy scenarios.
SSP2-4.5 assumes current trends continue with moderate mitigation efforts. Temperature increase approximately 2.7°C by 2100. This is the "business as usual" scenario most commonly used for planning.
SSP3-7.0 assumes regional rivalry, slow economic growth, and weak climate policies. Temperature increase approximately 3.6°C by 2100. Use for stress-testing designs.
SSP5-8.5 assumes fossil-fuel intensive development with minimal climate policy. Temperature increase approximately 4.4°C by 2100. Worst-case scenario for resilience analysis.
The tool applies scaling factors based on these scenarios: SSP1-2.6 at 65% of baseline warming, SSP2-4.5 at 100%, SSP3-7.0 at 135%, SSP5-8.5 at 170%.
Time Periods
Future years determine how far the climate projections extend:
2030: Near-term, 0.45× baseline warming factor
2050: Mid-century, 1.0× baseline warming factor
2080: Late-century, 1.55× baseline warming factor
2090: End-century, 1.75× baseline warming factor
These factors compound with the SSP scenario factors. A 2090 SSP5-8.5 projection shows substantially more warming than a 2030 SSP1-2.6 projection.
GCM Models
Global Climate Models simulate Earth's climate system. Different modeling groups produce different results based on their approaches to representing atmospheric physics, ocean circulation, and feedback mechanisms.
Ensemble Average combines results from multiple models. This smooths out individual model biases and provides a consensus projection. Best for general-purpose work.
CanESM5 (Canadian Earth System Model) has high climate sensitivity. Projections run warmer than average. Use for conservative design margins.
EC-Earth3 (European consortium) has moderate sensitivity close to the ensemble mean. Good alternative to the ensemble for European projects.
ACCESS-CM2 (Australian model) has lower climate sensitivity. Projections run cooler than average. Use for lower-bound estimates.
The tool applies model-specific factors for temperature, humidity, solar radiation, cloud cover, wind speed, and precipitation. High-sensitivity models amplify all climate changes; low-sensitivity models dampen them.
Radiation Models
Diffuse Separation (N15)
Three models separate global horizontal radiation into direct and diffuse components. This affects illuminance calculations and daylighting analysis.
Ridley, Boland & Lauret (2010): Default model. Uses clearness index, solar time, altitude angle, daily clearness, and persistence. Well-validated across multiple climates.
Engerer (2015): Method 3 from the Engerer series. Similar inputs with different coefficient fitting. May perform better in specific climates.
Paulescu & Blaga (2019): Two-predictor model using clearness index and solar altitude. Simpler formulation, still accurate for most applications.
For most work, stick with Ridley. Switch models only if you have specific validation data for your climate zone.
Illuminance Model (N16-N19)
Two methods are available for calculating illuminance values:
Luminous Efficacy (default): Converts radiation to illuminance using sky-condition-dependent efficacy values (lm/W). More robust across edge cases and produces physically consistent results. Recommended for most applications.
Perez et al. 1990: The coefficient-based model from the original FWG methodology. Uses 8 sky clearness bins with tabulated coefficients for precipitable water content, zenith angle, and atmospheric brightness. Use this when you need results comparable to official FWG outputs or other tools using the Perez model.
Ground Temperature Update
When enabled, the tool recalculates monthly ground temperatures at three depths (0.5m, 2.0m, 4.0m) using the Kusuda-Achenbach (1965) equations. Ground temperatures lag air temperature changes and affect basement heat transfer, ground-source heat pump performance, and slab-on-grade calculations.
Leave this enabled unless you have separate ground temperature data from a geotechnical study.
Advanced Options
Click the "⚙️ Advanced Options" panel header to expand advanced settings. These options give sophisticated users more control over the morphing process.
Climate Data Source
Choose how climate change deltas are calculated:
Built-in scaling factors (default): Uses the tool's internal scaling approach based on SSP scenarios, time periods, GCM model sensitivities, and latitude. Suitable for most design work.
Custom monthly deltas (CSV): Import your own monthly climate change values. Use this when you have:
Output from the official FWG tool for your specific location
Regional climate study data
Custom projections from a climate consultant
Importing Custom Deltas
Click "📤 Template" to download a CSV template with the required format
Edit the template with your monthly delta values (12 rows, one per month)
Click "📥 Import CSV" to load your custom data
The status indicator confirms successful import
CSV Format:
Column definitions:
dTAS
°C
Mean temperature change
dTASMAX
°C
Daily maximum temperature change
dTASMIN
°C
Daily minimum temperature change
dHUSS
kg/kg
Specific humidity change (e.g., 0.0012)
dPSL
Pa
Sea level pressure change
dRSDS_pct
decimal
Solar radiation change as fraction (0.02 = +2%)
dCLT
tenths
Cloud cover change (0-10 scale)
windRel
ratio
Wind speed multiplier (1.05 = +5%)
snowRel
ratio
Snow depth multiplier (0.85 = -15%)
precRel
ratio
Precipitation multiplier
When custom deltas are loaded, the SSP and year selections are ignored—your values are applied directly. Output files are named with _custom_ instead of the GCM model name.
Regional Adjustments
Fine-tune the built-in scaling factors for your region:
Latitude amplification (default 0.35): Controls polar amplification—how much more high latitudes warm compared to the equator. Higher values increase warming at high latitudes. The formula is: warming × (1 + |latitude|/90 × amplification)
Seasonal amplitude (default 0.15): Controls how much warming varies by season. Higher values create larger differences between summer and winter warming. Set to 0 for uniform year-round warming.
These settings only affect the built-in scaling factors, not custom imported deltas.
Output Options
Embed methodology in EPW: When enabled, adds a comment line to the EPW header documenting how the file was created:
This creates an audit trail when files are shared between team members or archived for future reference.
Export validation CSV: Generates a CSV file comparing original and morphed file statistics. Useful for QA documentation and verifying that morphing produced expected results.
Export applied deltas: Generates a CSV file listing the exact monthly delta values that were applied. Useful for:
Documenting methodology for reports
Comparing built-in values to official FWG outputs
Replicating results in other tools
How the Calculation Works
The tool implements the morphing methodology from the Future Weather Generator documentation. Each weather variable gets adjusted differently.
Temperature (Dry Bulb, Dew Point)
Temperature uses shift-and-stretch morphing:
Where:
ΔT_mean is the projected mean temperature change
α is the stretch factor calculated from changes in daily maximum and minimum temperatures
T_mean_present is the baseline monthly mean temperature
This preserves the diurnal temperature range while shifting and potentially amplifying extremes. Hot days get hotter; cold nights may warm more or less depending on the GCM projections.
Dew point is derived from morphed humidity using Newton-Raphson iteration to find the temperature where saturation vapor pressure equals the calculated vapor pressure.
Humidity
Relative humidity morphs through specific humidity to preserve physical consistency:
Calculate present specific humidity from relative humidity and saturation pressure
Add projected specific humidity change (ΔQ)
Calculate new vapor pressure from modified specific humidity
Derive relative humidity from vapor pressure and saturation pressure at the new temperature
This prevents impossible values (humidity over 100%) that simple scaling would produce.
Solar Radiation
Global horizontal radiation uses stretch morphing:
Diffuse horizontal radiation is separated from global using the selected diffuse model (Ridley, Engerer, or Paulescu), which considers:
Hourly clearness index (ratio of measured to extraterrestrial radiation)
Daily clearness index
Clearness persistence (previous and next hour values)
Solar altitude angle
Solar time
Direct normal radiation is calculated from global and diffuse using solar geometry:
Values are capped at 1200 W/m² to prevent unrealistic spikes at low solar angles.
Illuminance
Two calculation methods are available:
Luminous Efficacy Method (Default)
Converts radiation to illuminance using sky-condition-dependent efficacy values:
Global Horizontal Illuminance (N16): GHR × efficacy × altitude factor
Direct Normal Illuminance (N17): DNR × efficacy × altitude factor
Diffuse Horizontal Illuminance (N18): DHR × efficacy × altitude factor
Zenith Luminance (N19): Derived from diffuse illuminance and solar altitude
Luminous efficacy values (lm/W) vary by sky clearness:
Overcast (ε < 1.5)
115
95
130
Intermediate (1.5 ≤ ε < 3.0)
110
100
120
Clear (ε ≥ 3.0)
105
105
110
The altitude factor adjusts efficacy based on solar position: 0.9 + 0.2 × sin(altitude).
Perez Model (FWG-Compatible)
Uses the Perez et al. (1990) coefficient tables with 8 sky clearness bins. Each illuminance type uses different coefficients based on:
Precipitable water content (from dew point)
Atmospheric brightness (from diffuse radiation and air mass)
Zenith angle
Sky clearness category
This method matches the official FWG implementation for comparison purposes.
Other Variables
Atmospheric pressure: Simple offset based on projected sea level pressure change.
Wind speed: Multiplicative scaling based on projected wind speed ratio.
Sky cover: Additive offset for total sky cover; opaque sky cover scaled proportionally.
Snow depth: Multiplicative scaling (typically decreasing with warming).
Precipitation: Multiplicative scaling (may increase or decrease depending on region).
Ground Temperatures
Kusuda-Achenbach (1965) equations calculate monthly ground temperatures from:
Annual mean air temperature (from morphed data)
Annual temperature amplitude
Soil thermal diffusivity (0.056 m²/day default)
Phase lag from coldest month
Depth below surface
Temperature variation decreases with depth. At 4m, seasonal swing is minimal; at 0.5m, it tracks air temperature with a phase delay.
Header Updates
The tool updates EPW header lines:
Line 1 (LOCATION): Appends GCM model, scenario, and year to city name
Line 4 (GROUND TEMPERATURES): Replaces with calculated values at three depths
Line 6 (COMMENTS 2): Adds methodology stamp when "Embed methodology" is enabled
Reading the Results
Charts
Three time-series charts show:
Temperature (°C): Dry bulb temperature over the selected period. Future files shift upward and may show amplified peaks.
Global Horizontal Radiation (Wh/m²): Solar radiation reaching a horizontal surface. Changes are typically smaller than temperature changes.
Relative Humidity (%): Bounded 0-100%. May increase or decrease depending on the balance between temperature and moisture changes.
Each selected file displays as a different color. The legend identifies which line corresponds to which file.
Statistics Table
When visible, shows for each selected file:
Min
Minimum value in filtered period
Mean
Average value
Max
Maximum value
P5
5th percentile (value exceeded 95% of the time)
P95
95th percentile (value exceeded 5% of the time)
P5 and P95 are more robust than min/max for identifying typical extremes without outlier sensitivity.
Generated Files
Output files follow the naming convention:
Example: Christchurch_ensemble_ssp245_2050.epw
When using custom deltas:
Example: Christchurch_custom_ssp245_2050.epw
Files are valid EPW format compatible with EnergyPlus, IES VE, TRNSYS, and other simulation engines that read EPW weather data.
Common Scenarios
Scenario 1: Code Compliance Future-Proofing
Design team wants to show the building performs adequately under 2050 climate conditions.
Load the site's TMY weather file
Select SSP2-4.5 (likely scenario) and SSP5-8.5 (stress test)
Select 2050 only
Use ensemble GCM
Generate files
Run energy simulations with both future files
Document that design cooling capacity handles projected loads
Result: Evidence that the HVAC system won't be undersized mid-building-lifespan.
Scenario 2: Climate Risk Assessment
Owner wants full range of climate projections for a 60-year building life.
Load weather file
Select all four SSP scenarios
Select 2050 and 2080
Use ensemble GCM
Generate files (8 total)
Compare peak cooling loads across all scenarios
Identify when/if existing systems become inadequate
Result: Risk matrix showing probability and timing of climate-related performance issues.
Scenario 3: Resilience Design
Architect designing for climate adaptation wants worst-case summer conditions.
Load weather file
Select SSP5-8.5 only
Select 2080 and 2090
Use CanESM5 (high sensitivity)
Generate files
Use Peak Summer Week filter to view extreme conditions
Design passive cooling strategies for these extremes
Result: Design parameters for a building that maintains habitability during future heat waves.
Scenario 4: Sensitivity Analysis
Researcher comparing how different GCMs affect results.
Load weather file
Select SSP2-4.5
Select 2050
Generate with ensemble, then repeat with CanESM5 and ACCESS-CM2
Compare statistics across all three
Document uncertainty range in projections
Result: Confidence intervals on climate projections for academic or policy work.
Scenario 5: FWG-Compatible Output
User needs results that match official FWG tool outputs for comparison or validation.
Run official FWG tool for your location to get monthly deltas
Export the deltas to CSV format matching the template
Load weather file in this tool
Expand Advanced Options
Select "Custom monthly deltas (CSV)"
Import your FWG-derived deltas
Select Perez illuminance model for full compatibility
Enable "Export applied deltas" for documentation
Generate files
Result: Future weather files using official FWG climate data with this tool's visualization and comparison features.
Scenario 6: Regional Calibration
User has regional climate study data suggesting different warming patterns.
Load weather file
Expand Advanced Options
Adjust latitude amplification if regional studies show different polar amplification
Adjust seasonal amplitude if regional studies show different seasonal patterns
Enable validation CSV export
Generate files
Review validation CSV to confirm expected temperature changes
Result: Future weather files calibrated to regional climate research.
Limitations
This tool implements a simplified morphing methodology suitable for building energy analysis.
What it is:
A tool for generating future weather files for simulation
Suitable for climate-responsive design analysis
Appropriate for code compliance future-proofing studies
Based on peer-reviewed FWG methodology
Extensible via custom delta import for location-specific accuracy
What it isn't:
A substitute for regional climate modeling
Suitable for detailed meteorological analysis
Going to capture changes in weather patterns, storm frequency, or extreme event timing
Applicable for agricultural, hydrological, or ecological studies
Morphing limitations: The methodology assumes climate change shifts and stretches existing weather patterns rather than fundamentally changing them. A location that never had heat waves won't suddenly generate realistic heat wave sequences. The morphed file shows higher temperatures but maintains the original temporal patterns.
Spatial resolution: Built-in GCM projections are applied uniformly based on latitude. Real climate change varies locally based on urban heat islands, coastal effects, and topography. The tool doesn't downscale to microclimate level. For location-specific projections, import custom deltas from regional climate studies.
Extreme events: The methodology handles gradual shifts in means and variances. It doesn't insert new extreme events or change the frequency distribution of existing ones. For extreme event analysis, use dedicated climate impact tools.
Uncertainty: Climate projections have inherent uncertainty. The spread between SSP scenarios and GCM models gives some indication, but doesn't capture all sources of uncertainty. Use results for planning ranges, not precise predictions.
Built-in vs. Official FWG: The built-in scaling factors are simplified approximations. For research-grade accuracy or when comparing to peer-reviewed studies using FWG, import actual GCM deltas from the official FWG tool using the custom delta feature.
Technical Reference
EPW Fields Modified
N6
Dry Bulb Temperature
Shift + Stretch
N7
Dew Point Temperature
Derived from humidity
N8
Relative Humidity
Via specific humidity
N9
Atmospheric Pressure
Offset
N10
Extraterrestrial Horizontal Radiation
Recalculated
N11
Extraterrestrial Direct Normal Radiation
Recalculated
N12
Horizontal Infrared Radiation
Recalculated
N13
Global Horizontal Radiation
Stretch
N14
Direct Normal Radiation
Derived (capped 1200 W/m²)
N15
Diffuse Horizontal Radiation
Diffuse model
N16
Global Horizontal Illuminance
Efficacy or Perez
N17
Direct Normal Illuminance
Efficacy or Perez
N18
Diffuse Horizontal Illuminance
Efficacy or Perez
N19
Zenith Luminance
Efficacy or Perez
N21
Wind Speed
Scale
N22
Total Sky Cover
Offset
N23
Opaque Sky Cover
Proportional
N30
Snow Depth
Scale
N33
Liquid Precipitation
Scale
Key Constants
Solar Constant
1367 W/m²
Standard
Stefan-Boltzmann
5.67×10⁻⁸ W/(m²·K⁴)
Physical constant
Ground Diffusivity
0.056 m²/day
Typical soil
DNR Maximum Cap
1200 W/m²
Physical limit
Clear Sky Efficacy
105 lm/W
Typical daylight
Overcast Efficacy
115-130 lm/W
Typical daylight
Default Latitude Amplification
0.35
Calibrated
Default Seasonal Amplitude
0.15
Calibrated
Output Files
When output options are enabled, the tool generates:
*_[scenario].epw
Morphed weather files
epw_validation_[date].csv
Original vs morphed statistics comparison
applied_deltas_[date].csv
Monthly delta values actually applied
References
Rodrigues, E., Fernandes, M.S., Carvalho, D., & Andrade, C. (2023). Future Weather Generator methodology documentation. Building and Environment, 233, 110104.
Perez, R., Ineichen, P., Seals, R., Michalsky, J., & Stewart, R. (1990). Modeling daylight availability and irradiance components from direct and global irradiance. Solar Energy, 44(5), 271-289.
Kusuda, T., & Achenbach, P.R. (1965). Earth temperature and thermal diffusivity at selected stations in the United States. ASHRAE Transactions, 71(1), 61-75.
Ridley, B., Boland, J., & Lauret, P. (2010). Modelling of diffuse solar fraction with multiple predictors. Renewable Energy, 35(2), 478-483.
Engerer, N.A. (2015). Minute resolution estimates of the diffuse fraction of global irradiance for southeastern Australia. Solar Energy, 116, 215-237.
Paulescu, M., & Blaga, R. (2019). A simple and reliable empirical model with two predictors for estimating 1-minute diffuse fraction. Solar Energy, 180, 75-84.
Keyboard Shortcuts
Tab
Navigate between input fields
Enter
Trigger calculation when focused on input
Escape
Close modal dialogs
Last updated
Was this helpful?