Skip to content

Future Climate Projections

POWER provides downscaled, bias-corrected future climate projections to support long-term planning, climate resilience, and analysis of the future under different emissions scenarios. Unlike the observation-driven satellite retrievals and atmospheric reanalyses (CERES SYN1deg, FLASHFlux, MERRA-2, GEOS-IT) featured on the POWER website, the projections are forward-looking model outputs generated by state-of-the-art global climate models. Statistical downscaling and bias correction (e.g., the BCSD method) are standard techniques for translating coarse GCM output into higher-resolution fields suitable for impacts analysis while reducing systematic model bias against an observational reference Wood et al., 2004; Thrasher et al., 2022. The basis of these projections are a group of climate models that are part of the Coupled Model Intercomparison Project Phase 6 (CMIP6), a coordinated international effort that coordinates and provides a standard framework for the content and format of the model outputs Eyring et al., 2016; IPCC AR6 WG1, 2021.

POWER provides future climate projections curated by NASA's RASI workgroup, derived from NEX-GDDP-CMIP6 statistical downscaling of CMIP6 GCMs from a ~2° spatial resolution of the raw CMIP6 model output to a ~0.25° resolution. The NEX-GDDP-CMIP6 product provides daily, bias-corrected, spatially downscaled projections on a 0.25° (~25 km) global land grid spanning 1950–2100 NEX-GDDP-CMIP6 landing; Thrasher et al., 2022.

RASI/GEX-GDDP-CMIP6

The RASI projections served through POWER are supported in part by NASA's Risk Analysis and Solutions Innovators (RASI) workgroup at the NASA Goddard Institute for Space Studies (GISS). RASI builds on NASA's Earth Science Division and Office of Strategic Infrastructure (OSI) partnership and produces co-produced, hazard-based climate projections for NASA Centers and the surrounding regions. RASI is GISS's hazard- and risk-focused climate-projection activity supporting NASA institutional resilience planning NASA GISS RASI Rosenzweig et al., 2014.

The RASI projections served through POWER are derived from the NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) product, which statistically downscales an ensemble of CMIP6 Global Climate Models (GCMs) to a finer spatial grid using bias-correction spatial disaggregation (BCSD). Groups of experts on the RASI team curate and create data products from NEX-GDDP so that users can explore projected changes in temperature, precipitation, and related hazards under multiple emissions pathways. The full NEX-GDDP-CMIP6 contains 35 CMIP6 climate models. The list of models used by the RASI team has been reduced to 22 models. Models that are considered outliers or have global temperature changes that are on the high end of the range of possibilities contained in the set of models ('hot' models, Hausfather et al., 2022) are not included in the final list. NEX-GDDP-CMIP6 downscales a subset of CMIP6 GCMs to 0.25° daily resolution using the BCSD algorithm, which corrects model bias against an observational reference before spatial disaggregation Thrasher et al., 2022; Wood et al., 2004.

The following models are included in the 22 model subset used by the RASI team in the creation of the products made available by POWER:

  • ACCESS-CM2
  • ACCESS-ESM1-5
  • BCC-CSM2-MR
  • CESM2
  • CMCC-ESM2
  • CNRM-CM6-1
  • CNRM-ESM2-1
  • EC-Earth3
  • FGOALS-g3
  • GFDL-ESM4
  • GISS-E2-1-G
  • INM-CM4-8
  • INM-CM5-0
  • IPSL-CM6A-LR
  • KACE-1-0-G
  • MIROC6
  • MIROC-ES2L
  • MPI-ESM1-2-HR
  • MPI-ESM1-2-LR
  • MRI-ESM2-0
  • NorESM2-LM
  • NorESM2-MM
Component Description
Upstream model ensemble CMIP6 GCMs (subset selected by NEX-GDDP)
Downscaling product NASA NEX-GDDP-CMIP6 (statistical, BCSD)
POWER-facing curator NASA GISS RASI workgroup
Temporal coverage Historical baseline (1981–2014) and projection horizons through 2100
Emissions scenarios SSP1-2.6 (low), SSP2-4.5 (intermediate), SSP3-7.0 (high)

The downscaling product, ensemble framing, and SSP scenario set in the table above are documented in the NEX-GDDP-CMIP6 and CMIP6/SSP literature Thrasher et al., 2022; Eyring et al., 2016; O'Neill et al., 2016.

Provided Variables and Indicators

RASI on POWER currently provides the following variables and indicators:

  • Temperature — projected change in mean, maximum, and minimum air temperature versus a 1981–2000 baseline (subset of total historical baseline).
  • Precipitation — projected change in mean, intensity, and frequency of precipitation.
  • Hazard indicators — workgroup-curated derived metrics (extreme heat days, heavy-precipitation events, freeze-thaw, drought, flood-relevant proxies).

Reporting projected change relative to a fixed historical baseline (rather than absolute downscaled values) is the recommended practice for impacts work because it partially cancels residual model bias Thrasher et al., 2022; IPCC AR6 WG1, 2021.

The catalog of variables and indicators is curated by the RASI workgroups (Energy Infrastructure, Extreme Weather, Wildfires & Air Quality, Flooding, Water Resources, Temperature & Precipitation, and Ecosystems) and may be expanded in future POWER releases.

Validation and Uncertainty

The RASI projections contain uncertainty from three different sources:

  1. GCM structural uncertainty — different CMIP6 models simulate the climate system with different physics and parameterizations.
  2. Internal climate variability — decadal-scale natural variability is indistinguishable from forced response on short time horizons.
  3. Scenario uncertainty — future emissions are not predictable from physics alone; they depend on policy and technology.

These three sources of uncertainty (model, internal variability, and scenario) and the relative growth of the components over time are assessed in AR6 WG1 IPCC AR6 WG1, 2021; Eyring et al., 2016. In the IPCC AR6 WG1 report, the relative magnitude of each component is estimated to be the following for global precipitation (Figure 8.23 in IPCC AR6 WG1 report):

Uncertainty Type % of 2021–2040 Global Error % of 2041–2060 Global Error % of 2061–2080 Global Error
Internal variability 41 17 5
Model Uncertainty 58 77 78
Scenario Uncertainty 2 6 17

Although NEX-GDDP is bias-corrected against an observational reference, residual systematic errors remain — particularly for precipitation extremes and in regions of complex topography.

Statistical downscaling improves mean-state agreement but cannot fully remove biases in precipitation extremes or in topographically complex regions, and it does not add physical processes absent from the parent GCM Thrasher et al., 2022; Wood et al., 2004.

Warning

RASI projections are scenario-conditional, not probabilistic forecasts. Always report alongside the SSP, the GCM ensemble (or member), the historical baseline period, and the projection horizon.

  • Long-term infrastructure resilience and adaptation planning.
  • Bracketing plausible futures for facility-level risk assessments.
  • Comparing change between a historical baseline and a future horizon (preferred over reporting absolute values from a single GCM).
  • Short-range or sub-daily weather forecasting.
  • Backfilling missing observations in the MERRA-2 / GEOS-IT time series.
  • Direct comparison to a single ground station without context (the downscaled grid is a regional climate signal, not a station-level observation).

References