AMIP

AMIP
Name	Atmospheric Model Intercomparison Project
Abbreviation	AMIP
Established	1990
Discipline	Climate science
Type	International research project
Parent organization	World Climate Research Programme

AMIP The Atmospheric Model Intercomparison Project is an international coordinated experiment that evaluates atmospheric general circulation models against observed sea surface temperatures and sea ice conditions. Designed to benchmark model performance, foster model development, and support intercomparison across institutions, AMIP has intersected with major programs and assessments in Intergovernmental Panel on Climate Change, World Climate Research Programme, Coupled Model Intercomparison Project, Global Climate Models and operational centers like National Center for Atmospheric Research, Met Office, and European Centre for Medium-Range Weather Forecasts. The project produced standardized datasets and protocols widely cited in assessments such as the IPCC Third Assessment Report and activities within Climate Model Intercomparison Project Phase 5.

Overview

AMIP provides a stabilized framework for testing atmospheric components of climate models by prescribing historical sea surface temperatures and sea ice fields, enabling direct comparison among models from diverse institutions. It connects to initiatives and datasets produced by National Aeronautics and Space Administration, National Oceanic and Atmospheric Administration, International Satellite Cloud Climatology Project, Hadley Centre, and observational programs like ARGO and Global Surface Temperature. The experiment emphasizes reproducibility and has been central to evaluation activities at agencies including European Space Agency, Japan Meteorological Agency, Canadian Centre for Climate Modelling and Analysis, and academic groups at Massachusetts Institute of Technology, Columbia University, and Scripps Institution of Oceanography.

History and development

Conceived in the late 1980s, AMIP arose from community needs articulated in workshops sponsored by World Climate Research Programme and the International Geosphere-Biosphere Programme. Early planning involved collaborations among GFDL, HadCM3 developers, and centers such as Geophysical Fluid Dynamics Laboratory, Lawrence Livermore National Laboratory, and Max Planck Institute for Meteorology. The original protocol published in 1995 set the stage for multi-model intercomparison used by the IPCC Second Assessment Report and subsequent model evaluation activities. Successive iterations influenced the development of coupled experiments in CMIP and informed model improvements at institutions like Princeton University and University of Reading.

Experimental design and protocol

AMIP prescribes a fixed set of boundary conditions: observed monthly sea surface temperatures and sea ice concentrations, following datasets compiled by National Oceanic and Atmospheric Administration, Met Office Hadley Centre, and ERSST reconstructions. Model participants run atmospheric general circulation models for multi-decadal periods, typically from 1979 onwards to take advantage of satellite-era observations from missions such as TIROS-N, ERS-1, and Nimbus. Output variables and diagnostic fields are standardized to facilitate comparison with observations from ARGO, Tropical Rainfall Measuring Mission, CloudSat, and reanalyses like ERA-Interim and NCEP/NCAR Reanalysis. The protocol specifies temporal resolution, vertical levels, and archived fields for variables including radiation fluxes, precipitation, temperature, and circulation indices measured against phenomena like the El Niño–Southern Oscillation and the North Atlantic Oscillation.

Participating models and institutions

AMIP attracted participation from a wide array of modeling centers, spanning national laboratories, universities, and meteorological agencies. Notable contributors have included Geophysical Fluid Dynamics Laboratory, Hadley Centre, Max Planck Institute for Meteorology, European Centre for Medium-Range Weather Forecasts, Japan Agency for Marine-Earth Science and Technology, CSIRO, Canadian Centre for Climate Modelling and Analysis, NOAA GFDL, Institute Pierre-Simon Laplace, Princeton University, University of Washington, and Columbia University. Models ranged from early spectral dynamical cores to more recent finite-volume and cubed-sphere dynamical cores developed at institutions like NCAR and MIT, each submitting standardized output to repositories overseen by coordinating groups within WCRP.

Key findings and scientific impact

AMIP provided systematic evaluation of atmospheric response to surface forcing, revealing robust biases common across models—such as tropical precipitation distribution errors, cloud-radiative feedback uncertainties, and jet-stream position biases—issues also highlighted in assessments by IPCC, National Research Council, and reviewers at Royal Society. AMIP comparisons helped quantify model spread in simulating phenomena like El Niño–Southern Oscillation, monsoon variability in regions examined by Indian Institute of Tropical Meteorology, and extratropical storm-track dynamics studied at Lamont–Doherty Earth Observatory. The experiment informed parameterization improvements in convection, cloud microphysics, and radiation at groups including NCAR, GFDL, and Met Office, and its diagnostics were incorporated into evaluation frameworks used by CMIP5 and CMIP6.

Data access and usage

AMIP output and associated observational datasets have been archived at international data centers and made available to researchers via portals managed by World Data Center for Climate, PCMDI, ESGF, and national archives at NOAA National Centers for Environmental Information. Users have leveraged AMIP data for model tuning, development of emergent constraints cited by Nature and Science publications, and for training machine-learning emulators at institutions such as Google DeepMind collaborations and university research groups. AMIP datasets remain a reference for comparing atmospheric-only model performance against coupled experiments in CMIP.

Limitations and future directions

While AMIP isolates atmospheric response to prescribed surface forcing, it omits ocean-atmosphere feedbacks central to phenomena studied by Sverdrup, Bjerknes, and coupled-model initiatives like CMIP6. Limitations include dependence on observed SST forcing quality, inadequate representation of aerosol interactions documented by Aerosols and Climate studies, and challenges reproducing decadal variability highlighted by Pacific Decadal Oscillation research. Future directions emphasize integrating higher-resolution models developed at Lawrence Berkeley National Laboratory and Los Alamos National Laboratory, coupling with interactive chemistry modules from NCAR and NASA Goddard Institute for Space Studies, and aligning AMIP-like protocols with cloud-resolving experiments at facilities such as ARM Climate Research Facility and community initiatives coordinated by WCRP.

Category:Climate modeling