Climate Modelling, Climate Prediction and Model Validation W.L. GATES, J.F.B. MITCHELL, G.J. BOER, U. CUBASCH, V.P. MELESHKO Contributors: D. Anderson; W. Broecker; D. Cariolle; H. Cattle; R.D. Cess; F. Giorgi; M.I. Hojfert; B.G. Hunt; A. Kitoh; P. Lemke; H. Le Treut; R.S. Lindzen; S. Manabe; B.J. McAvaney; L. Mearns; G.A. Meehl; J.M. Murphy; T.N. Palmer; A.B. Pittock; K. Puri; D.A. Randall; D. Rind; PR. Rowntree; M.E. Schlesinger; C.A. Senior; I.H. Simmonds; R. Stoujfer; S. Tibaldi; T. Tokioka; G. Visconti; J.E. Walsh; W.-C. Wang; D. Webb CONTENTS Executive Summary 101 B4.2 Diurnal Cycle 119 B4.3 Day-to-Day Variability 119 Bl Introduction 103 B4.4 Extreme Events 119 B2 Advances in Modelling Climate Change due to B4.5 Blocking and Storm Activity 119 Increased Greenhouse Gases 103 B4.6 Inlra-Seasonal Variability 120 B2.1 Introduction 103 B4.7 Interannual Variability 120 B2.2 New Transient Results from Coupled B4.8 ENSO and Monsoons 121 Atmosphere-Ocean GCMs 103 B4.9 Decadal Variability 122 B2.3 New Equilibrium Results from Atmospheric B5 Advances in Modelling the Oceans and Sea-Ice 122 GCMs and Mixed-Layer Ocean Models 108 B5.1 Introduction 122 B2.4 Regional Climate Simulations 112 B5.2 Eddy-Resolving Models 123 B3 Advances in Analysis of Climate Feedbacks and B5.3 Thermohaline Circulation 124 Sensitivity 114 B5.4 Sea-Ice Models 125 B3.1 Introduction 114 B3.2 Water Vapour Feedback 114 B6 Advances in Model Validation 126 B6.1 Introduction 126 B3.3 Cloud Feedback 116 B6.2 Systematic Errors and Model Intercomparison 126 B3.4 Surface Albedo Feedback 117 B6.3 Data for Model Validation 129 B3.5 Climate Sensitivity 118 B4 Advances in Modelling Atmospheric Variability 118 References 129 B4.1 Introduction 119 EXECUTIVE SUMMARY Coupled Model Experiments the equilibrium sensitivity to doubled CO2 from the range The new transient climate simulations with coupled of 1.5 to 4.5°C as given by IPCC 1990. atmosphere-ocean general circulation models (GCMs) • There is no compelling evidence that water vapour generally confirm the findings of Section 6 in the 1990 feedback is anything other than the positive feedback it has Intergovernmental Panel on Climate Change report (IPCC, been conventionally thought to be, although there may be 1990), although the number of coupled model simulations difficulties with the treatments of upper-level water vapour is still small. in current models. • There is broad agreement among the four current models • The effects of clouds remain a major area of uncertainty in in the simulated large-scale patterns of change and in their the modelling of climate change. While the treatment of temporal evolution. clouds in GCMs is becoming more complex, a clear • The large-scale patierns of temperature change remain understanding of the consequences of different cloud essentially fixed with time, and they become more evident parametrizations has not yet emerged. with longer averaging intervals and as the simulations progress. Atmospheric Variability • The large-scale patterns of change are similar to those • Model experiments with doubled CO2 give no clear obtained in comparable equilibrium experiments except indication of a systematic change in the variability of that the warming is retarded in the southern high latitude temperature on daily to interannual time-scales, while ocean and the northern North Atlantic Ocean where deep changes of variability for other climate features appear to water is formed. be regionally (and possibly model) dependent. All but one of the models show slow initial warming (which may be an artefact of the experimental design) Ocean Modelling followed by a nearly linear trend of approximately 0.3°C • Results from eddy-resolving ocean models show a broadly per decade. realistic portrayal of oceanic variability, although the All models simulate a peak-to-trough natural variability of climatic role of eddies remains unclear. about 0.3°C in global surface air temperature on decadal Ocean-only GCMs show considerable sensitivity of the time-scales. thermohaline circulation on decadal and longer time-scales • The rate of sea level rise due to thermal expansion to changes in the surface fresh-water flux, although increases with time to between 2 and 4cm/decade at the coupled models may be less sensitive. time of doubling of equivalent CO2. • Sea-ice dynamics may play an important role in the freezing process, and should therefore be included in Regional Changes models for the simulation of climate change under • Although confidence in the regional changes simulated by increased CO2 GCMs remains low, progress in the simulation of regional climate is being obtained with both statistical and one-way Model Validation nested model techniques. In both cases the quality of the • There have been improvements in the accuracy of large-scale flow provided by the GCM is critical. individual atmospheric and oceanic GCMs, although the ranges of intermodel error and sensitivity remain large. Climate Feedbacks and Sensitivity • The lack of adequate observational data remains a serious There is no compelling new evidence to warrant changing impediment to climate model improvement. B Climate Modelling, Climate Prediction & Model Validation 103 Bl Introduction B2 Advances in Modelling Climate Change due to Increased Greenhouse Gases Since the publication of the first IPCC Scientific B2.1 Introduction Assessment of Climate Change (IPCC, 1990) there have Simulation of the climatic response to increases in been significant advances in many areas of climate atmospheric greenhouse gases has continued to dominate research as part of a continuing worldwide acceleration of climate modelling. Preliminary results from new interest in the assessment of possible anthropogenic integrations of coupled global atmosphere-ocean models climate changes. In this section we concentrate on with progressive increases of CO2 show that the patterns of advances in the modelling of climate change due to the transient response are similar to those in an equilibrium increased greenhouse gases, improvements in the analysis response, except over the high-latitude southern ocean and of climate processes and feedbacks, and on advances in northern North Atlantic ocean; here the delayed warming climate model validation that were not available to the has highlighted the critical role of the oceanic IPCC in early 1990. This section is thus intended as an thermohaline circulation. Computing limitations have update to selected portions of the 1990 assessment rather continued to restrict the resolution that can be used in than as a comprehensive revision, and an effort has been GCM simulations of the climate changes due to increased made to keep the discussion both concise and focussed in greenhouse gases, although progress is being made in the accordance with the stringent space limitations placed simulation of regional climate by both statistical upon this supplementary report. techniques and by locally nesting a higher-resolution In Section B2 recent simulations of climate change are model within a global GCM. assessed, with emphasis on results from coupled ocean- atmosphere models. In Section B3, recent research on B2.2 New Transient Results from Coupled Atmosphere- modelling climate feedbacks is discussed, and the IPCC Ocean GCMs 1990 estimates of chmate sensitivity are reviewed. The At the time of the 1990 IPCC report, preliminary results simulation of atmospheric variability and its changes due were available from only two coupled model integrations to increased atmospheric CO2 are discussed in Section B4 with transient CO2, namely those made at NCAR while developments in ocean and sea-ice modelling are (Washington and Meehl, 1989) and at GFDL (Stouffer et presented in Section B5. Finally, Section B6 discusses al, 1989). Of these only the GFDL integration had been advances in climate model validation. carried to the point of CO2 doubling (which occurred after Table Bl: Summary of transient CO2 experiments with coupled ocean-atmosphere GCMs GFDL MPI NCAR UKMO AGCM R15L9 T21 L19 R15L9 2.5° X 3.75° LI 1 OGCM 4.5° X 3.75° L12 4°L11 5°L4 2.5° X 3.75° LI7 Features no diurnal cycle. prognostic CLW, quasi- prognostic CLW, isopycnal ocean geostrophic ocean isopycnal ocean diffusion diffusion Flux adjustment seasonal, heat, fresh seasonal, heat, fresh none seasonal, heat, fresh water water, wind stress water Control CO2 (ppm) 300 390 t 330 323 CO2 (t) l%yr-' IPCCa&d(1990 1% yr' l%yr-l (compound) Scenarios), 2XCO2 (linear) (compound) Length (yr) 100 100 60 75 CO2 doubling time (yr) 70 60 (IPCCa) 100 70 Warming (°C) at COj 2.3 1.3 2.3 1.7 doubling (projected) 2xC02 sensitivity (°C) 4.0 2.6 4.5 2.7 tt (with mixed layer ocean) L - number of vertical levels; R - number of spectral waves (rhomboidal truncation); T - no spectral waves (triangular truncation), t - equivalent CO2 value for trace gases ++ - estimate from low resolution experiment 104 Climate Modelling, Climate Prediction & Model Validation B approximately 70 years as a result of a 1% per year compound increase of CO2). Here we present further 4 resuhs from the transient CO2 experiments with both the • GFDL GFDL and NCAR coupled models, along with preliminary • MPI results from new transient CO2 integrations recently — — UKMO completed at the Max-Planck-Institute for Meteorology NCAR 0) —— IPCC A (MPI) in Hamburg and at the Hadley Centre of the United S 1 - Kingdom Meteorological Office (UKMO). A fuller ^ . E description and intercomparison of these results is being ^ 0- prepared (WCRP, 1992). -1 A summary of the new transient results is presented in 0 20 40 60 80 100 Table Bl. Note that there are differences in the CO2 Year equivalent doubling times (by a factor of almost two) and Figure Bl: Decadal mean changes in globally averaged surface that the equilibrium sensitivity of the models is different.