Differences in simulations of climate feedbacks are sources of significant divergence in climate models' temperature response to anthropogenic forcing. Snow albedo feedback is particularly critical for climate change prediction in heavily‐populated northern hemisphere land masses. Here we show its strength in current models exhibits a factor‐of‐three spread. These large intermodel variations in feedback strength in climate change are nearly perfectly correlated with comparably large intermodel variations in feedback strength in the context of the seasonal cycle. Moreover, the feedback strength in the real seasonal cycle can be measured and compared to simulated values. These mostly fall outside the range of the observed estimate, suggesting many models have an unrealistic snow albedo feedback in the seasonal cycle context. Because of the tight correlation between simulated feedback strength in the seasonal cycle and climate change, eliminating the model errors in the seasonal cycle will lead directly to a reduction in the spread of feedback strength in climate change. Though this comparison to observations may put the models in an unduly harsh light because of uncertainties in the observed estimate that are difficult to quantify, our results map out a clear strategy for targeted observation of the seasonal cycle to reduce divergence in simulations of climate sensitivity.

We developed a three‐dimensional radiative transfer model simulating solar fluxes over mountain surfaces precisely given distributions of atmospheric scatterers and absorbers. The model quantifies direct, diffuse, terrain‐reflected, and coupling (i.e., photons reflected and scattered more than once) fluxes. We applied it to a midlatitude mountainous surface to study these components' diurnal, seasonal, and geographical variability under clear skies. Domain‐averaged direct and diffuse fluxes together comprise over 96% of the flux year‐round, with diffuse fluxes' relative importance varying inversely with that of direct radiation. Direct fluxes generally account for at least 80% of the total. However, the domain‐averaged diffuse flux proportion increases to nearly 40% at high zenith angles, and approaches 100% when neighboring slopes obscure the surface from the Sun. Terrain‐reflected and coupling components each account for less than 1% throughout much of the year. However, together they comprise ∼3% when surface albedo increases during winter and are similarly nonnegligible in deep valleys all year. We also studied controls on geographical variations in flux components: The sky view factor, a conventional predictor of diffuse fluxes, is surprisingly weakly correlated with them, posing a parameterization challenge. Terrain‐reflected and coupling fluxes may be easier to parameterize given topography. Finally, we assessed shortwave errors in General Circulation Models with smoothed topography by comparing results with the mountainous surface to identical calculations for a flat surface with the same mean elevation. The differences range from 5 to 20 W/m² and arise because the atmosphere absorbs a different amount of sunshine when underlying topography is smoothed.

The primary regimes of local atmospheric variability are examined in a 6-km regional atmospheric model of the southern third of California, an area of significant land surface heterogeneity, intense topography, and climate diversity. The model was forced by reanalysis boundary conditions over the period 1995–2003. The region is approximately the same size as a typical grid box of the current generation of general circulation models used for global climate prediction and reanalysis product generation, and so can be thought of as a laboratory for the study of climate at spatial scales smaller than those resolved by global simulations and reanalysis products. It is found that the simulated circulation during the October–March wet season, when variability is most significant, can be understood through an objective classification technique in terms of three wind regimes. The composite surface wind patterns associated with these regimes exhibit significant spatial structure within the model domain, consistent with the complex topography of the region. These regimes also correspond nearly perfectly with the simulation’s highly structured patterns of variability in hydrology and temperature, and therefore are the main contributors to the local climate variability. The regimes are approximately equally likely to occur regardless of the phase of the classical large-scale modes of atmospheric variability prevailing in the Pacific–North American sector. The high degree of spatial structure of the local regimes and their tightly associated climate impacts, as well as their ambiguous relationship with the primary modes of large-scale variability, demonstrate that the local perspective offered by the high-resolution model is necessary to understand and predict the climate variations of the region.

Processes in the climate system that can either amplify or dampen the climate response to an external perturbation are referred to as climate feedbacks. Climate sensitivity estimates depend critically on radiative feedbacks associated with water vapor, lapse rate, clouds, snow, and sea ice, and global estimates of these feedbacks differ among general circulation models. By reviewing recent observational, numerical, and theoretical studies, this paper shows that there has been progress since the Third Assessment Report of the Intergovernmental Panel on Climate Change in (i) the understanding of the physical mechanisms involved in these feedbacks, (ii) the interpretation of intermodel differences in global estimates of these feedbacks, and (iii) the development of methodologies of evaluation of these feedbacks (or of some components) using observations. This suggests that continuing developments in climate feedback research will progressively help make it possible to constrain the GCMs’ range of climate feedbacks and climate sensitivity through an ensemble of diagnostics based on physical understanding and observations.

In this paper, the two factors controlling Northern Hemisphere springtime snow albedo feedback in transient climate change are isolated and quantified based on scenario runs of 17 climate models used in the Intergovernmental Panel on Climate Change Fourth Assessment Report. The first factor is the dependence of planetary albedo on surface albedo, representing the atmosphere's attenuation effect on surface albedo anomalies. It is potentially a major source of divergence in simulations of snow albedo feedback because of large differences in simulated cloud fields in Northern Hemisphere land areas. To calculate the dependence, an analytical model governing planetary albedo was developed. Detailed validations of the analytical model for two of the simulations are shown, version 3 of the Community Climate System Model (CCSM3) and the Geophysical Fluid Dynamics Laboratory global coupled Climate Model 2.0 (CM2.0), demonstrating that it facilitates a highly accurate calculation of the dependence of planetary albedo on surface albedo given readily available simulation output. In all simulations it is found that surface albedo anomalies are attenuated by approximately half in Northern Hemisphere land areas as they are transformed into planetary albedo anomalies. The intermodel standard deviation in the dependence of planetary albedo on surface albedo is surprisingly small, less than 10% of the mean. Moreover, when an observational estimate of this factor is calculated by applying the same method to the satellite-based International Satellite Cloud Climatology Project (ISCCP) data, it is found that most simulations agree with ISCCP values to within about 10%, despite further disagreements between observed and simulated cloud fields. This suggests that even large relative errors in simulated cloud fields do not result in significant error in this factor, enhancing confidence in climate models. The second factor, related exclusively to surface processes, is the change in surface albedo associated with an anthropogenically induced temperature change in Northern Hemisphere land areas. It exhibits much more intermodel variability. The standard deviation is about ⅓ of the mean, with the largest value being approximately 3 times larger than the smallest. Therefore this factor is unquestionably the main source of the large divergence in simulations of snow albedo feedback. To reduce the divergence, attention should be focused on differing parameterizations of snow processes, rather than intermodel variations in the attenuation effect of the atmosphere on surface albedo anomalies.

Milankovitch proposed that variations in the earth’s orbit cause climate variability through a local thermodynamic response to changes in insolation. This hypothesis is tested by examining variability in an atmospheric general circulation model coupled to an ocean mixed layer model subjected to the orbital forcing of the past 165 000 yr. During Northern Hemisphere summer, the model’s response conforms to Milankovitch’s hypothesis, with high (low) insolation generating warm (cold) temperatures throughout the hemisphere. However, during Northern Hemisphere winter, the climate variations stemming from orbital forcing cannot be solely understood as a local thermodynamic response to radiation anomalies. Instead, orbital forcing perturbs the atmospheric circulation in a pattern bearing a striking resemblance to the northern annular mode, the primary mode of simulated and observed unforced atmospheric variability. The hypothesized reason for this similarity is that the circulation response to orbital forcing reflects the same dynamics generating unforced variability. These circulation anomalies are in turn responsible for significant fluctuations in other climate variables: Most of the simulated orbital signatures in wintertime surface air temperature over midlatitude continents are directly traceable not to local radiative forcing, but to orbital excitation of the northern annular mode. This has paleoclimate implications: during the point of the model integration corresponding to the last interglacial (Eemian) period, the orbital excitation of this mode generates a 1°–2°C warm surface air temperature anomaly over Europe, providing an explanation for the warm anomaly of comparable magnitude implied by the paleoclimate proxy record. The results imply that interpretations of the paleoclimate record must account for changes in surface temperature driven not only by changes in insolation, but also by perturbations in atmospheric dynamics.

Climatological planetary albedo obtained from the International Satellite Cloud Climatology Project (ISCCP) D-series flux dataset is broken down into contributions from the surface and atmosphere in cryosphere regions. The atmosphere accounts for much more of climatological planetary albedo (≥75%) than the surface at all times of the year. The insignificance of the surface contribution over highly reflective cryosphere regions is attributed mostly to the damping effect of the atmosphere. The overlying atmosphere attenuates the surface’s contribution to climatological planetary albedo by reducing the number of solar photons initially reaching the surface and the number of photons initially reflected by the surface that actually reach the top of the atmosphere.

The ISCCP datasets were also used to determine the relative contributions of the surface and atmosphere to seasonal and interannual planetary albedo variability in cryosphere regions. Even damped by the atmosphere to the same degree as in the climatological case, the surface contribution dominates the variability in planetary albedo on seasonal and interannual time scales. The surface accounts for about 75% of the change in climatological planetary albedo from one season to another with similar zenith angle and more than 50% of its interannual variability at nearly all times of the year, especially during seasons with extensive snow and sea ice extent. The dominance of the surface in planetary albedo variability is because surface albedo variability associated with snow and ice fluctuations is significantly larger than atmospheric albedo variability due to cloud fluctuations. The large effect of snow and ice variations on planetary albedo variability suggests that if cloud fields do not change much in a future warmer climate, a retreat of snow cover or sea ice would lead to a significant increase in net incoming solar radiation, resulting in an enhancement of high-latitude climate sensitivity.

The depth of the planetary boundary layer (PBL) is a climatologically important quantity that has received little attention on regional to global scales. Here a 10-yr climatology of PBL depth from the University of California, Los Angeles (UCLA) atmospheric GCM is analyzed using the PBL mass budget. Based on the dominant physical processes, several PBL regimes are identified. These regimes tend to exhibit large-scale geographic organization. Locally generated buoyancy fluxes and static stability control PBL depth nearly everywhere, though convective mass flux has a large influence at tropical marine locations. Virtually all geographical variability in PBL depth can be linearly related to these quantities. While dry convective boundary layers dominate over land, stratocumulus-topped boundary layers are most common over ocean. This division of regimes leads to a dramatic land–sea contrast in PBL depth. Diurnal effects keep mean PBL depth over land shallow despite large daytime surface fluxes. The contrast arises because the large daily exchange of heat and mass between the PBL and free atmosphere over land is not present over the ocean, where mixing is accomplished by turbulent entrainment. Consistent treatment of remnant air from the deep, daytime PBL is necessary for proper representation of this diurnal behavior over land. Many locations exhibit seasonal shifts in PBL regime related to changes in PBL clouds. These shifts are controlled by seasonal variations in buoyancy flux and static stability.

A coarse resolution coupled ocean–atmosphere simulation in which surface albedo feedback is suppressed by prescribing surface albedo, is compared to one where snow and sea ice anomalies are allowed to affect surface albedo. Canonical CO₂-doubling experiments were performed with both models to assess the impact of this feedback on equilibrium response to external forcing. It accounts for about half the high-latitude response to the forcing. Both models were also run for 1000 yr without forcing to assess the impact of surface albedo feedback on internal variability. Surprisingly little internal variability can be attributed to this feedback, except in the Northern Hemisphere continents during spring and in the sea ice zone of the Southern Hemisphere year-round. At these locations and during these seasons, it accounts for, at most, 20% of the variability. The main reason for this relatively weak signal is that horizontal damping processes dilute the impact of surface albedo feedback.

When snow albedo feedback in Northern Hemisphere continents is isolated from horizontal damping processes, it has a similar strength in the CO₂-doubling and internal variability contexts; a given temperature anomaly in these regions is associated with approximately the same change in snow depth and surface albedo whether it was externally forced or internally generated. This suggests that the presence of internal variability in the observed record is not a barrier to extracting information about snow albedo feedback's contribution to equilibrium climate sensitivity. This is demonstrated in principle in a “scenario run,” where estimates of past, present, and future changes in greenhouse gases and sulfate aerosols are imposed on the model with surface albedo feedback. This simulation contains a mix of internal variations and externally forced anomalies similar to the observed record. The snow albedo feedback to the scenario run's climate anomalies agrees very well with the snow albedo feedback in the CO₂-doubling context. Moreover, the portion of the scenario run corresponding to the present-day satellite record is long enough to capture this feedback, suggesting this record could be used to estimate snow albedo feedback's contribution to equilibrium climate sensitivity.

Past research on the climate response to orbital forcing has emphasized the glacial-interglacial variations in global ice volume, global-mean temperature, and the global hydrologic cycle. This emphasis may be inappropriate in the tropics, where the response to precessional forcing is likely to be somewhat independent of the glacial-interglacial variations, particularly in variables relating to the hydrologic cycle. To illustrate this point, we use an atmospheric general circulation model coupled to a slab ocean model, performing experiments that quantify the tropical climate’s response to (1) opposite phases of precessional forcing, and (2) Last Glacial Maximum boundary conditions. While the glacially-forced tropical temperature changes are typically more than an order of magnitude larger than those arising from precessional forcing, the hydrologic signals stemming from the two forcings are comparable in magnitude. The mechanisms behind these signals are investigated and shown to be quite distinct for the precessional and glacial forcing. Because of strong dynamical linkages in the tropics, the model results illustrate the impossibility of predicting the local hydrologic response to external forcing without understanding the response at much larger spatial scales. Examples from the paleoclimate record are presented as additional evidence for the importance of precessional signals in past variations of the tropical climate.

Zonally symmetric fluctuations of the midlatitude westerly winds characterize the primary mode of atmospheric variability in the Southern Hemisphere during all seasons. This is true not only in observations but also in an unforced 15 000-yr integration of a coarse-resolution (R15) coupled ocean–atmosphere model. Here it is documented how this mode of atmospheric variability, known as the Southern Annular Mode (SAM), generates ocean circulation and sea ice variations in the model integration on interannual to centennial timescales that are tightly in phase with the SAM. The positive phase of the SAM is associated with an intensification of the surface westerlies over the circumpolar ocean (around 60°S), and a weakening of the surface westerlies farther north. This induces Ekman drift to the north at all longitudes of the circumpolar ocean, and Ekman drift to the south at around 30°S. Through mass continuity, the Ekman drift generates anomalous upwelling along the margins of the Antarctic continent, and downwelling around 45°S. The anomalous flow diverging from the Antarctic continent also increases the vertical tilt of the isopycnals in the Southern Ocean, so that a more intense circumpolar current is also closely associated with positive SAM. In addition, the anomalous divergent flow advects sea ice farther north, resulting in an increase in sea ice coverage. Finally, positive SAM drives increases in poleward heat transport at about 30°S, while decreases occur in the circumpolar region. Ocean and sea ice anomalies of the opposite sign occur when the SAM is negative. The ocean and sea ice fluctuations associated with the SAM constitute a significant fraction of simulated ocean variability poleward of 30°S year-round. The robustness of the mechanisms relating the SAM to oceanic variability suggests that the SAM is likely an important source of large-scale variability in the real Southern Hemisphere ocean.

Using two versions of the GFDL coupled ocean‐atmosphere model, one where water vapor anomalies are allowed to affect the longwave radiation calculation and one where they are not, we examine the role of water vapor feedback in internal precipitation variability and greenhouse‐gas‐forced intensification of the hydrologic cycle. Without external forcing, the experiment with water vapor feedback produces 44% more annual‐mean, global‐mean precipitation variability than the one without. We diagnose the reason for this difference: In both experiments, global‐mean surface temperature anomalies are associated with water vapor anomalies. However, when water vapor interacts with longwave radiation, the temperature anomalies are associated with larger anomalies in surface downward longwave radiation. This increases the temperature anomaly damping through latent heat flux, creating an evaporation anomaly. The evaporation anomaly, in turn, leads to an anomaly of nearly the same magnitude in precipitation. In the experiment without water vapor feedback, this mechanism is absent. While the interaction between longwave and water vapor has a large impact on the global hydrologic cycle internal variations, its effect decreases as spatial scales decrease, so water vapor feedback has only a very small impact on grid‐scale hydrologic variability. Water vapor feedback also affects the hydrologic cycle intensification when greenhouse gas concentrations increase. By the 5th century of global warming experiments where CO₂ is increased and then fixed at its doubled value, the global‐mean precipitation increase is nearly an order of magnitude larger when water vapor feedback is present. The cause of this difference is similar to the cause of the difference in internal precipitation variability: When water vapor feedback is present, the increase in water vapor associated with a warmer climate enhances downward longwave radiation. To maintain surface heat balance, evaporation increases, leading to a similar increase in precipitation. This effect is absent in the experiment without water vapor feedback. The large impact of water vapor feedback on hydrologic cycle intensification does not weaken as spatial scales decrease, unlike the internal variability case. Accurate representations of water vapor feedback are therefore necessary to simulate global‐scale hydrologic variability and intensification of the hydrologic cycle in global warming.

Temperature reconstructions from the North Atlantic region indicate frequent abrupt and severe climate fluctuations during the last glacial and Holocene periods. The driving forces for these events are unclear and coupled atmosphere–ocean models of global circulation have only simulated such events by inserting large amounts of fresh water into the northern North Atlantic Ocean. Here we report a drastic cooling event in a 15,000-yr simulation of global circulation with present-day climate conditions without the use of such external forcing. In our simulation, the annual average surface temperature near southern Greenland spontaneously fell 6–10 standard deviations below its mean value for a period of 30–40 yr. The event was triggered by a persistent northwesterly wind that transported large amounts of buoyant cold and fresh water into the northern North Atlantic Ocean. Oceanic convection shut down in response to this flow, concentrating the entire cooling of the northern North Atlantic by the colder atmosphere in the uppermost ocean layer. Given the similarity between our simulation and observed records of rapid cooling events, our results indicate that internal atmospheric variability alone could have generated the extreme climate disruptions in this region.

We examine 800-year time series of internally generated variability in both a coupled ocean-atmosphere model where water vapor anomalies are not allowed to interact with longwave radiation and one where they are. The ENSO-like phenomenon in the experiment without water vapor feedback is drastically suppressed both in amplitude and geographic extent relative to the experiment with water vapor feedback. Surprisingly, the reduced amplitude of ENSO-related sea surface temperature anomalies in the model without water vapor feedback cannot be attributed to greater longwave damping of sea surface temperature. (Differences between the two experiments in radiative feedback due to clouds counterbalance almost perfectly the differences in radiative feedback due to water vapor.) Rather, the interaction between water vapor anomalies and longwave radiation affects the ENSO-like phenomenon through its influence on the vertical structure of radiative heating: Because of the changes in water vapor associated with it, a given warm equatorial Pacific sea surface temperature anomaly is associated with a radiative heating profile that is much more gravitationally unstable when water vapor feedback is present. The warm sea surface temperature anomaly therefore results in more convection in the experiment with water vapor feedback. The increased convection, in turn, is related to a larger westerly wind-stress anomaly, which creates a larger decrease in upwelling of cold water, thereby enhancing the magnitude of the original warm sea surface temperature anomaly. In this manner, the interaction between water vapor anomalies and longwave radiation magnifies the air-sea interactions at the heart of the ENSO phenomenon; without this interaction, the coupling between sea surface temperature and wind stress is effectively reduced, resulting in smaller amplitude ENSO episodes with a more limited geographical extent.

Sea surface temperature (SST) and salinity (SSS) time series from four ocean weather stations and data from an integration of the GFDL coupled ocean-atmosphere model are analyzed to test the applicability of local linear stochastic theory to the mixed-layer ocean. According to this theory, mixed-layer variability away from coasts and fronts can be explained as a ‘red noise’ response to the ‘white noise’ forcing by atmospheric disturbances. At one weather station, Papa (northeast Pacific), this stochastic theory can be applied to both salinity and temperature, explaining the relative redness of the SSS spectrum. Similar results hold for a model grid point adjacent to Papa, where the relationships between atmospheric energy and water fluxes and actual changes in SST and SSS are what is expected from local linear stochastic theory. At the other weather stations, this theory cannot adequately explain mixed-layer variability. Two oceanic processes must be taken into account: at Panulirus (near Bermuda), mososcale eddies enhance the observed variability at high frequencies. At Mike and India (North Atlantic), variations in SST and SSS advection, indicated by the coherence and equal persistence of SST and SSS anomalies, contribute to much of the low frequency variability in the model and observations. To achieve a global perspective, TOPEX altimeter data and model results are used to identify regions of the ocean where these mechanisms of variability are important. Where mesoscale eddies are as energetic as at Panulirus, indicated by the TOPEX global distribution of sea level variability, one would expect enhanced variability on short time scales. In regions exhibiting signatures of variability similar to Mike and India, variations in SST and SSS advection should dominate at low frequencies. According to the model, this mode of variability is found in the circumpolar ocean and the northern North Atlantic, where it is associated with the irregular oscillations of the model’s thermohaline circulation.

To understand the role of water vapor feedback in unperturbed surface temperature variability, a version of the Geophysical Fluid Dynamics Laboratory coupled ocean–atmosphere model is integrated for 1000 yr in two configurations, one with water vapor feedback and one without. For all spatial scales, the model with water vapor feedback has more low-frequency (timescale ≥ 2 yr) surface temperature variability than the one without. Thus water vapor feedback is positive in the context of the model’s unperturbed variability. In addition, water vapor feedback is more effective the longer the timescale of the surface temperature anomaly and the larger its spatial scale.

To understand the role of water vapor feedback in global warming, two 500-yr integrations were also performed in which CO₂ was doubled in both model configurations. The final surface global warming in the model with water vapor feedback is 3.38°C, while in the one without it is only 1.05°C. However, the model’s water vapor feedback has a larger impact on surface warming in response to a doubling of CO₂than it does on internally generated, low-frequency, global-mean surface temperature anomalies. Water vapor feedback’s strength therefore depends on the type of temperature anomaly it affects. The authors found that the degree to which a surface temperature anomaly penetrates into the troposphere is a critical factor in determining the effectiveness of its associated water vapor feedback. The more the anomaly penetrates, the stronger the feedback. It is also shown that the apparent impact of water vapor feedback is altered by other feedback mechanisms, such as albedo and cloud feedback. The sensitivity of the results to this fact is examined.

Finally, the authors compare the local and global-mean surface temperature time series from both unperturbed variability experiments to the observed record. The experiment without water vapor feedback does not have enough global-scale variability to reproduce the magnitude of the variability in the observed global-mean record, whether or not one removes the warming trend observed over the past century. In contrast, the amount of variability in the experiment with water vapor feedback is comparable to that of the global-mean record, provided the observed warming trend is removed. Thus, the authors are unable to simulate the observed levels of variability without water vapor feedback.