Sensitivities of the net heat flux into the deep ocean (Q_net) and of the deep-ocean heat content (DOC) below 700 m are studied using an ocean general circulation model and its adjoint. Both are found to have very similar sensitivities. The sensitivity to the surface freshwater flux (E-P-R) is positive in the Atlantic but negative in the Pacific and Southern Ocean. A positive sensitivity to the downward net surface heat flux is found only in the North Atlantic north of 40°N and the Southern Ocean. The diapycnal diffusivity of temperature affects Q_net and DOC positively in a large area of the tropics and subtropics in both the Pacific and Atlantic Oceans. The isopycnal diffusivity contributes to Q_net and DOC mainly in the Southern Ocean. Detailed analysis indicates that the surface freshwater flux affects Q_net and DOC by changing vertical velocity, temperature stratification, and overturning circulation. The downward net surface heat flux appears to increase Q_net and DOC by strengthening vertical advection and isopycnal mixing. The contribution of isopycnal diffusivity to Q_net and DOC is largely associated with the vertical heat flux due to isopycnal mixing. Similarly, the diapycnal diffusivity of temperature modulates Q_net and DOC through the downward heat flux due to diapycnal diffusion. The uncertainties of Q_net and DOC are estimated on the basis of the sensitivities and error bars of observed surface forcing and oceanic diffusivities. For DOC they are about 0.7 K (1 K = 3.7 × 10²⁴J) for the isopycnal diffusivity, 0.4 K for the diapycnal diffusivity of temperature, 0.3 K for the surface freshwater flux, and 0.1 K for the net surface heat flux and zonal wind stress. Our results suggest that the heat uptake by ocean general circulation models in climate experiments is sensitive to the isopycnal diffusivity as well to the diapycnal thermal diffusivity.

© 2003 American Geophysical Union

The MIT 2D climate model is used to make probabilistic projections for changes in global mean surface temperature and for thermosteric sea level rise under a variety of forcing scenarios. The uncertainties in climate sensitivity and rate of heat uptake by the deep ocean are quantified by using the probability distributions derived from observed 20th century temperature changes. The impact on climate change projections of using the smallest and largest estimates of 20th century deep ocean warming is explored. The impact is large in the case of global mean thermosteric sea level rise. In the MIT reference ("business as usual") scenario the median rise by 2100 is 27 and 43 cm in the respective cases. The impact on increases in global mean surface air temperature is more modest, 4.9 C and 3.9 C in the two respective cases, because of the correlation between climate sensitivity and ocean heat uptake required by 20th century surface and upper air temperature changes. The results are also compared with the projections made by the IPCC AR4's multi-model ensemble for several of the SRES scenarios. The multi-model projections are more consistent with the MIT projections based on the largest estimate of ocean warming. However the range for the rate of heat uptake by the ocean suggested by the lowest estimate of ocean warming is more consistent with the range suggested by the 20th century changes in surface and upper air temperatures, combined with expert prior for climate sensitivity.

Part I: The diapycnal diffusivity in the ocean is one of the least known parameters in current climate models. Measurements of this diffusivity are sparse and insufficient for compiling a global map. Inferences from inverse methods and energy budget calculations suggest as much as a factor of 5 difference in the global mean value of the diapycnal diffusivity. Yet, the climate is extremely sensitive to the diapycnal diffusivity. In this paper we study the sensitivity of the current climate to the diapycnal diffusivity using a coupled model with a 3-dimensional global ocean component with idealized geometry. In a subsequent paper we analyze the sensitivity of the climate change to the same parameter.
      Our results show that, at equilibrium, the strength of the thermohaline circulation in the North Atlantic scales with the 0.44 power of the diapycnal diffusivity, in contrast to the theoretical value of 2/3. On the other hand, the strength of the circulation in the South Pacific scales with the 0.63 power of the diapycnal diffusivity. The implication is that the amount of water upwelling from the deep ocean may be regulated by the diapycnal diffusion in the Indo-Pacific Ocean.
     The vertical heat balance in the ocean is controlled by: in the downward direction, (i) advection and (ii) diapycnal diffusion; in the upward direction, (iii) isopycnal diffusion and (iv) bolus velocity (GM) advection. The size of the latter three fluxes increases with diapycnal diffusivity, because the thickness of the thermocline also increases with diapycnal diffusivity leading to greater isopycnal slopes at high latitudes, and hence enhanced isopycnal diffusion and GM advection. Thus larger diapycnal diffusion is compensated for by changes in isopycnal diffusion and GM advection. Little change is found for the advective flux because of compensation between downward and upward advection.
     We present sensitivity results for the hysteresis curve of the thermohaline circulation. The stability of the climate system to slow freshwater perturbations is reduced as a consequence of a smaller diapycnal diffusivity. This result confirms the findings of 2-dimensional climate models. However, contrary to the results of these studies, a common threshold for the shutdown of the thermohaline circulation is not found in our model.

Part II: The sensitivity of the transient climate to the diapycnal diffusivity in the ocean is studied for a global warming scenario in which CO2 increases by 1% per year for 75 years. The thermohaline circulation slows down for about 100 years and recovers afterward, for any value of the diapycnal diffusivity. The rates of slowdown and of recovery, as well as the percentage recovery of the circulation at the end of 1000-year integrations, are variable but a direct relation with the diapycnal diffusivity cannot be found. At year 70 (when CO2, has doubled) an increase of the diapycnal diffusivity from 0.1 cm²/s to 1.0 cm²/s leads to a decrease in surface air temperature of about 0.4 K and an increase in sea level rise of about 4 cm The steric height gradient is divided into thermal component and haline component. It appears that, in the first 60 years of simulated global warming, temperature variations dominate the salinity ones in weakly diffusive models, whereas the opposite occurs in strongly diffusive models.
      The analysis of the vertical heat balance reveals that deep ocean heat uptake is due to reduced upward isopycnal diffusive flux and Gent-McWilliams advective flux. Surface warming, induced by enhanced CO2 in the atmosphere, leads to a reduction of the isopycnal slope which translates into a reduction of the above fluxes. The amount of reduction is directly related to the magnitude of the isopycnal diffusive flux and GM advective flux at equilibrium. These latter fluxes depend on the thickness of the thermocline at equilibrium, hence on the diapycnal diffusion. Thus, the increase of deep-ocean heat uptake with diapycnal diffusivity is an indirect effect that the latter parameter has on the isopycnal diffusion and GM advection.

© 2005 American Meteorological Society

This paper investigates the sensitivity of regional hydrology to climate change using a physically based model. The model partitions precipitation into surface runoff, groundwater runoff, and evapotranspiration by describing these fluxes first at the local instantaneous scale and then integrating over spatial and temporal distributions of soil saturation, precipitation, and wet environment evapotranspiration to calculate basin-wide climatic mean fluxes and soil saturation. The sensitivities of the mean fluxes are calculated by changing the mean precipitation and wet environment evapotranspiration. The model is applied to the Illinois River basin, and the impact of the basin's characteristics on the sensitivities is studied. For a relatively broad range of conditions the runoff processes tend to amplify climate change signals in precipitation and wet environment evapotranspiration, while evapotranspiration processes tend to dampen the same signals. These results indicate that it may be easier to detect climate changes in runoff measurements than in precipitation measurements.

© 2005 American Geophysical Union