The contribution of the land surface energy balance complexity to differences in means, variances and extremes using the AMIP-II methodology

Abstract

This paper explores the relationship between the complexity of the land surface energy balance parameterization and the simulation of means, variances and extremes in a climate model. We used the BMRC climate model combined with the protocol of AMIP-II to perform six ensemble simulations for each of four levels of surface energy balance complexity. Our results were then compared with other AMIP-II results in terms of the mean, variance and extremes of temperatures and precipitation. In terms of the zonally averaged mean and the maximum temperatures and precipitation, the surface energy balance complexity did not systematically affect the BMRC climate model results. The zonal minimum temperature was affected by the inclusion of tiling and/or a temporally variable canopy conductance. We found no evidence that surface energy balance complexity affected the globally- or zonally-averaged variances. Some quite large differences were identified in the probability density functions of maximum (10 K) and minimum (4 K) temperature caused by surface tiling and/or the inclusion of a time-varying canopy conductance. With these included, the model simulated a higher probability of cooler minima and warmer maxima and therefore a different diurnal temperature range. Adding interception of precipitation led to an increase in the likelihood of more extreme precipitation. Thus, provided interception, surface tiling and a time-variable stomatal conductance are included in a land surface model, the impact of other uncertainties in the parameterization of the surface energy balance are unlikely to limit the use of climate models for simulating changes in the extremes. Most published results indicating changes to precipitation and temperature extremes due toincreasing carbon dioxide are therefore unlikely to be significantly limited by uncertainty in how to parameterize the surface energy balance. Given that the variations in surface energy balance complexity included in our experiments approximates the range included in the AMIP-II models, we conclude that it this is unlikely to explain the differences found between the AMIP-II simulations. This does not mean that AMIP-II differences are not caused to a significant degree by differences in their respective LSMs, rather it limits the potential role of the land surface to non-surface energy balance components, or components (such as carbon) that are not considered here.