2009, Jung, M., Reichstein, M., Bondeau, A.

Global, spatially and temporally explicit estimates of carbon and water fluxes derived from empirical up-scaling eddy covariance measurements would constitute a new and possibly powerful data stream to study the variability of the global terrestrial carbon and water cycle. This paper introduces and validates a machine learning approach dedicated to the upscaling of observations from the current global network of eddy covariance towers (FLUXNET). We present a new model TRee Induction ALgorithm (TRIAL) that performs hierarchical stratification of the data set into units where particular multiple regressions for a target variable hold. We propose an ensemble approach (Evolving tRees with RandOm gRowth, ERROR) where the base learning algorithm is perturbed in order to gain a diverse sequence of different model trees which evolves over time. We evaluate the efficiency of the model tree ensemble (MTE) approach using an artificial data set derived from the Lund-Potsdam-Jena managed Land (LPJmL) biosphere model. We aim at reproducing global monthly gross primary production as simulated by LPJmL from 1998–2005 using only locations and months where high quality FLUXNET data exist for the training of the model trees. The model trees are trained with the LPJmL land cover and meteorological input data, climate data, and the fraction of absorbed photosynthetic active radiation simulated by LPJmL. Given that we know the "true result" in the form of global LPJmL simulations we can effectively study the performance of the MTE upscaling and associated problems of extrapolation capacity. We show that MTE is able to explain 92% of the variability of the global LPJmL GPP simulations. The mean spatial pattern and the seasonal variability of GPP that constitute the largest sources of variance are very well reproduced (96% and 94% of variance explained respectively) while the monthly interannual anomalies which occupy much less variance are less well matched (41% of variance explained). We demonstrate the substantially improved accuracy of MTE over individual model trees in particular for the monthly anomalies and for situations of extrapolation. We estimate that roughly one fifth of the domain is subject to extrapolation while MTE is still able to reproduce 73% of the LPJmL GPP variability here. This paper presents for the first time a benchmark for a global FLUXNET upscaling approach that will be employed in future studies. Although the real world FLUXNET upscaling is more complicated than for a noise free and reduced complexity biosphere model as presented here, our results show that an empirical upscaling from the current FLUXNET network with MTE is feasible and able to extract global patterns of carbon flux variability.

2009, Calov, R., Ganopolski, A., Kubatzki, C., Claussen, M.

We investigate glacial inception and glacial thresholds in the climate-cryosphere system utilising the Earth system model of intermediate complexity CLIMBER-2, which includes modules for atmosphere, terrestrial vegetation, ocean and interactive ice sheets. The latter are described by the three-dimensional polythermal ice-sheet model SICOPOLIS. A bifurcation which represents glacial inception is analysed with two different model setups: one setup with dynamical ice-sheet model and another setup without it. The respective glacial thresholds differ in terms of maximum boreal summer insolation at 65° N (hereafter referred as Milankovitch forcing (MF)). The glacial threshold of the configuration without ice-sheet dynamics corresponds to a much lower value of MF compared to the full model. If MF attains values only slightly below the aforementioned threshold there is fast transient response. Depending on the value of MF relative to the glacial threshold, the transient response time of inland-ice volume in the model configuration with ice-sheet dynamics ranges from 10 000 to 100 000 years. Due to these long response times, a glacial threshold obtained in an equilibrium simulation is not directly applicable to the transient response of the climate-cryosphere system to time-dependent orbital forcing. It is demonstrated that in transient simulations just crossing of the glacial threshold does not imply large-scale glaciation of the Northern Hemisphere. We found that in transient simulations MF has to drop well below the glacial threshold determined in an equilibrium simulation to initiate glacial inception. Finally, we show that the asynchronous coupling between climate and inland-ice components allows one sufficient realistic simulation of glacial inception and, at the same time, a considerable reduction of computational costs.

2009, Kawa, S.R., Stolarski, R.S., Newman, P.A., Douglass, A.R., Rex, M., Hofmann, D.J., Santee, M.L., Frieler, K.

The impact and significance of uncertainties in model calculations of stratospheric ozone loss resulting from known uncertainty in chemical kinetics parameters is evaluated in trajectory chemistry simulations for the Antarctic and Arctic polar vortices. The uncertainty in modeled ozone loss is derived from Monte Carlo scenario simulations varying the kinetic (reaction and photolysis rate) parameters within their estimated uncertainty bounds. Simulations of a typical winter/spring Antarctic vortex scenario and Match scenarios in the Arctic produce large uncertainty in ozone loss rates and integrated seasonal loss. The simulations clearly indicate that the dominant source of model uncertainty in polar ozone loss is uncertainty in the Cl2O 2 photolysis reaction, which arises from uncertainty in laboratory-measured molecular cross sections at atmospherically important wavelengths. This estimated uncertainty in JCl 2O2 from laboratory measurements seriously hinders our ability to model polar ozone loss within useful quantitative error limits. Atmospheric observations, however, suggest that the Cl2O2 photolysis uncertainty may be less than that derived from the lab data. Comparisons to Match, South Pole ozonesonde, and Aura Microwave Limb Sounder (MLS) data all show that the nominal recommended rate simulations agree with data within uncertainties when the Cl2O2 photolysis error is reduced by a factor of two, in line with previous in situ ClOx measurements. Comparisons to simulations using recent cross sections from Pope et al. (2007) are outside the constrained error bounds in each case. Other reactions producing significant sensitivity in polar ozone loss include BrO + ClO and its branching ratios. These uncertainties challenge our confidence in modeling polar ozone depletion and projecting future changes in response to changing halogen emissions and climate. Further laboratory, theoretical, and possibly atmospheric studies are needed.

2009, Schepanski, K., Tegen, I., MacKe, A.

We present a study of Saharan dust export towards the tropical North Atlantic using the regional dust emission, transport and deposition model LM-MUSCAT. Horizontal and vertical distribution of dust optical thickness, concentration, and dry and wet deposition rates are used to describe seasonality of dust export and deposition towards the eastern Atlantic for three typical months in different seasons. Deposition rates strongly depend on the vertical dust distribution, which differs with seasons. Furthermore the contribution of dust originating from the Bod́eĺe Depression to Saharan dust over the Atlantic is investigated. A maximum contribution of Bod́eĺe dust transported towards the Cape Verde Islands is evident in winter when the Bod́eĺe source area is most active and dominant with regard to activation frequency and dust emission. Limitations of using satellite retrievals to estimate dust deposition are highlighted.