Charles S. Jackson

Error Reduction and Convergence in Climate Prediction

Abstract Although climate models have steadily improved their ability to reproduce the observed climate, over the years there has been little change to the wide range of sensitivities exhibited by different models to a doubling of atmospheric CO2 concentrations. Stochastic optimization is used to mimic how six independent climate model development efforts might use the same atmospheric general circulation model, set of observational constraints, and model skill criteria to choose different settings for parameters thought to be important sources of uncertainty related to clouds and convection. Each optimized model improved its skill with respect to observations selected as targets of model development. Of particular note were the improvements seen in reproducing observed extreme rainfall rates over the tropical Pacific, which was not specifically targeted during the optimization process. As compared to the default model sensitivity of 2.4°C, the ensemble of optimized model configurations had a larger and n...

An Efficient Stochastic Bayesian Approach to Optimal Parameter and Uncertainty Estimation for Climate Model Predictions

Abstract One source of uncertainty for climate model predictions arises from the fact that climate models have been optimized to reproduce observational means. To quantify the uncertainty resulting from a realistic range of model configurations, it is necessary to estimate a multidimensional probability distribution that quantifies how likely different model parameter combinations are, given knowledge of the uncertainties in the observations. The computational cost of mapping a multidimensional probability distribution for a climate model using traditional means (e.g., Monte Carlo or Metropolis/Gibbs sampling) is impractical, requiring 104–106 model evaluations for problems involving less than 10 parameters. This paper examines whether such a calculation is more feasible using a particularly efficient but approximate algorithm called Bayesian stochastic inversion, based on multiple very fast simulated annealing (VFSA). Investigated here is how the number of model parameters, natural variability, and the d...

Improving land-surface model hydrology: Is an explicit aquifer model better than a deeper soil profile?

We use Monte Carlo analysis to show that explicit representation of an aquifer within a land-surface model (LSM) decreases the dependence of model performance on accurate selection of subsurface hydrologic parameters. Within the National Center for Atmospheric Research Community Land Model (CLM) we evaluate three parameterizations of vertical water flow: (1) a shallow soil profile that is characteristic of standard LSMs; (2) an extended soil profile that allows for greater variation in terrestrial water storage; and (3) a lumped, unconfined aquifer model coupled to the shallow soil profile. North American Land Data Assimilation System meteorological forcing data (1997–2005) drive the models as a single column representing Illinois, USA. The three versions of CLM are each run 22,500 times using a random sample of the parameter space for soil texture and key hydrologic parameters. Other parameters remain constant. Observation-based monthly changes in state-averaged terrestrial water storage (dTWS) are used to evaluate the model simulations. After single-criteria parameter exploration, the schemes are equivalently adept at simulating dTWS. However, explicit representation of groundwater considerably decreases the sensitivity of modeled dTWS to errant parameter choices. We show that approximate knowledge of parameter values is not sufficient to guarantee realistic model performance: because interaction among parameters is significant, they must be prescribed as a congruent set.

Insights into spatial sensitivities of ice mass response to environmental change from the SeaRISE ice sheet modeling project I: Antarctica

Sophie Nowicki, Robert A. Bindschadler, Ayako Abe-Ouchi, Andy Aschwanden, Ed Bueler, Hyeungu Choi, Jim Fastook, Glen Granzow, Ralf Greve, Gail Gutowski, Ute Herzfeld, Charles Jackson, Jesse Johnson, Constantine Khroulev, Eric Larour, Anders Levermann, William H. Lipscomb, Maria A. Martin, Mathieu Morlighem, Byron R. Parizek, David Pollard, Stephen F. Price, Diandong Ren, Eric Rignot, Fuyuki Saito, Tatsuru Sato, Hakime Seddik, Helene Seroussi, Kunio Takahashi, Ryan Walker, and Wei Li Wang

Reliability of multi-model and structurally different single-model ensembles

The performance of several state-of-the-art climate model ensembles, including two multi-model ensembles (MMEs) and four structurally different (perturbed parameter) single model ensembles (SMEs), are investigated for the first time using the rank histogram approach. In this method, the reliability of a model ensemble is evaluated from the point of view of whether the observations can be regarded as being sampled from the ensemble. Our analysis reveals that, in the MMEs, the climate variables we investigated are broadly reliable on the global scale, with a tendency towards overdispersion. On the other hand, in the SMEs, the reliability differs depending on the ensemble and variable field considered. In general, the mean state and historical trend of surface air temperature, and mean state of precipitation are reliable in the SMEs. However, variables such as sea level pressure or top-of-atmosphere clear-sky shortwave radiation do not cover a sufficiently wide range in some. It is not possible to assess whether this is a fundamental feature of SMEs generated with particular model, or a consequence of the algorithm used to select and perturb the values of the parameters. As under-dispersion is a potentially more serious issue when using ensembles to make projections, we recommend the application of rank histograms to assess reliability when designing and running perturbed physics SMEs.

The Importance of Atmospheric Dynamics in the Northern Hemisphere Wintertime Climate Response to Changes in the Earth's Orbit

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.

Computational methods for parameter estimation in climate models

Intensive computational methods have been used by Earth scientists in a wide range of problems in data inversion and uncertainty quantication such as earthquake epicenter location and climate projections. To quantify the uncer- tainties resulting from a range of plausible model congurations it is necessary to estimate a multidimensional probability distribution. The computational cost of estimating these distributions for geoscience applications is impractical using traditional methods such as Metropolis/Gibbs algorithms as simulation costs limit the number of experiments that can be obtained reasonably. Several alternate sampling strategies have been proposed that could improve on the sampling e- ciency including Multiple Very Fast Simulated Annealing (MVFSA) and Adaptive Metropolis algorithms. The performance of these proposed sampling strategies are evaluated with a surrogate climate model that is able to approximate the noise and response behavior of a realistic atmospheric general circulation model (AGCM). The surrogate model is fast enough that its evaluation can be embed- ded in these Monte Carlo algorithms. We show that adaptive methods can be superior to MVFSA to approximate the known posterior distribution with fewer forward evaluations. However the adaptive methods can also be limited by inad- equate sample mixing. The Single Component and Delayed Rejection Adaptive Metropolis algorithms were found to resolve these limitations, although challenges remain to approximating multi-modal distributions. The results show that these advanced methods of statistical inference can provide practical solutions to the cli- mate model calibration problem and challenges in quantifying climate projection uncertainties. The computational methods would also be useful to problems out- side climate prediction, particularly those where sampling is limited by availability of computational resources.

Sensitivity of stationary wave amplitude to regional changes in Laurentide ice sheet topography in single-layer models of the atmosphere

Climate variability on millennial timescales has been observed in many geologic records covering the last glacial cycle. A potential source of this variability is the Laurentide ice sheet (LIS) in its periodic discharge of large quantities of icebergs to the North Atlantic. The present analysis considers whether regional variations in LIS topography could exert a significant influence on the atmospheres stationary wave circulation. The maximum effect that regional changes in LIS topography have on the atmospheres stationary wave circulation is determined using single-layer models of the atmosphere. Model experiments measure the individual contribution of 4.5°×7.5° sections of the LIS and Greenland topography to global mean stationary wave amplitude. Results show the possibility for a limited region of topography to control a disproportionate amount of the atmospheres total response to topography. Moreover, the possibility exists for a reduction in topographic forcing to increase stationary wave amplitude. These results can be understood by considering how the mean flow controls the horizontal propagation of wave energy and superposition of wave amplitude. The location of regions with enhanced stationary wave sensitivity to topographic alteration is found to be sensitive to mean topographic height but not mean wind strength. The latter is found to primarily affect the overall amplitude of sensitivity rather than the pattern. The impact of two hypothetical changes in LIS topography is considered, and they are found to have widely different effects on the global stationary wave field. Stationary wave sensitivity to topography within the single-layer models suggests that variations in the size or shape of the LIS can be one factor important to climate variability on millennial timescales.

A multivariate empirical‐orthogonal‐function‐based measure of climate model performance

[1] A measure of the average distance between climate model predictions of multiple fields and observations has been developed that is based on the use of empirical orthogonal functions (EOFs). The application of EOFs provides a means to use information about spatial correlations in natural variability to provide a more balanced view of the significance of changes in model predictions across multiple fields, seasons, and regions. A comparison is made between the EOF-based measure and measures that are normalized by grid point variance and spatial variance for changes in the National Center for Atmospheric Research Community Climate Model, Version 3.10 (CCM3.10), parameter controlling initial cloud downdraft mass flux (ALFA), an important parameter within the Zhang and McFarlane [1995] convection scheme. All measures present consistent views that increasing ALFA from its default value creates significant improvements in precipitation, shortwave radiation reaching the surface, and surface latent heat fluxes at the expense of degrading predictions of total cloud cover, near-surface air temperature, net shortwave radiation at the top of the atmosphere, and relative humidity. However, the relative importance of each of these changes, and therefore the average view of the change in model performance, is significantly impacted by the details of how each measure of model performance handles regions with little or no internal variability. In general, the EOF-based measure emphasizes regions where modeledobservational differences are large, excluding those regions where internal variability is small. INDEX TERMS: 3309 Meteorology and Atmospheric Dynamics: Climatology (1620); 3314 Meteorology and Atmospheric Dynamics: Convective processes; 3337 Meteorology and Atmospheric Dynamics: Numerical modeling and data assimilation; 3394 Meteorology and Atmospheric Dynamics: Instruments and techniques; KEYWORDS: climate prediction, skill scores, numerical modeling

Reliability and importance of structural diversity of climate model ensembles

We investigate the performance of the newest generation multi-model ensemble (MME) from the Coupled Model Intercomparison Project (CMIP5). We compare the ensemble to the previous generation models (CMIP3) as well as several single model ensembles (SMEs), which are constructed by varying components of single models. These SMEs range from ensembles where parameter uncertainties are sampled (perturbed physics ensembles) through to an ensemble where a number of the physical schemes are switched (multi-physics ensemble). We focus on assessing reliability against present-day climatology with rank histograms, but also investigate the effective degrees of freedom (EDoF) of the fields of variables which makes the statistical test of reliability more rigorous, and consider the distances between the observation and ensemble members. We find that the features of the CMIP5 rank histograms, of general reliability on broad scales, are consistent with those of CMIP3, suggesting a similar level of performance for present-day climatology. The spread of MMEs tends towards being “over-dispersed” rather than “under-dispersed”. In general, the SMEs examined tend towards insufficient dispersion and the rank histogram analysis identifies them as being statistically distinguishable from many of the observations. The EDoFs of the MMEs are generally greater than those of SMEs, suggesting that structural changes lead to a characteristically richer range of model behaviours than is obtained with parametric/physical-scheme-switching ensembles. For distance measures, the observations and models ensemble members are similarly spaced from each other for MMEs, whereas for the SMEs, the observations are generally well outside the ensemble. We suggest that multi-model ensembles should represent an important component of uncertainty analysis.