Doug Nychka

Quantifying Uncertainty in Projections of Regional Climate Change: A Bayesian Approach to the Analysis of Multimodel Ensembles

Abstract A Bayesian statistical model is proposed that combines information from a multimodel ensemble of atmosphere–ocean general circulation models (AOGCMs) and observations to determine probability distributions of future temperature change on a regional scale. The posterior distributions derived from the statistical assumptions incorporate the criteria of bias and convergence in the relative weights implicitly assigned to the ensemble members. This approach can be considered an extension and elaboration of the reliability ensemble averaging method. For illustration, the authors consider the output of mean surface temperature from nine AOGCMs, run under the A2 emission scenario from the Synthesis Report on Emission Scenarios (SRES), for boreal winter and summer, aggregated over 22 land regions and into two 30-yr averages representative of current and future climate conditions. The shapes of the final probability density functions of temperature change vary widely, from unimodal curves for regions where...

Spatiotemporal Hierarchical Bayesian Modeling Tropical Ocean Surface Winds

Spatiotemporal processes are ubiquitous in the environmental and physical sciences. This is certainly true of atmospheric and oceanic processes, which typically exhibit many different scales of spatial and temporal variability. The complexity of these processes and the large number of observation/prediction locations preclude the use of traditional covariance-based spatiotemporal statistical methods. Alternatively, we focus on conditionally specified (i.e., hierarchical) spatiotemporal models. These methods offer several advantages over traditional approaches. Primarily, physical and dynamical constraints can be easily incorporated into the conditional formulation, so that the series of relatively simple yet physically realistic conditional models leads to a much more complicated spatiotemporal covariance structure than can be specified directly. Furthermore, by making use of the sparse structure inherent in the hierarchical approach, as well as multiresolution (wavelet) bases, the models can be computed with very large datasets. This modeling approach was necessitated by a scientifically meaningful problem in the geosciences. Satellite-derived wind estimates have high spatial resolution but limited global coverage. In contrast, wind fields provided by the major weather centers provide complete coverage but have low spatial resolution. The goal is to combine these data in a manner that incorporates the space-time dynamics inherent in the surface wind field. This is an essential task to enable meteorological research, because no complete high-resolution surface wind datasets exist over the world oceans. High-resolution datasets of this type are crucial for improving our understanding of global air–sea interactions affecting climate and tropical disturbances, and for driving large-scale ocean circulation models.

Bayesian Modeling of Uncertainty in Ensembles of Climate Models

Projections of future climate change caused by increasing greenhouse gases depend critically on numerical climate models coupling the ocean and atmosphere (global climate models [GCMs]). However, different models differ substantially in their projections, which raises the question of how the different models can best be combined into a probability distribution of future climate change. For this analysis, we have collected both current and future projected mean temperatures produced by nine climate models for 22 regions of the earth. We also have estimates of current mean temperatures from actual observations, together with standard errors, that can be used to calibrate the climate models. We propose a Bayesian analysis that allows us to combine the different climate models into a posterior distribution of future temperature increase, for each of the 22 regions, while allowing for the different climate models to have different variances. Two versions of the analysis are proposed: a univariate analysis in which each region is analyzed separately, and a multivariate analysis in which the 22 regions are combined into an overall statistical model. A cross-validation approach is proposed to confirm the reasonableness of our Bayesian predictive distributions. The results of this analysis allow for a quantification of the uncertainty of climate model projections as a Bayesian posterior distribution, substantially extending previous approaches to uncertainty in climate models.

An algorithm for the construction of spatial coverage designs with implementation in SPLUS

Abstract Space-filling “coverage” designs are spatial sampling plans which optimize a distance-based criterion. Because they do not depend on the covariance structure of the process to be sampled, coverage designs are computed more efficiently than designs that are optimal for mean-squared-error criteria. This paper presents an efficient algorithm for the construction of coverage designs and evaluates its performance in terms of computation time and effectiveness at finding “good” designs. Results suggest that near-optimal designs for reasonably large problems can be computed efficiently. The algorithm is implemented in the statistical programming language SPLUS and examples of the construction of coverage designs are given involving an existing network of ozone monitoring sites.

Multi-resolution time series analysis applied to solar irradiance and climate reconstructions

A better understanding of natural climate variability is crucial for global climate change studies and the evaluation of the sensitivity of the climate system to imposed perturbations. External forcing factors might contribute substantially to both high and low frequency variations in climate but a clear separation of their impact from internally generated fluctuations is difficult. We employ wavelet decomposition to identify common characteristics in forcing and climatic time series of the last four centuries. Here, we focus on solar irradiance variations by applying this statistical method to a selection of widely used proxy-based reconstructions. Major variability components are isolated through time-scale decomposition. Two classical solar modes (85 and 11 years) are not only identified within the limited time period covered by the solar datasets, but their relative influences on climate as represented by hemispheric surface temperature reconstructions are also estimated. While the low-frequency component shows close ties between solar variations and surface climate, a relationship between the 11-year sunspot cycle and temperature reconstructions is more difficult to attribute. However, the statistical multi-resolution analysis appears to be an ideal tool to uncover relationships and their changes at different temporal scales normally hidden by the strong background noise in the climate system.

Changes in Surface Air Temperature Caused by Desiccation of the Aral Sea

A statistical method for establishing the cause‐effect relationship between a land surface modification and some component of observed climatic change is presented. This method aids attribution in two ways. First, the climatic changes that are unique to the area influenced by some land surface modification are identified. This isolates changes caused by the spatially restricted forcing from changes caused by other factors. Second, most of the short-term climatic variability in the records from the affected area is removed based on information from the surrounding region. This makes it possible to identify smaller climatic changes. This method is used to identify the changes in surface air temperature that have resulted from desiccation of the Aral Sea (1960‐97). Desiccation has weakened the ‘‘lake effect’’ of the Aral Sea, so regional climatic changes are expected. Substantial temperature trends, unrelated to desiccation, are observed across a broad region of central Asia (;2000 km) between 1960 and 1997. These trends are similar in magnitude to the changes from desiccation. These trends are removed from the records from the Aral region because they would enhance or offset the local temperature changes caused by desiccation. There is also substantial year-to-year temperature variability that is spatially coherent across central Asia. The method used here removes ;80%‐90% of this short-term variability in the observed temperature records from the Aral region. This lowers the climate change detection limit from ;38‐88 Ct o;18‐28C, which improves the identification of the spatial extent of the desiccation-induced changes. The climate records from around the Aral Sea show dramatic temperature changes between 1960 and 1997, once regionally coherent trends and variability are removed. Mean, maximum, and minimum temperature near the Aral Sea have changed by up to 68C. Warming (cooling) is observed during spring and summer (autumn and winter), as expected to accompany the diminished lake effect caused by desiccation. The magnitude of temperature changes decreases with increasing distance from the 1960 shoreline, with changes extending up to ;200 km from the shoreline in the downwind direction. An increase in diurnal temperature range of 28‐38 Ci s observed in all months, demonstrating a weakening of the lake’s damping effect on the diurnal temperature cycle.

Parameter estimation for computationally intensive nonlinear regression with an application to climate modeling

Nonlinear regression is a useful statistical tool, relating observed data and a nonlinear function of unknown parameters. When the parameter-dependent nonlinear function is computationally intensive, a straightforward regression analysis by maximum likelihood is not feasible. The method presented in this paper proposes to construct a faster running surrogate for such a computationally intensive nonlinear function, and to use it in a related nonlinear statistical model that accounts for the uncertainty associated with this surrogate. A pivotal quantity in the Earths climate system is the climate sensitivity: the change in global temperature due to doubling of atmospheric CO 2 concentrations. This, along with other climate parameters, are estimated by applying the statistical method developed in this paper, where the computationally intensive nonlinear function is the MIT 2D climate model.

A Nonparametric Regression Approach to Syringe Grading for Quality Improvement

Abstract In the biomedical products industry, measures of the quality of individual clinical specimens or manufacturing production units are often available in the form of high-dimensional data such as continuous recordings obtained from an analytical instrument. These recordings are then examined by experts in the field who extract certain features and use these to classify individuals. To formalize and quantify this procedure, an approach for extracting features from recordings based on nonparametric regression is described. These features are then used to build a classification model that incorporates the knowledge of the expert. The procedure is illustrated with the problem of grading of syringes from associated friction profile data. Features of the syringe friction profiles used in the classification are extracted via smoothing splines, and grades of the syringes are assigned by an expert tribologist. A nonlinear classification model is constructed to predict syringe grades based on the extracted fe...

A methodology for evaluating the impact of data compression on climate simulation data

High-resolution climate simulations require tremendous computing resources and can generate massive datasets. At present, preserving the data from these simulations consumes vast storage resources at institutions such as the National Center for Atmospheric Research (NCAR). The historical data generation trends are economically unsustainable, and storage resources are already beginning to limit science objectives. To mitigate this problem, we investigate the use of data compression techniques on climate simulation data from the Community Earth System Model. Ultimately, to convince climate scientists to compress their simulation data, we must be able to demonstrate that the reconstructed data reveals the same mean climate as the original data, and this paper is a first step toward that goal. To that end, we develop an approach for verifying the climate data and use it to evaluate several compression algorithms. We find that the diversity of the climate data requires the individual treatment of variables, and, in doing so, the reconstructed data can fall within the natural variability of the system, while achieving compression rates of up to 5:1.

Testing theories to explore the drivers of cities' atmospheric emissions.

Abstract Despite a growing body of evidence demonstrating the importance of cities as sources of many local, regional, and global impacts on the atmosphere, ecosystems, and human populations, most theories on the relationship between society and the environment have focused on the global or national level. A variety of theories exist on human–environment interactions; for example, ecological modernization, urban transitions, and human ecology. However, with the exception of urban transitions, these theories have been mainly concerned with nation states and have ignored the subnational and local (city) levels. This article aims at filling this gap by employing ordinary least squares regression to examine these theories at the city level using the STIRPAT formula. It finds that with the exception of population (which shows an unstable relationship with the impacts indicators applied in the analysis) a remarkable level of variation exists in the importance of drivers across the three exercises. This led us to conclude that urban atmospheric pollutants result from diverse activities (e.g., transportation, industrial), are formed through different processes (vehicle combustion, biomass burning), have a residence time ranging from hours to years, and are the outcome of diverse sets of societal and environmental drivers.