Studies in the Atmospheric Sciences

ISBN-13: 9780387987576 / Angielski / Miękka / 2000 / 199 str.

M. Berliner; D. Nychka; T. Hoar

Opis

The National Center for Atmospheric Research (NCAR) in Boulder, Colorado, is a major institution performing research in the environmental sciences. After an open competition, the National Science Foundation gave its financial sup- port to NCAR for the establishment of a much-needed statistics/probability program in the geophysical sciences. The result is the NCAR Geophysical Statistics Project (GSP). The primary mission of this program is: - To foster excellence in the use of modern statistical science in the geo- physical and environmental sciences. - To attract and support statistical scientists for collaborative research. To fulfill its mission, GSP engages in research in statistical science and its application to the atmospheric and allied sciences. It supports a variety of collaborative efforts between statistical scientists and disciplinary scientists. Senior statisticians, both members of and visitors to GSP, work in conjunc- tion with researchers at NCAR and elsewhere in the application of current statistical methods as well as in. the development of new statistical models and techniques. Senior GSP members and NCAR researchers also collaborate in the mentoring of junior members, typically post-doctoral level or statistics graduate students.

Spis treści

1 Introduction.- 1 Statistics in the Climate and Weather Sciences.- 2 A Guide to this Volume.- 2.1 Chapter Outline.- 2.2 What Is Missing?.- 3 Software, Datasets, and the Web Companion.- 2 A Statistical Perspective on Data Assimilation in Numerical Models.- 1 Introduction.- 2 Assimilation and Penalized Least Squares.- 3 Time-Dependent Assimilation Methods.- 3.1 The Kalman Filter.- 4 Ensemble Forecasting.- 4.1 Methodology.- 4.2 Statistical Perspective.- 5 Numerical Studies.- 5.1 Precipitation Model.- 6 Conclusions.- 3 Multivariate Spatial Models.- 1 Introduction.- 1.1 Motivating Example: Ozone and Meteorology.- 1.2 Optimal Spatial Prediction—Kriging.- 1.3 Universal Kriging.- 1.4 Multivariate Approaches.- 2 Cokriging.- 3 Kriging with External Drift.- 3.1 Prediction Variance Misspecification in the KED Model.- 4 A Hierarchical Model.- 5 Miscellaneous Topics.- 5.1 K-Variate Extensions.- 5.2 Latent Process Models.- 5.3 Modeling Orthogonal Contrasts.- 5.4 Multivariate Space—Time Models.- 6 Applications.- 6.1 Modeling Wind Fields.- 6.2 Modeling Temperature Fields.- 7 Conclusions.- 4 Hierarchical Space—Time Dynamic Models.- 1 Introduction.- 1.1 A Brief Review of Space—Time Modeling.- 1.2 Space—Time Dynamic Models.- 2 Hierarchical Space—Time Dynamic Modeling.- 2.1 A General Space—Time Dynamic Model.- 2.2 Reformulated Space—Time Dynamic Model.- 2.3 A Bayesian Model.- 3 Tropical Wind Process.- 3.1 Deterministic View.- 3.2 Stochastic View.- 3.3 Combined Stochastic/Dynamic View.- 3.4 Physically Informative Priors.- 4 Ocean Wind Implementation.- 4.1 Results.- 5 Discussion.- 5 Experimental Design for Spatial and Adaptive Observations.- 1 Introduction.- 1.1 Scientific Background.- 1.2 Overview of Statistical Design.- 2 Experimental Design: Spatial Fields.- 2.1 Optimal Design.- 2.2 Special Solutions.- 2.3 Greedy Algorithms.- 3 Experimental Design in Space—Time.- 4 Discussion.- 6 Seasonal Variation in Stratospheric Ozone Levels, a Functional Data Analysis Study.- 1 Introduction.- 2 Stratospheric Ozone Data.- 3 Principal Component Analysis.- 3.1 Principal Component Analysis of the Interpolated Profiles.- 4 Continuous Basis Functions to Represent Ozone.- 4.1 Basis Function Coefficients.- 5 Varying Coefficient Models.- 6 Discussion.- 7 Neural Networks: Cloud Parameterizations.- 1 Introduction.- 1.1 The Essence of Parameterization.- 1.2 Neural Networks and Fitting Nonlinear Models.- 2 Cloud Parameterizations.- 2.1 A Preliminary Model.- 2.2 Toward a GCM Cloud Parameterization.- 3 Simulation of Cloud Cover.- 4 Conclusions and Future Work.- 8 Exploratory Statistical Analysis of Tropical Oceanic Convection Using Discrete Wavelet Transforms.- 1 Introduction.- 1.1 Atmospheric Motivation.- 1.2 Statistical Motivation.- 1.3 Objectives.- 2 Description of the Dataset.- 2.1 Cloud System Regimes.- 3 Discrete Wavelets.- 3.1 Wavelets in One Dimension.- 3.2 Discrete Wavelet Transform in Two Dimensions.- 4 Statistical Study of Cloud Systems.- 4.1 Detecting Squall-Lines.- 4.2 The Orientation Problem.- 4.3 Examples.- 4.4 Identifying the Scattered Cloud Regime.- 5 Conclusions and Future Work.- 9 Predicting Clear-Air Turbulence.- 1 Introduction.- 2 Indices Derived from the RUC-60 Model.- 3 Data Structure.- 4 The Single Index Approach.- 5 Modeling Strategy.- 5.1 MARS.- 5.2 FDA.- 5.3 FDA + MARS Algorithm.- 6 Implementing FDA + MARS for CAT Forecast.- 6.1 Training the Procedure: In-Sample Performances.- 6.2 Testing the Procedure: Prediction Ability.- 7 MARS as Variable Subset Selection.- 8 Conclusions.- 10 Spatial Structure of the SeaWiFS Ocean Color Data for the North Atlantic Ocean.- 1 Introduction.- 2 SeaWiFS Ocean Color Data.- 3 Semivariograms and Other Tools in Spatial Statistics.- 3.1 Stationarity and Isotropy.- 3.2 Estimating the Empirical Semivariogram..- 3.3 Semivariogram Parameters: Nugget, Sill, and Range.- 3.4 Some Isotropic Semivariogram Models.- 4 Ocean Color Semivariograms.- 4.1 Spatial Analysis for the Ocean Chlorophyll.- 4.2 Semivariograms in the Original Scale.- 4.3 Effect of Sampling Resolution.- 5 Spatial Patterns for the North Atlantic Ocean.- 6 Conclusions and Final Remarks.- References.