Alexis Hannart

Causal Counterfactual Theory for the Attribution of Weather and Climate-Related Events

AbstractThe emergence of clear semantics for causal claims and of a sound logic for causal reasoning is relatively recent, with the consolidation over the past decades of a coherent theoretical corpus of definitions, concepts, and methods of general applicability that is anchored into counterfactuals. The latter corpus has proved to be of high practical interest in numerous applied fields (e.g., epidemiology, economics, and social science). In spite of their rather consensual nature and proven efficacy, these definitions and methods are to a large extent not used in detection and attribution (D&A). This article gives a brief overview of the main concepts underpinning the causal theory and proposes some methodological extensions for the causal attribution of weather and climate-related events that are rooted into the latter. Implications for the formulation of causal claims and their uncertainty are finally discussed.

An Improved Bayesian Information Criterion for Multiple Change-Point Models

In multiple change-point analysis, inferring the number of change points is often achieved by minimizing a selection criterion that trades off data fidelity with complexity. We address the open problem of defining a selection criterion adapted to the context of multiple change-point analysis. Our approach is inspired by the Schwarz seminal formulation of the Bayesian information criterion (BIC): similarly, we introduce priors—here describing the occurrence of change points—and we use the Laplace approximation to derive a closed-form expression of the criterion. Differently from this previous work, we take advantage of the a priori information introduced, instead of asymptotically eliminating the dependence on priors. Results obtained on simulated series show a substantial gain in performance versus recent alternative criteria used in multiple change-point analysis. Results also show that the a priori information introduced in our criterion on the regularity of interevent times is the main driver of this substantial performance gain. Methods are motivated by and demonstrated on a meteorological application involving the homogenization of a temperature series.

DADA: data assimilation for the detection and attribution of weather and climate-related events

We describe a new approach that allows for systematic causal attribution of weather and climate-related events, in near-real time. The method is designed so as to facilitate its implementation at meteorological centers by relying on data and methods that are routinely available when numerically forecasting the weather. We thus show that causal attribution can be obtained as a by-product of data assimilation procedures run on a daily basis to update numerical weather prediction (NWP) models with new atmospheric observations; hence, the proposed methodology can take advantage of the powerful computational and observational capacity of weather forecasting centers. We explain the theoretical rationale of this approach and sketch the most prominent features of a “data assimilation–based detection and attribution” (DADA) procedure. The proposal is illustrated in the context of the classical three-variable Lorenz model with additional forcing. The paper concludes by raising several theoretical and practical questions that need to be addressed to make the proposal operational within NWP centers.

Optimal fingerprinting under multiple sources of uncertainty

Detection and attribution studies routinely use linear regression methods referred to as optimal fingerprinting. Within the latter methodological paradigm, it is usually recognized that multiple sources of uncertainty affect both the observations and the simulated climate responses used as regressors. These include for instance internal variability, climate model error, or observational error. When all errors share the same covariance, the statistical inference is usually performed with the so-called total least squares procedure, but to date no inference procedure is readily available in the climate literature to treat the general case where this assumption does not hold. Here we address this deficiency. After a brief outlook on the error-in-variable models literature, we describe an inference procedure based on likelihood maximization, inspired by a recent article dealing with a similar situation in geodesy. We evaluate the performance of our approach via an idealized test bed. We find the procedure to outperform existing procedures when the latter wrongly neglect some sources of uncertainty.

Disconcerting learning on climate sensitivity and the uncertain future of uncertainty

How will our estimates of climate uncertainty evolve in the coming years, as new learning is acquired and climate research makes further progress? As a tentative contribution to this question, we argue here that the future path of climate uncertainty may itself be quite uncertain, and that our uncertainty is actually prone to increase even though we learn more about the climate system. We term disconcerting learning this somewhat counter-intuitive process in which improved knowledge generates higher uncertainty. After recalling some definitions, this concept is connected with the related concept of negative learning that was introduced earlier by Oppenheimer et al. (Clim Change 89:155–172, 2008). We illustrate disconcerting learning on several real-life examples and characterize mathematically certain general conditions for its occurrence. We show next that these conditions are met in the current state of our knowledge on climate sensitivity, and illustrate this situation based on an energy balance model of climate. We finally discuss the implications of these results on the development of adaptation and mitigation policy.

Combining Analog Method and Ensemble Data Assimilation: Application to the Lorenz-63 Chaotic System

Nowadays, ocean and atmosphere sciences face a deluge of data from space, in situ monitoring as well as numerical simulations. The availability of these different data sources offers new opportunities, still largely underexploited, to improve the understanding, modeling, and reconstruction of geophysical dynamics. The classical way to reconstruct the space-time variations of a geophysical system from observations relies on data assimilation methods using multiple runs of the known dynamical model. This classical framework may have severe limitations including its computational cost, the lack of adequacy of the model with observed data, and modeling uncertainties. In this paper, we explore an alternative approach and develop a fully data-driven framework, which combines machine learning and statistical sampling to simulate the dynamics of complex system. As a proof concept, we address the assimilation of the chaotic Lorenz-63 model. We demonstrate that a nonparametric sampler from a catalog of historical datasets, namely, a nearest neighbor or analog sampler, combined with a classical stochastic data assimilation scheme, the ensemble Kalman filter and smoother, reaches state-of-the-art performances, without online evaluations of the physical model.

Estimating model evidence using data assimilation

Author(s): Carrassi, A; Bocquet, M; Hannart, A; Ghil, M | Abstract:

Integrated Optimal Fingerprinting: Method Description and Illustration

AbstractThe present paper introduces and illustrates methodological developments intended for so-called optimal fingerprinting methods, which are of frequent use in detection and attribution studies. These methods used to involve three independent steps: preliminary reduction of the dimension of the data, estimation of the covariance associated to internal climate variability, and, finally, linear regression inference with associated uncertainty assessment. It is argued that such a compartmentalized treatment presents several issues; an integrated method is thus introduced to address them. The suggested approach is based on a single-piece statistical model that represents both linear regression and control runs. The unknown covariance is treated as a nuisance parameter that is eliminated by integration. This allows for the introduction of regularization assumptions. Point estimates and confidence intervals follow from the integrated likelihood. Further, it is shown that preliminary dimension reduction is ...

Causal Influence of Anthropogenic Forcings on the Argentinian Heat Wave of December 2013

Fil: Hannart, Alexis. Consejo Nacional de Investigaciones Cientificas y Tecnicas. Oficina de Coordinacion Administrativa Ciudad Universitaria. Centro de Investigaciones del Mar y la Atmosfera. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Centro de Investigaciones del Mar y la Atmosfera; Argentina