Frameworks for Analysis of Regional, concurrent, Conditional and Non-Stationary Extremes in Geosciences

Abstract

The objectives of this STIR project was to investigate the merit of the following model concepts: (1) A model for regional non-stationary analysis of extremes with constant and time-varying exceedance probability concepts. This will allow analysis of extremes in geosciences across different spatial scales under non-stationary assumption. The model is named Process-informed Nonstationary Extreme Value Analysis (ProNEVA) and can integrate time or a physically-based covariate to describe change in statistics of extremes. The source code of the toolbox along with a Graphical User Interface (GUI) is already freely available to the public. (2) An empirical Bayesian-based extreme value model for assessing concurrent and conditional extremes. This will allow deriving and assessing the full distribution functions of concurrent (joint) extremes in a changing environment.(3) A comprehensive and generalized framework for uncertainty assessment of extremes using the concept of Differential Evolution Markov Chain (DE-MC). This model will allow deriving quantitative uncertainty estimates for extremes in a non-stationary world.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 30, 2015
Accession Number
AD1094579

Entities

People

  • Amir AghaKouchak

Organizations

  • University of California, Irvine

Tags

DTIC Thesaurus Topics

  • Bayesian Inference
  • Bayesian Networks
  • Climate Change
  • Computational Science
  • Data Mining
  • Data Science
  • Databases
  • Earth Sciences
  • Geography
  • Information Science
  • Knowledge Management
  • Monte Carlo Method
  • Risk Analysis
  • Sea Level Rise
  • Statistical Algorithms
  • Storm Surges
  • Surveys

Readers

  • Computational Modeling and Simulation
  • Database Systems and Applications
  • Statistical inference.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference