Detailed information about the course

[ Back ]
Title

Climate Time Series Analysis

Dates

3 days in June (or September) 2020

Responsable de l'activité

Anneleen Foubert

Organizer(s)

Dr. Andres Rüggeberg, UNIFR

Speakers

Dr Manfred Mudelsee, Climate Risk Analysis

Description

The three-day course, given by Dr. Manfred Mudelsee, consists of lectures and computer tutorials. The course is in English and is designed for PhD students. The course is limited to 15 participants maximum allowing to give some individual feedback during the tutorials. The PhD students are very welcome to bring your own data for analysis. An own laptop is required as all course material is in electronic form (PDF, Windows executables, Fortran source codes), and printed copy of the Springer book'Climate time series analysis' will be offered.

Program

Day 1:

- Introduction into Climate archives, Timescales and Fundamental concepts of time series analysis

- Persistence models: Stochastic processes, Short versus long memory

- Bootstrap confidence intervals: Statistical estimation, Standard errors, bias, confidence intervals, Classical methods, Bootstrap resampling, Monte Carlo methods, Hypothesis tests

- Regression I: Linear regression, Nonlinear regression, Nonparametric regression or smoothing, Example: climate transitions

Day 2:

- Spectral Analysis: Spectrum and periodogram, Multitaper estimation, Lomb–Scargle periodogram, Example: solar cycles

- Extreme value time series: Data types and risk estimation, Stationary models (GEV, GP), Nonstationary models (Poisson), Example: flood risk analysis and hurricane risk

- Correlation: Pearson's correlation coefficient, Spearman's rank correlation coefficient, Example: runoff variations

Day 3: 

- Regression II: Errors-in-variables model, Prediction, Example: climate sensitivity and proxy calibration

- Future directions: Timescale modelling, Higher dimensions, Climate models, Optimal estimation Climate-related computer tutorial exercises (selection):

- Quantification of the Northern Hemisphere Glaciation in the Pliocene by means of change-point regression

- Comparing climate-model simulated runoff in different scenarios

- Nonparametric trends in regional Holocene climate with bootstrap confidence band

- Sunspot spectrum estimation

- Hurricane activity over the past 1000 years examined with nonstationary risk analysis

- Correlations between measured river runoff series from different hydrological stations

- Calibration of a paleoclimatic proxy variable

Location

University of Fribourg

Credits

1.5

Information
Places

15

Deadline for registration 01.06.2020
short-url short URL

short-url URL onepage