A number of factors make it difficult to obtain accurate results from climate time series. Some of these factors are:
- uneven time spacing
- autocorrelation (persistence, serial dependence)
- non-normal distributions
- uncertain timescales
CRA has developed, adapted, tested and proven numerous statistical analysis techniques which enhance the
accuracy and reliability of time series analysis data. The success of these techniques, as well as the
dual-disciplinary (both climate and statistics) approach to data analysis, has earned CRA a position of
respect in professional journals which are highly regarded by the climate change industry.
Analysis types
- Regression: linear, nonlinear, nonparametric, lagged
- Spectral analysis: Lomb–Scargle, WOSA, red-noise test, cross-spectra, multitaper
- Extreme value time series: risk analysis
- Correlation: Pearson's, Spearman's, nonlinear
- Stochastic processes: long memory, nonlinear dynamics
Estimation methods
- Maximum likelihood
- Nonparametric kernel
- Bootstrap simulations: moving block, parametric, stationary, ordinary, pairwise
- Timescale construction and simulation
Robustness evaluation
This answers how strongly the result depends on made assumptions.
- Sensitivity analyses
- Monte Carlo tests
|
|
