GEV trend test via likelihood ratio — gev_trend

Fits stationary GEV(mu, sigma, xi) and nonstationary GEV(mu0 + mu1*year, sigma, xi) via extRemes::fevd, then compares with a likelihood ratio test (chi-squared, df=1).

Usage

gev_trend_test(annual_maxima, years)

Arguments

annual_maxima: Numeric vector of annual maxima.
years: Numeric vector of corresponding years (same length).

Value

A list with components:

mu0: Intercept of nonstationary location.
mu1: Slope of nonstationary location (units per year).
lrt_stat: Likelihood ratio test statistic.
lrt_pvalue: P-value from chi-squared(df=1).
aic_stat: AIC of stationary model.
aic_nonstat: AIC of nonstationary model.
nllh_stat: Negative log-likelihood of stationary model.
nllh_nonstat: Negative log-likelihood of nonstationary model.