Skip to content

GeoscienceAustralia/tcha

Repository files navigation

Tropical Cyclone Hazard Assessment

Evaluating the likelihood and magnitude of tropical cyclone winds based on a stochastic TC model (TCRM: https://github.com/geoscienceaustralia/tcrm).

Analysis of observations

Scripts to analyse the observational records of TCs and automatic weather station observations

TC frequency

frequency/tc_frequency.py - calculates mean frequency and trends for a range of TC datasets and time periods of those datasets. frequency/jtwc_frequency.py - Uses JTWC data to evaluate frequency frequency/frequency_distribution.py - fits a negative binomial distribution to annual frequency, for consideration as the source model for TCRM. Negative binomial initially selected over poisson distribution, as the distribution is very slightly overdispersed ([mu / sigma] < 1). frequency/tc_frequency_bayesian.py - use Bayesian MCMC methods to fit Poisson distribution to TC frequency, and generate posterior samples that can be used for sampling annual TC counts.

Track density

density/track_density.py - calculates TC frequency on a grid, counting the number of unique events intersecting each grid point. Currently uses the BoM best track dataset (IDCKMSTM0S.csv) as input, and a 0.5x0.5 degree grid over the simulation domain.

Compares 1981-2020 and 1951-2020 periods.

Uses jackknife (leave-one-out) bootstrap resampling to evaluate mean track density, by iteratively excluding seasons from the dataset for calculating track density.

To run:

``python density/track_density.py``

TC landfall rates

Lifetime maximum intensity

lmi/extractLMI.py lmi/extractLMI_IDCKMSTM0S.py

Potential intensity analysis

Using the theory of potential intensity to guide estimation of simulated TC intensity.

Basin-wide trends

Monthly trends

Potential intensity from climate models

Deep layer mean winds

dlm-climatology/climatology.py -

TC-related rainfall

precip/extract_precip.py - extracts ERA5 precipitation within a defined distance of the cyclone centre.

Contact:

Craig Arthur [email protected] Last updated: 2023-07-20