Calibrated Probabilistic Forecasts of Sea Ice Concentration

This module is for making probabilistic forecasts of sea ice concentration (SIC) using the methods described in Dirkson et al, 2018. The user should refer to this paper for further details.

The module consists of two classes: beinf and taqm. The beinf class is used to fit SIC data to the zero- and one- inflated beta (BEINF) distribution (Ospina and Ferrari, 2010), and to freeze BEINF distribution objects that can be used to compute e.g. its pdf, its cdf, random variates. The taqm class is used to carry out the trend adjusted quantile mapping (TAQM) calibration method. The methods in both of these classes should be applied at the individual grid cell level. Examples are included in the tutorial below, and a template for using the taqm class is available at https://github.com/adirkson/SIC-probability (see the ‘README.md’ file for accessing this file).

This project was built in Python v2.7 and relies on the classes in the ‘beinf.py’ and ‘taqm.py’ files available at https://github.com/adirkson/SIC-probability (see the ‘README.md’ file for accessing these files).

Contents

• Tutorial
  • Zero- and One- Inflated Beta Distribution
  • Trend-adjusted Quantile Mapping
• Documentation
  • beinf (zero- and one- inflated beta distribution)
  • taqm (trend-adjusted quantile mapping)

Indices and tables

• Index
• Module Index
• Search Page

References

Dirkson, A., Merryfield, W. J., & Monahan, A. (2018). Calibrated Probabilistic Forecasts of Arctic Sea Ice Concentration. Journal of Climate, Accepted.

Ospina, R., & Ferrari, S. L. (2010). Inflated beta distributions. Statistical Papers, 51(1), 111.