Back to search

University of British Columbia

Investigating the utility of weighting schemes on a 51-member short-range ensemble forecast

Abstract

dc:description

Accurate weather predictions are crucial to the wellbeing of modern society. The Weather Forecast Research Team (WFRT) runs an ensemble of numerical weather prediction models daily (Short Range Ensemble Forecast; SREF 00Z). The ensemble can be combined into an “average” which is usually the best deterministic forecast. In this study, we experiment with ways of improving its skill for southwestern British Columbia. To do this, four different weighting techniques for ensemble averaging are employed using all 51 members of the SREF. These are constructed using 1 year (October 2021-September 2022) of forecasts and observation data pairs from 90 stations in the region, for temperature, wind speed and precipitation. The weights were then applied to a second year of data (October 2022-September 2023) and compared to the current SREF. To investigate the use of a shorter training set on the weights, as suggested by the literature, a sliding window approach was also used with the weights being calculated with the forecast and observation data pairs for the week and month prior to the forecast of interest. We also experimented with varying the number of ensemble members to explore the sensitivity of the SREF to the number of members. Spatially, there were some special cases where the weighted ensembles performed better or poorer than the current SREF. However, all four weighting schemes did not perform statistically significantly compared to the current equally-weighted SREF (pvalue > 0.05). As for the varying number of ensemble members, it was found that the skill of all of the ensembles varied throughout the year and was never consistently in the small or large range for all tested variables. However, we did note that the skill converged for all members greater than 30. This indicated a lower limit of the necessary ensemble members to match the current SREF skill. Future studies should include more data (i.e. longer datasets; climatology) to get a better picture of the state of the SREF and how its members are interacting and performing.

Degree

thesis:*
Name thesis:degree_name
Master of Science - MSc
Level thesis:degree_level
master's
Discipline thesis:degree_discipline
Atmospheric Science
Grantor dc:publisher
University of British Columbia
Year dc:date
2024

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • McKinney, Reagan
Contributors dc:contributor
  • Stull, Roland B., 1950-

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/87478

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

McKinney, Reagan. Investigating the utility of weighting schemes on a 51-member short-range ensemble forecast. master's thesis, University of British Columbia, 2024. http://hdl.handle.net/2429/87478