S-SOM v1.0: a structural self-organizing map algorithm for weather typing

This study proposes a novel structural self-organizing map (S-SOM) algorithm for synoptic weather typing. A novel feature of the S-SOM compared with traditional SOMs is its ability to deal with input data with spatial or temporal structures. In detail, the search scheme for the best matching unit (BMU) in a S-SOM is built based on a structural similarity (S-SIM) index rather than by using the traditional Euclidean distance (ED). S-SIM enables the BMU search to consider the correlation in space between weather states, such as the locations of highs or lows, that is impossible when using ED. The S-SOM performance is evaluated by multiple demo simulations of clustering weather patterns over Japan using the ERA-Interim sea-level pressure data. The results show the S-SOM's superiority compared with a standard SOM with ED (or ED-SOM) in two respects: clustering quality based on silhouette analysis and topological preservation based on topological error. Better performance of S-SOM versus ED is consistent with results from different tests and node-size configurations. S-SOM performs better than a SOM using the Pearson correlation coefficient (or COR-SOM), though the difference is not as clear as it is compared to ED-SOM.

A structural self-organizing map (S-SOM) algorithm for weather typing

In this study, we propose a novel structural self-organizing map (S-SOM) algorithm for synoptic weather typing. A novel feature of the S-SOM compared with traditional SOMs is its ability to deal with input data that have spatial or temporal structures. In detail, the search scheme for the best matching unit (BMU) in a S-SOM is built based upon a structural similarity (S-SIM) index rather than by using the traditional Euclidean distance (ED). S-SIM enables the BMU search to consider the correlation in space between weather states, such as the location of highs of lows, that is impossible when using ED. The S-SOM performance is evaluated by multiple demo simulations of clustering weather patterns over Japan using the ERA-Interim sea-level pressure data. The results show the superiority of the S-SOM compared with a standard SOM with ED (or ED-SOM) in two respects: clustering quality based on silhouette analysis and topological preservation based on topological error analysis. The superior performance of the S-SOM compared with an ED-SOM is probably independent of both the input data and SOM configuration.

Evaluation of climate change impacts on extreme rainfall events characteristics using a synoptic weather typing-based daily precipitation downscaling model

Climate change has impacted all phenomena in the hydrologic cycle, especially extreme events. General circulation models (GCMs) are used to investigate climate change impacts but because of their low resolution, downscaling methods are developed to provide data with high enough resolution for regional studies from GCM outputs. The performance of rainfall downscaling methods is commonly acceptable in preserving average characteristics, but they do not preserve the extreme event characteristics especially rainfall amount and distribution. In this study, a novel downscaling method called synoptic statistical downscaling model is proposed for daily precipitation downscaling with an emphasis on extreme event characteristics preservation. The proposed model is applied to a region located in central Iran. The results show that the developed model can downscale all percentiles of precipitation events with an acceptable performance and there is no assumption about the similarity of future rainfall data with the historical observations. The outputs of CCSM4 GCM for two representative concentration pathways (RCPs) of RCP4.5 and RCP8.5 are used to investigate the climate change impacts in the study region. The results show 40% and 30% increase in the number of extreme rainfall events under RCP4.5 and RCP8.5, respectively.

A Synoptic Weather-Typing Approach to Project Future Daily Rainfall and Extremes at Local Scale in Ontario, Canada

This paper attempts to project possible changes in the frequency of daily rainfall events late in this century for four selected river basins (i.e., Grand, Humber, Rideau, and Upper Thames) in Ontario, Canada. To achieve this goal, automated synoptic weather typing as well as cumulative logit and nonlinear regression methods was employed to develop within-weather-type daily rainfall simulation models. In addition, regression-based downscaling was applied to downscale four general circulation model (GCM) simulations to three meteorological stations (i.e., London, Ottawa, and Toronto) within the river basins for all meteorological variables (except rainfall) used in the study. Using downscaled GCM hourly climate data, discriminant function analysis was employed to allocate each future day for two windows of time (2046–65, 2081–2100) into one of the weather types. Future daily rainfall and its extremes were projected by applying within-weather-type rainfall simulation models together with downscaled future GCM climate data. A verification process of model results has been built into the whole exercise (i.e., statistical downscaling, synoptic weather typing, and daily rainfall simulation modeling) to ascertain whether the methods are stable for projection of changes in frequency of future daily rainfall events. Two independent approaches were used to project changes in frequency of daily rainfall events: method I—comparing future and historical frequencies of rainfall-related weather types, and method II—applying daily rainfall simulation models with downscaled future climate information. The increases of future daily rainfall event frequencies and seasonal rainfall totals (April–November) projected by method II are usually greater than those derived by method I. The increase in frequency of future daily heavy rainfall events greater than or equal to 25 mm, derived from both methods, is likely to be greater than that of future daily rainfall events greater than or equal to 0.2 mm: 35%–50% versus 10%–25% over the period 2081–2100 derived from method II. In addition, the return values of annual maximum 3-day accumulated rainfall totals are projected to increase by 20%–50%, 30%–55%, and 25%–60% for the periods 2001–50, 2026–75, and 2051–2100, respectively. Inter-GCM and interscenario uncertainties of future rainfall projections were quantitatively assessed. The intermodel uncertainties are similar to the interscenario uncertainties, for both method I and method II. However, the uncertainties are generally much smaller than the projection of percentage increases in the frequency of future seasonal rain days and future seasonal rainfall totals. The overall mean projected percentage increases are about 2.6 times greater than overall mean intermodel and interscenario uncertainties from method I; the corresponding projected increases from method II are 2.2–3.7 times greater than overall mean uncertainties.

A Synoptic Weather Typing Approach to Simulate Daily Rainfall and Extremes in Ontario, Canada: Potential for Climate Change Projections

An automated synoptic weather typing and stepwise cumulative logit/nonlinear regression analyses were employed to simulate the occurrence and quantity of daily rainfall events. The synoptic weather typing was developed using principal component analysis, an average linkage clustering procedure, and discriminant function analysis to identify the weather types most likely to be associated with daily rainfall events for the four selected river basins in Ontario. Within-weather-type daily rainfall simulation models comprise a two-step process: (i) cumulative logit regression to predict the occurrence of daily rainfall events, and (ii) using probability of the logit regression, a nonlinear regression procedure to simulate daily rainfall quantities. The rainfall simulation models were validated using an independent dataset, and the results showed that the models were successful at replicating the occurrence and quantity of daily rainfall events. For example, the relative operating characteristics score is greater than 0.97 for rainfall events with daily rainfall ≥10 or ≥25 mm, for both model development and validation. For evaluation of daily rainfall quantity simulation models, four correctness classifications of excellent, good, fair, and poor were defined, based on the difference between daily rainfall observations and model simulations. Across four selected river basins, the percentage of excellent and good simulations for model development ranged from 62% to 84% (of 20 individuals, 16 cases ≥ 70%, 7 cases ≥ 80%); the corresponding percentage for model validation ranged from 50% to 76% (of 20 individuals, 15 cases ≥ 60%, 6 cases ≥ 70%).

Possible impacts of climate change on freezing rain in south-central Canada using downscaled future climate scenarios

Freezing rain is a major atmospheric hazard in mid-latitude nations of the globe. Among all Canadian hydrometeorological hazards, freezing rain is associated with the highest damage costs per event. Using synoptic weather typing to identify the occurrence of freezing rain events, this study estimates changes in future freezing rain events under future climate scenarios for south-central Canada. Synoptic weather typing consists of principal components analysis, an average linkage clustering procedure (i.e., a hierarchical agglomerative cluster method), and discriminant function analysis (a nonhierarchical method). Meteorological data used in the analysis included hourly surface observations from 15 selected weather stations and six atmospheric levels of six-hourly National Centers for Environmental Prediction (NCEP) upper-air reanalysis weather variables for the winter months (November–April) of 1958/59–2000/01. A statistical downscaling method was used to downscale four general circulation model (GCM) scenarios to the selected weather stations. Using downscaled scenarios, discriminant function analysis was used to project the occurrence of future weather types. The within-type frequency of future freezing rain events is assumed to be directly proportional to the change in frequency of future freezing rain-related weather types The results showed that with warming temperatures in a future climate, percentage increases in the occurrence of freezing rain events in the north of the study area are likely to be greater than those in the south. By the 2050s, freezing rain events for the three colder months (December–February) could increase by about 85% (95% confidence interval – CI: ±13%), 60% (95% CI: &plusmn9%), and 40% (95% CI: ±6%) in northern Ontario, eastern Ontario (including Montreal, Quebec), and southern Ontario, respectively. The increase by the 2080s could be even greater: about 135% (95% CI: ±20%), 95% (95% CI: ±13%), and 45% (95% CI: ±9%). For the three warmer months (November, March, April), the percentage increases in future freezing rain events are projected to be much smaller with some areas showing either a decrease or little change in frequency of freezing rain. On average, northern Ontario could experience about 10% (95% CI: ±2%) and 20% (95% CI: ±4%) more freezing rain events by the 2050s and 2080s, respectively. However, future freezing rain events in southern Ontario could decrease about 10% (95% CI: ±3%) and 15% (95% CI: ±5%) by the 2050s and 2080s, respectively. In eastern Ontario (including Montreal, Quebec), the frequency of future freezing rain events is projected to remain the same as it is currently.