A fuzzy rule based classification method has been adapted to identifying the atmo- spheric circulation patterns that drive regional wave climates. The east coast of South Africa was used as a case study. The method is based on normalized anomalies in daily 700 hPa geo-potential heights. The CP classes are derived from an optimization procedure which is guided by a variable of interest, in this case wave heights. The classification shows a strong anomaly pattern east-southeast of South Africa which explains 30 − 60% of extreme wave events. This CP type explains extreme events in all seasons. However it occurs infrequently (∼ 8% of the time) and is associated with large wave heights ranging from 5.0−8.5 m. Frequently occurring CP classes have similar structure to mid-latitude cyclones or translational low pressure systems
(followed by a zone of high pressure) that occur south of South Africa (Taljaard, 1967).
The methodology discussed here appears to be new in the context of wave cli- mate analysis and has potential for application to risk assessment studies in coastal management and engineering.
On linking atmospheric circulation
patterns to extreme wave events for
coastal vulnerability assessments.
Atmospheric circulation patterns (CPs) are fundamental drivers of regional wave cli- mates. A fuzzy rule based classification algorithm has recently been developed to identify these atmospheric features. The algorithm is guided by wave heights and op- timises the location, shape and strength of a set of CP classes in order to find features that drive extreme waves. This paper focuses on a method for evaluating the perfor- mance of CP classification algorithms and reducing the subjectivity in the selection of classification parameters. We suggest a method based on entropy to quantify the classification quality and provide a means to objectively define an optimal number of CP classes. We also explore the sensitivity to the temporal resolution of the data. For our case study site the entropy measure indicates that a good quality classification requires 15–20 CP classes. However regardless of the number of classes used there is a persistent, common class that explains a large proportion of extreme wave events.
The methods described here contribute to developing a new framework for improved statistical wave modelling in coastal vulnerability risk assessments.
4.1 Introduction
Atmospheric features drive regional wave climates. Their occurrence and persistence control wave development including extreme wave events that can cause severe coastal erosion. Therefore atmospheric circulation patterns (CPs) are fundamental drivers of coastal vulnerability.
The physical links between ocean waves and atmospheric circulation is complex since it involves processes over a range of spatial and temporal scales. Current global wave models rely heavily on reanalysis with data assimilation to produce reasonable hind-cast wave data, but remain poor at reproducing extreme events nearshore (Caires et al., 2004; Chawla et al., 2013; Stopa & Cheung, 2014; Swail & Cox, 1999; Tolman et al., 2002). Furthermore they are limited in their ability to predict future climate
scenarios. However event timing is generally well captured by models, so they can be used to identify the drivers of regional wave climates. Recently, the classification of synoptic scale atmospheric circulation with links to regional wave climates has provided insights that suggest that CPs can be used to statistically model regional wave climates (Corbella et al., 2015; Espejoet al., 2014; Pringleet al., 2014). A more detailed discussion of this approach is presented in § 4.4.2.
There are several reasons why classifying atmospheric drivers and their links to waves should have an important role in coastal vulnerability and climate change as- sessments. It provides a natural way to identify independent storm events for risk assessment. Corbella & Stretch (2012b) identified independent events based on auto- correlation, while Li et al. (2013) separated independent events by requiring a min- imum inter-event time, and Callaghan et al. (2008) manually assessed independent events based on the meteorological features associated with them. The transition of CPs between classes seems to be the most physically meaningful method for de- lineating independent events. Atmospheric CPs also contain important information relevant to the distributions of wave height, direction and period. For example when a particular CP type occurs the associated wave height, direction and period can be (statistically) predicted. CP classification can also be used to extend current data sets and to infill missing data (Hewitson & Crane, 2002). Finally the prediction and evaluation of climate change impacts on coastal vulnerability would be more robust if they are linked to changes in the atmospheric circulation patterns that are the basic drivers of extreme wave events.
Objective atmospheric classification algorithms have been used extensively in the past to cluster similar circulation patterns into classes representing different atmo- spheric states Bárdossy (2010); Bárdossyet al. (1995, 2002); Hess & Brezowsky (1952);
Huthet al. (2008); Lamb (1972). Automated classification techniques that have been used to great effect include clustering analysis, principle component analysis (PCA) and self organizing maps (SOMS) (Hewitson & Crane, 2002; Huthet al., 2008). While these methods have been shown to provide insight into the CPs that are typical of particular regions, links to surface weather phenomena (such as precipitation, tem- perature or the wave climate) are only found after the classes are derived and are not part of the class derivation itself. This study aims to elucidate the links between atmospheric CPs and coastal vulnerability events. In such cases coupling a variable of interest associated with those events to CPs after the classes have been derived may not be the most effective approach. Alternatively, if the variable of interest is used
to guide the CP classification to an optimal solution we can gain more direct insights into the types of CPs that drive those events. In fact Bárdossy et al. (2015) have shown how two guiding variables (wave height and precipitation) used by the same classification algorithm produce significantly different results.
Pringle et al. (2014) were the first to apply statistical methods to finding a set of atmospheric states (or circulation patterns) with strong links to wave behaviour. This new approach to wave climate analysis may have important implications for coastal vulnerability and risk assessment. It uses a fuzzy rule based classification algorithm to identify CPs that drive wave development and builds on the approach previously introduced by Bárdossy et al. (1995) for linking CPs to rainfall. The classification uses objective reasoning and is guided by the wave climate data to give an optimal solution. However Huth et al. (2008) argue that so-called objective classifications are still subjective to some extent, such as in the selection of certain key parameters. For example in their classification study Pringleet al. (2014) used eight CP classes whose patterns were derived from daily data. The decision to use eight classes was subjective and the effect on the model performance was not evaluated. Nor were the effects of the selected temporal resolution of the data. The sensitivity of the algorithm to such parameters can be important. Therefore this paper addresses these issues and suggests an objective means to select the best combination of model parameters. Applications of CP based wave modelling techniques, which are currently under development for coastal vulnerability assessments, are then discussed.