Australia: The Land Where Time Began
Southwestern Australia – Rainfall Changes and Their Relationship to the Southern Annular Mode and ENSO
Rainfall over coastal southwestern Australia has decreased by 10%-20% since the 1970s (IOCI, 2002), while summer rainfall has increased by 40%-50% in the inland areas that are semiarid. In this paper Raut et al. used a K-means algorithm to cluster rainfall patterns directly as opposed to an approach that was more conventional of clustering synoptic conditions, which is usually the mean sea level pressure, and inferring the rainfall that is associated with it. It was shown that fewer westerly fronts in June and July were the main reason for the reduced coastal rainfall during winter. The frequency of strong front reduction in June is responsible for half of the decrease in rainfall in June-August (JJA), whereas the weaker fronts frequency reduction in June and July accounts for ⅓ of the total decrease. The rainfall increase in the inland in December to February (DJF) results from an increase in frequency of easterly troughs in December and February. These rainfall patterns are linked to the southern annular mode (SAM) and the Southern Oscillation index (SOI). The coastal rainfall and the inland rainfall are both related to the predominantly positive phase of SAM, especially in a neutral ENSO phase.
As well as the coastal southwestern Australia rainfall reduction after the 1970s droughts on seasonal scales have increased in intensity and longevity in the region (Gallant et al., 2013). This rainfall reduction also reduced the dam flows by more than 50% in the region (Bates et al., 2008). Contrasting with this there was an increase of rainfall totals and frequency of extreme rainfall events during the summer over inland SWA (Suppiah & Hennessy, 1998; Fierro & Leslie, 2013).
There are a number of studies that have sought an explanation for the variability of the coastal rainfall and declining winter rainfall since the 1970s (see IOCI, 2002; Nichols, 2006, and references therein). The role of large-scale climate modes have been explored by the earliest studied, including, though not only the ENSO, the southern annular mode (SAM), and the temperatures in the Indian Ocean, in annual and seasonal variability of rainfall (McBride & Nichols, 1983; IOCI, 2002; Pezza et al., 2008). It has been found (McBride & Nicholls, 1983), (Allan & Haylock, 1993) that there is a strong relationship between the declining rainfall along the southwest coast of Australia and the long-term anomalies in mean sea level pressure, though there is a weak connection between rainfall and ENSO over the coastal regions of southwestern Australia compared to the remainder of the continent. They speculated that the ENSO may have influenced the fluctuations in the circulation driving these mean sea level pressure anomalies. A weak, though signification correlation of the June to August (JJA) rainfall with the Southern Oscillation index and the dipole mode index (DMI) has been reported (Risbey et al., 2009), which is consistent with the speculation of Raut et al. Nevertheless there has not been a significant trend in the Southern Oscillation index during the period over which the rainfall over southwestern Australia has decreased and, therefore, it cannot be linked to the long-term trends in rainfall over the region (Chowdhury & Beecham, 2010; Nicholls, 2010).
The reduction of winter rainfall over the coastal regions, on the other hand, has been found to be associated with the positive phase of the daily SAM index (Hendon et al., 2007), though the correlation to the coastal June to August rainfall with the SAM index was found to be insignificant; only during SON over the inland region (Risbey et al., 2009). It was found, similarly, (Meneghini et al., 2007) that there was no long term association between the seasonal SAM index and seasonal rainfall in southwestern Australia. Year to year variations in the rainfall over southern Australia, however, are correlated with the SAM index (Nicholls, 2010). It was reported (Feng et al., 2010) that the correlation of rainfall with Southern Annular Mode index is significant if the year 1964 is excluded from the time series. It appears, therefore that the effect of SAM on the changing rainfall over southwestern Australia is still not certain. Moreover, the various correlations between climate indices and rainfall that have been found in all the above studies are either insignificant or barely exceed 0.35 and as a consequence can explain only 12% of the variance in the rainfall over southwestern Australia.
An explanation for the declining rainfall, through changes in the frequency and the strength of the synoptic systems in the region has been sought by some studies. It has been found that there is a strong inverse relationship between rainfall in coastal southwestern Australia and the MSLP over the region (Allan & Haylock, 1993; Ansell et al., 2000; Li et al., 2005), indicating the role of large-scale circulation patterns in the control of the rainfall. The declining winter rainfall, in particular, has been associated with the reduced frequency of the front-like pressure systems over the region (Hope et al., 2006; Alexander et al., 2010) and the increases in both station MSLP over southwestern Australia and the surface temperature of the Indian Ocean (Smith et al., 2000). It has been found (e.g., Risbey et al., 2013) that the reduced June to August rainfall between 1985 and 2009 is associated mostly with the frontal systems. In this region, the mean annual frequency of fronts, as was deduced from several reanalyses, has increased over the period 1989-2009 (Berry et al., 2011a), in spite of the number of the low pressure systems. According to Raut et al. it is likely these conflicting conclusions can be attributed to the different methods used to identify and classify synoptic systems that are rain-bearing such as fronts, troughs, and cut-off lows (Hope et al., 2014).
As well as fronts, cut-off lows frequently affect southwestern Australian during winter (Qi et al., 1999). The contribution of cut-offs to the rainfall in June and July is less than that at any other time of year (Pook et al., 2013), though about ⅓ of the rainfall to southwestern Australia during April and October is provided by the cut-off lows. A negative trend in the intensity of the cutoff lows was reported (Pook et al., 2012), though there is no significant trend in their frequency. It was found similarly (Risbey et al., 2013) that the rainfall over inland southwestern Australia that results from cutoff lows has decreased 1985-2009.
It is believed that local land cover-type changes also contribute to reduced rainfall along the coast and increased rainfall over the inland area (Pitman et al., 2004), though it is difficult to quantify their contribution. The way in which land cover change could change the frequency of fronts and troughs, as was summarised in cluster analysis (Hope e al., 2006), or change storm tracks in the mid-latitudes and position and strength of the jet stream, as has been reported (Frederiksen & Frederiksen, 2007; Frederiksen et al., 2011), is also difficult to explain.
Cluster analysis is commonly used to classify synoptic and surface conditions (Stone, 1989; Hope et al., 2006) or profiles from radiosondes (Pope et al., 2009) into distinct weather regimes. Inferences are drawn from the properties of these weather regimes and their trends, about the physical processes that are responsible for the associated rainfall and it trends. When synoptic processes are the main focus of the study and the relationship between the weather regimes that are defined and rainfall is strong, this approach works well. Also, the patterns change smoothly from cluster to cluster and the clusters are of comparable size, for a variable that is normally distributed. It is shown by experience that it takes many more clusters to separate the synoptic patterns that are associated with heavy rainfall. Changes in a few heavy rain events may lead to changes in the accumulation for seasonal rainfall.
The work described in this paper takes a simple and more direct approach and clusters the daily rainfall directly, which is in contrast to the traditional approach. K-means clustering always produces a large cluster for the light rain and clusters for the moderate and heavy rainfall events that are comparatively smaller, which automatically separates days of light rain from the extreme events. Therefore clustering on rainfall compared with other variables is an advantage.
The composite MSLP and horizontal wind are inferred for each cluster. The trend in each rainfall cluster is calculated and the monthly changes in rainfall are attributed to the changes in the frequency and intensity of the clusters. Rainfall clusters also help in the study of the combined effects of SOI and SAM.
Discussion & conclusions
It is shown by the results that it is a useful technique to cluster on rainfall patterns when studying changes in rainfall as a function of synoptic conditions, and also on linking these changes to large-scale climate modes, such as ENSO and SAM. It is shown by the study that the major decline in the winter rainfall over coastal southwestern Australia results from the overall reduction in the frequency of westerly fronts, strong fronts in particular. Also, the rainfall increase over inland southwestern Australia results from an increased frequency of easterly troughs in December and February. It was shown that the reduction in winter rainfall and the increase in summer rainfall depend on the phases of both SAM and ENSO; the positive phase of SAM is associated with reduced winter rainfall and increased summer rainfall in all 3 phases of ENSO. Also, in the post 1970s the neutral phase of ENSO in combination with the positive phase of SAM occurred more often than in the earlier period and such a combination is associated with a reduction of rainfall from westerly fronts and increased rainfall from easterly toughs.
The K-means clustering method bins rainfall according to magnitude and geographic distribution, which results in separate classes for the rainfall on the coast and the inland and separate classes for heavy events and light or moderate events. An important aspect of the study that is presented in this paper is clustering of rainfall patterns that allows for an estimation of the change, or trend, which is direct and consistent, in the rainfall that is associated with any cluster. It is shown by experience that clustering on MSLP (Hope et al., 2006; Alexander et al., 2010) or any other smooth variable that does not differentiate clearly heavy rainfall conditions from the light rain conditions that are more frequent. Moreover, dry days dominate all such clusters. Changes in heavy rainfall events that are comparatively less frequent can cause significant changes in the mean rainfall as is evident in clusters 3 and 5 in this study.
The results are consistent, qualitatively, with the earlier studies (Hope et al., 2006; Alexander et al., 2010) over coastal southwestern Australia which show that the decreasing frequency of fronts is the main reason for the declining rainfall. Also, the results that are reported in this paper support the conclusion that a large fraction of the decline in rainfall, more than 50%, in June to August is the result of fewer strong fronts occurring. It appears that after the 1970s the conditions have reduced the number of fronts in general and prevented the frontal systems from strengthening. More has been contributed to the decline in rainfall by the decrease in the number of strong fronts than the decrease in weak fronts. It is shown by this study that the reduction of rainfall in the 1970s was due mainly to fewer fronts, though an increase in the number of dry days, which were associated with the persistent high, is responsible for the recent decline in the 1990s. The reduced frequency of fronts did not lead to an increase in the number of dry days in the 1970s and the change in days of light rain was also small.
The frequency of raining easterly troughs bring increased rainfall over the inland areas, though it is difficult to know whether the frequency of all troughs, both wet and dry, changed during this period, without any objective method of identifying these troughs. A front detection method that has been used (Berry et al., 2011b) has recently shown a high frequency of fronts in December to February as compared to June to August and a trend that increases in annual frequency of fronts over southwestern Australia (Berry et al., 2011a). It was shown by the use of the same method that a large fraction of the December to January rainfall is connected to warm fronts, while June to August rainfall is connected mainly to cold fronts. Raut et al. suggest it is likely that many of the fronts over Western Australia that were detected (Berry et al., 2011b) are associated with the easterly troughs and the annual increase in the number of troughs is a reflection of the increase in the number of easterly troughs in summer. This reduction is not included in the study (Berry et al., 2011a) as the data used in the study that began in 1989, though the frequency of frontal clusters (i.e., clusters 2 and 3) fell in the 1970s. According to Raut it is required that a study that is more focused using objective analysis to determine the seasonal trends in fronts, troughs and cutoff lows in this area.
It was concluded (Hendon et al., 2007) that the effect of the Southern Annular Mode is comparable to the effect of ENSO on coastal rainfall on southwestern Australia in winter. Also, it is suggested by the current results that a positive phase of SAM dramatically affects frontal development in winter as well as strengthening the easterly troughs in summer, which produces up to 5 times the rainfall in some situations. The effects of SAM on rainfall reaches its most pronounced in the neutral phase of ENSO, which occurs more often than the El Niño reaches its most pronounced in the neutral phase of ENSO, which occurs more often than the El Niño and La Niña combined. E.g., during 1948-2010 only 22 Julys were classified as El Niño or La Niña, but there were 41 neutral Julys. The phase of SAM consequently has much greater influence during these neutral months, which results in a long-term trend in southwestern Australian rainfall. Also, the change in ENSO from El Niño to neutral conditions affects significantly the monthly rainfall while the change from neutral to a La Niña affects it only slightly. The monthly SAM has been shown to influence strongly the rainfall in southwestern Australia, which contrasts with the results of Meneghini et al., 2007). Raut et al. suggest the difference might be due to the inability of correlation analysis to capture the exact strength of the nonlinear and interdependent relationship between SAM index and rainfall.
It has been found (Frederiksen & Frederiksen, 2007; Frederiksen et al., 2011) that reduced strength of the Southern Hemisphere subtropical jet stream and an associated displacement towards the Pole of storm tracks is linked to declining coastal rainfall in winter as well as enhanced rainfall in summer on inland areas. During the positive phase of SAM, the polewards shift in the subtropical jet increases the precipitation at the poleward flank of the jet and decreases it in winter over the subtropical latitudes (Hendon et al., 2014). A southwards shift in summer in the westerlies that are associated with the positive phase of SAM allows the penetration of the easterly troughs more frequently into higher latitudes.
The recent phase 5 of the Climate Model Intercomparison Project (CMIP5) simulations for this century that used the representative concentration pathway 4.5 (RCP4.5) greenhouse gas emission scenario, show a very weak negative trend in the SAM index, which contrasts with a strong positive trend that is projected when the RCP8.5 scenario is used (Zheng et al., 2013). It has also been shown (Polade et al., 2014) that there are 10-20 fewer rainy days and at least a 10% reduction in the total annual rainfall over southwestern Australia in the CMIP5 models for the RCP8.5 emission scenario compared to historical simulations. Raut et al. suggest that in light of these results the current study should be extended by use of the CMIP5 simulations of SAM, ENSO, and rainfall over southwestern Australia.
|Author: M.H.Monroe Email: firstname.lastname@example.org Sources & Further reading|