Relationships between rainfall in the southwest of Western Australia and near-global patterns of sea-surface temperature and mean sea-level pressure variability

Abstract

Rainfall in the southwest of Western Australia (SWWA) has decreased since the middle of last century. The relationships between near-global patterns of mean sea-level pressure (MSLP) and sea-surface temperatures (SST) with rainfall in SWWA have been explored using principal component analysis (PCA) and path analysis, with the aim to determine whether large-scale patterns of MSLP can be implicated as a causal factor in the SWWA rainfall decrease. By analysing the relationships between the most significant rotated MSLP PCA patterns and SWWA rainfall, over the period 1948 to 2000, we show that the first near-global MSLP pattern has a strong simultaneous relationship with May to July SWWA rainfall, and that this relationship is stronger than that between rainfall and sea-surface temperature patterns. The MSLP-rainfall relationship suggests that the prevalence of positive PCA values of MSLP over the last two decades is related to the observed decrease of SWWA rainfall, in the early part of the cool season. However, August to October rainfall has little relationship with the first MSLP pattern. In contrast, the second MSLP pattern has little relationship with May to July rainfall, but a strong relationship with August to October rainfall. By analysing both raw and detrended data, it was found that the relationship between MSLP and rainfall was mainly the result of a strong interannual relationship, rather than due to a trend or long time-scale MSLP-rainfall relationship. Significant lag relation ships, for example, between April MSLP pattern one and May to October rainfall, suggest that MSLP may be useful for predictive purposes. However, testing of these relationships in prediction mode indicated that little extra skill (above that achievable in current operational, SST-based, forecasts) seems feasible. Path analysis suggests that, simultaneously, SSTs affect SWWA rainfall primarily through their relationship with MSLP. However, for the lagged case, the results varied according to whether the data were detrended or not. For the lag detrended data, SSTs dominated the relationship; indicating that SSTs continue to be a good, as well as practical, choice for forecasting rainfall.