资源环境科技发展态势分析平台

Over the last century, significant declines in rainfall across the state of Hawai'i have been observed, and it is unknown whether these declines are due to natural variations in climate, or manifestations of human-induced climate change. Here, a statistical analysis of the observed rainfall variability was applied as first step towards better understanding causes for these long-term trends. Gridded seasonal rainfall from 1920 to 2012 is used to perform an empirical orthogonal function (EOF) analysis. The leading EOF components are correlated with three indices of natural climate variations (El Nio-Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO), and Pacific North American (PNA)), and multiple linear regression (MLR) is used to model the leading components with climate indices. PNA is the dominant mode of wet season (November-April) variability, while ENSO is most significant in the dry season (May-October). To assess whether there is an anthropogenic influence on rainfall, two methods are used: a linear trend term is included in the MLR, and pattern correlation coefficients (PCC) are calculated between recent rainfall trends and future changes in rainfall projected by downscaling methods. PCC results indicate that recent observed rainfall trends in the wet season are positively correlated with future expected changes in rainfall, while dry season PCC results do not show a clear pattern. The MLR results, however, show that the trend term adds significantly to model skill only in the dry season. Overall, MLR and PCC results give weak and inconclusive evidence for detection of anthropogenic signals in the observed rainfall trends.