Regional analysis for the purposes of designing a reference climate station network in New Zealand

The aim of this study is to contribute to the design of a reference climate station network that captures the regional variability in the climate and climatology of New Zealand (NZ).

We performed our analysis using New Zealand Re-Analysis which is a high resolution (1.5km) convection-permitting atmospheric regional reanalysis dataset over NZ spanning ~20 years. We analysed the dataset to identify climate regions that are co-varying, have similar climatology and are likely to have a similar response to climate change. To identify co-varying climate regions, we performed a Principal Component Analysis on the dataset and reconstructed it to retain 95% variance. The reconstructed data is then clustered using techniques such as k-means and self-organising maps, with the number of k-means clusters chosen based on a combination of Silhouette score and gap statistics. Secondly, we cluster on daily climatologies to isolate regions with similar climate. Finally, we cluster the differences between present daily climatology with respect to the future climatology using the Coupled Model Intercomparison Project to identify the regions which are likely to have similar climate change signals.

Using the regions derived, we present a method to optimally select a reference network representing the components of covariance, climatology and climate change.