The potential of coralline algae as an indicator of climate in the Southern Hemisphere and for the evaluation of global climate models: a New Zealand case study

General circulation models (GCMs) are currently the most important tools for obtaining projections about future climate. In addition, they provide data input for regional atmospheric models that translate global climate change to regional and local scales where humans and environments face the impacts. To ensure the accurateness of their simulations, GCMs need to be evaluated as thoroughly as possible against past climate records, where one focus is on the so-called "historical period" (1850–present). However, the evaluation task is difficult for the period before World War II and earlier due to a frequent lack of reliable observations. This problem is exacerbated for the Southern Hemisphere, which has been notoriously understudied in comparison to the climate of the Northern Hemisphere.

In this DFG-funded project (grant no. 453305163), we utilize crustose coralline algae – a rather recently discovered proxy archive – to extend the observational time series of sea surface temperature (SST) in the New Zealand region back to ~1850. The SST reconstruction is then employed in GCM evaluation to assess their skill in representing the large-scale climate of New Zealand. Finally, high-resolution sensitivity simulations are obtained from a regional atmospheric model to investigate the added value of the advanced GCM selection to regional climate modeling.

In New Zealand, variations in SST are reflected on a variety of spatial and temporal scales and are statistically detectable through to temperature anomalies and changes in glacier mass balance in the high mountains of the Southern Alps. Therefore, a key focus of the regional atmospheric modeling is to investigate the physical mechanisms that transform large-scale SST signals into local, high-mountain climate and glacier mass anomalies. For this purpose, an accurate representation of SST by GCMs is crucial.

综合环流模型（GCMs）目前是获取未来气候预测的重要工具。此外，它们还为区域大气模型提供数据输入，这些模型将全球气候变化转化为人类和环境面临的区域和地方尺度的影响。为确保其模拟的准确性，GCMs需要尽可能地与过去的气候记录进行详尽的评估，其中重点关注所谓的“历史时期”（1850年至今）。然而，由于缺乏可靠的观测数据，对二战前及更早时期的评估任务十分困难。这一问题在南部半球尤为严重，与北半球的气候相比，南部半球的气候研究一直处于被忽视的状态。

在本项由德国研究联合会（DFG）资助的项目（资助编号：453305163）中，我们利用石质珊瑚藻——一种最近才被发现的地层代用档案——将新西兰区域的海表温度（SST）观测时间序列扩展至约1850年。随后，通过大语言模型（LLM）对SST进行重建，用于评估GCM在表征新西兰大规模气候方面的能力。最后，通过区域大气模型获得高分辨率敏感性模拟，以探究高级GCM选择对区域气候模型增加的价值。

在新西兰，海表温度的变化在多种空间和时间尺度上得到反映，并通过温度异常和南部阿尔卑斯山脉高山冰川质量变化进行统计学检测。因此，区域大气模型的关键焦点是研究将大规模海表温度信号转化为局部、高山气候和冰川质量异常的物理机制。为此，GCMs对SST的准确表征至关重要。