The Indian Ocean Zonal Mode over the past millennium in observed and modeled precipitation isotopes

The Indian Ocean Zonal Mode (IOZM) has gained considerable attention in the past decade due to its role in causing widespread flooding and droughts in the continents and islands surrounding the Indian Ocean. The IOZM has also been observed to vary on a low-frequency (multi-decadal) basis, making its behavior important to understand both for mid-range 21st century climate prediction and for paleoclimate studies. Despite efforts to reconstruct the IOZM using corals and other high-resolution proxies, nonstationarities in the response of paleoclimate proxies to the IOZM have also been noted, raising the possibility that the IOZM may be difficult to reconstruct or to predict in the long-term. It is therefore critical to assess the low-frequency component of the IOZM in observed, modeled, and paleoclimate data from the Indian Ocean region in order to identify nonstationary behavior and to assess its role in lowfrequency climate variations. We present an analysis of low-frequency and nonstationary behavior in the IOZM on multi-decadal to centennial timescales using a combination of modeled, observed, and proxy reconstructions of d18O/ dDprecip. In order to assess multiple timescales of low-frequency variability, we focus on two key time periods: the historical period (1870e2003), and the past millennium (1000 C.E.-present). We find significant nonstationarities in the relationships between the IOZM, precipitation amount, and dOprecip/dDprecip during the historical period. These relationships vary on a multi-decadal basis in our model and in observed/reanalysis datasets. Air-sea interactions in the Indo-PacificWarmPool and teleconnections to the Pacific Ocean, as well as the phase of the IOZM itself, may contribute to this nonstationary behavior. We examine the potential ramifications of nonstationary IOZM behavior using a synthesis of spatially distributed proxy archives of dOprecip/dDprecip from both sides of the IOZM region spanning the past millennium. Our findings indicate that during the past millennium, a strong IOZM-like connection exists in the proxy data network, with anti-correlation between East Africa and Indonesia. However, the links are spatially limited and in some cases timescale-dependent. Nonlinear behaviors in these links suggest that the IOZM may be difficult to detect on a consistent basis in proxy records from the past millennium. Based on our modeling results, the inconsistent links in the IOZM proxy network may arise from temporally and spatially variable relationships between the IOZM, precipitation, and dOprecip/dDprecip. We conclude that the IOZM's potential to influence the climate of the Indian Ocean region is inconsistently reflected in proxy data, and that due to the changing strength of the IOZM/dOprecip/dDprecip relationship, its spatial “footprint” may be restricted on multi-decadal to multi-centennial timescales. © 2014 Elsevier Ltd. All rights reserved. .edu (B. Konecky).