Funded Projects

Program (s): Climate Variability and Predictability

Competition:

Award Number: NA13OAR4310156 OR NA13OAR4310154 | View Publications on Google Scholar

We propose to examine the sensitivities of tropical convective activity to humidity and temperature profiles in the environment, based on new field observations from the DYNAMO experiment. The sensitivities can be succinctly described in the framework of a time-independent matrix, which we propose to estimate as a novel activity in this project. In order to advance these objectives, we propose two linked lines of activity:

1. Data processing of special field observations to produce profiles of moisture, radar echo statistics (cloud systems) and wind divergence (indicative of convective activity and heating profiles), as input to the estimation;

2. Estimation of the sensitivity matrix (or linear response function), based on these and other data (soundings, satellite), guided by first-guess estimates derivable from existing cloud-resolving model results, and subdivided into convecting regimes as appropriate.

Knowing the sensitivity matrix (or a few such matrices, for different ‘regimes’ of convection activity) is desirable for several reasons. First, it helps advance understanding of convection variations and their impacts, including the Madden-Julian Oscillation (MJO) and other climate variations. The sensitivity matrix quantitatively attributes a relative importance to temperature and moisture anomalies at different altitudes in explaining convection or rainfall anomalies. The matrix can also be compared directly to its counterpart in climate models, acting as an inclusive standard and guide for the development of better moist physics packages to improve climate simulation and prediction.
In this way, our work will support core capabilities of NOAA in understanding and modeling of the climate system, as well as in observing system design and use. Through its vertical integral, the convection sensitivity matrix allows prediction of rainfall based on thermodynamic input profiles, which may be a useful tool informing the societal challenges of better understanding and prognosis of Climate Impacts on Water Resources, Changes in Extremes of Weather and Climate, and better models that can provide Information for Mitigating and Adapting to Climate Change, NOAA’s long-term goal in its Next-Generation Strategic Plan.

Program (s): Climate Variability and Predictability

Competition: Improved Understanding of Tropical Pacific Processes, Biases, and Climatology

Award Number: NA14OAR4310277 | View Publications on Google Scholar

A salient feature of the tropical Pacific is the pronounced east-west gradient in sea surface temperature along the equator. This temperature gradient is coupled to the atmospheric zonal circulation and oceanic thermocline tilt. Together they define the Warm pool, Cold tongue, Walker circulation Complex (WCWC). Asymmetry in the SST and precipitation fields about the equator sets the position of the Intertropical Convergence Zone (ITCZ). This is a proposal to study the effect of meridional variation in cloud albedo across the entire Pacific basin on this tropical climatological state in Earth system models.

Coupled model simulations of the tropical Pacific, while improved over the past decades, still show significant biases. A cold bias within the cold tongue region, a cold tongue that extends far west into the warm pool, and the double-ITCZ problem remain persistent issues in climate models. The mean east-west temperature gradient along the equator in the CMIP5 models varies between 3 and 8 degrees C, compared to nearly 6 degrees C in the observations. Developing a comprehensive understanding of what determines this temperature gradient and related characteristics of the fully coupled ocean-atmosphere system is critical for understanding tropical climatology and potential biases in climate models. Often, this question is treated as a local problem in the equatorial band – improvements are sought by tuning local parameters affecting the properties of deep convection in the warm pool or the amount of shortwave radiation reaching the eastern Pacific. However, the strength of the cold tongue is ultimately controlled by the temperature of waters subducted in the extra-tropics and transported to the equator by the ocean subtropical cells (STC). Consequently to understand tropical biases one needs to treat this coupled problem in a broader geographical context and consider latitudinal variations in the main dynamical factors.

Our preliminary analysis indicates that the meridional gradient in cloud albedo is one of these key factors. A close relationship exists across the pre-industrial CMIP5 simulations and our preliminary numerical experiments that connects the mean east-west gradient in upper-ocean temperatures and the contrast in cloud albedo between the extra-tropical and tropical Pacific. For example, when extra-tropical cloud albedo is higher than observed, or tropical cloud albedo is too low, the east-west temperature gradient is stronger than observed. Thus, we propose that in coupled climate models the zonal SST gradient and the related characteristics of the tropical ocean-atmosphere system (e.g. zonal winds and the thermocline tilt) are largely controlled by the meridional gradient in cloud albedo between the equator and the extra-tropics. Further, we propose that it is the inter-hemispheric albedo contrast that controls the position of the ITCZ. To investigate these problems we will conduct (i) sensitivity experiments with CESM, in which we systematically modify cloud properties affecting cloud albedo, (ii) a theoretical analysis of the coupled system with cloud feedbacks included, and (iii) an analysis of CMIP5 models focused on the effects of latitudinal variations in cloud albedo on the tropical climate.

Relevance to long-term NOAA goals and current solicitation: The overarching goal of this study is to understand the fundamental physical mechanisms that control tropical climatology and model biases. This objective is directly relevant to the current solicitation and longer-term NOAA goals, since simulating tropical climate correctly is critical for climate prediction on a variety of timescales from seasonal to interannual, to decadal and longer. Funding from this grant will support cross-disciplinary training of a postdoctoral associate at Yale, Dr. Natalie Burls.

Program (s): Climate Variability and Predictability

Competition: AMOC-Climate Linkages in NA/SA

Award Number: NA16OAR4310172 | View Publications on Google Scholar

Statement of the Problem: To understand the causes of decadal-scale variability in Atlantic overturning waters it is necessary to both recognize and connect the changes that are occurring in a moving ocean. That is, from a climate perspective, there is a need to use available information to better understand not only how ocean properties are changing, but also how dynamics may be affecting changes. While models and re-analyses look to provide a moving and even predictive three-dimensional picture, both correct and incorrect details are often lost in integration of  available outputand deep signals, in particular, may be missing from numerical integrations. On the other hand, observations capable of providing details on the characteristics of deep ocean properties and processes are, more often than not, disconnected from one another in time.

Speaking to the AMOC competition’s aim to refine present knowledge of the AMOC state,  variability and change as well as NOAA’s long term goal of an improved understanding of the  changing climate system, the goals of this project are twofold: 1) to improve understanding of  changes in the deep South Atlantic Meridional Overturning Circulation (SAMOC) properties and transports through a statistical analysis and comparison of observations and numerical model  output, and 2) to develop an understanding of where and how two particular models are  succeeding and/or failing to capture observed deep signals thought to be signatures of climate  change. Repeat hydrographic lines will be used in combination with float data and Lowered Acoustic Doppler Current Profiles together with output from two high-resolution assimilative models (HYCOM and ECCO) to develop an understanding of what changes are occurring in the observed fields, where they are occurring, how such changes are or are not reflected in the numerical fields, and whether this matters. That is, what are the overall consequences to numerical prediction of discrepancies between modeled and observed deep and bottom water changes and transport? The observation-model comparative analysis is relevant to the AMOC competition’s aim of combining existing observations with models to refine our understanding of present-day and past AMOC circulation and transport. It is also relevant to NOAA’s goal of  informing future climate-scale predictions as it lo oks to determine the importance of specific  deep/abyssal pathways and particular regions of mixing to decadal simulations.

The proposed work includes a formal collaboration with Elaine McDonagh and colleagues at the  National Oceanography Centre in Southampton and informal collaborations with Edmo Campos  and SAMOC group working towards a long-term South Atlantic observational network, as well as Tonia Capuano at the Université de Bretagne Occidentale to assist in the calculation of mixing  estimates.

Program (s): Climate Variability and Predictability

Competition:

Award Number: | View Publications on Google Scholar

Dust storms from Africa are a persistent feature in the skies over the northern tropical Atlantic, and strong variability in Atlantic dust cover has been observed on seasonal to decadal time scales. It is well known that over water surfaces the net radiative effect from mineral aerosols at the surface is negative, and recent work has shown that this reduction in downwelling radiation translates into localized cooling of the mixed layer. Recent studies have also demonstrated that over the last quarter century roughly 25% of the observed upward trend in sea surface temperatures can be attributed to declines in Atlantic dust cover over the same time period. 

While there is compelling evidence suggesting that African dust storms contribute to Atlantic surface temperatures on decadal time-scales, to-date studies investigating dust-forcing of temperatures have generally neglected other important environmental factors that are associated with Atlantic dust outbreaks; namely the dry air, mid-level warm anomaly, and increased surface wind speed. The Saharan Air Layer (SAL) is the term given to this dry air mass that is associated with the dust, and the net effect of the reduction in water vapor, warm anomaly, and increase in surface winds is to further cool the mixed layer via negative longwave radiative forcing at the surface, and wind-driven latent and sensible heat fluxes, and vertical turbulent mixing. Therefore, it is likely that the SAL, considered in its entirety, has a stronger role in shaping Atlantic temperatures on monthly to decadal time scales than does dust alone. 

At the same time, satellite, in-situ, and proxy dust records all show that Atlantic dust cover has strong decadal variability, and recent work has shown that a simple statistical model can reproduce month-to-month variability in Atlantic dust cover by considering reanalysis winds, climate indices, and observational records. Therefore, the opportunity exists to reconstruct spatial and temporal Atlantic dust storm variability from the mid-20th century to the present. 

Here we propose to capitalize on studies that demonstrate techniques for modeling the radiative effects of the SAL, and that provide a basis for developing a statistical model to reconstruct historical dust cover, in order to estimate the impact of the Saharan Air Layer on Atlantic Ocean 4 surface temperatures over the last 60 years. By exploring the effect of the SAL on temperatures using an ocean general circulation model (GCM), we will quantify the effect of the SAL in shaping decadal scale surface temperature variability. We will also analyze any effect of the SAL on deep ocean temperatures and determine if there is a SAL contribution to the Atlantic meridional overturning circulation (AMOC). 

In order to evaluate how aerosol cover and atmospheric winds have impacted Atlantic tropical cyclone activity on decadal time scales, we propose the following course of action: 

1) Use several data sets to create a climatology of the SAL that extends back to the mid-20th century, estimating SAL vertical profiles of dust optical depth, water vapor, temperature, and winds. 

2) Employ established techniques to model SAL-forced changes in surface radiative fluxes and surface turbulent heat fluxes, along with observation-based calculations of horizontal oceanic heat advection and vertical turbulent mixing, to determine the dominant processes associated with the mixed layer temperature response to the SAL. 

3) Force an ocean GCM with SAL-forced surface radiative fluxes and surface wind anomalies in order to quantify the ocean temperature response to SAL variability, and determine to what degree the SAL contributes to meridional heat transport, from 1950 through the present.

The main objective of this proposal is to understand the role of the SAL in shaping Atlantic surface and subsurface temperature anomalies over the last 60 years, focusing on the SAL’s contribution to decadal scale upper ocean temperature variability, and the AMOC. 

Program (s): Climate Variability & Predictability

Competition: Decadal Climate Variability and Predictability

Award Number: NA20OAR4310400, GC20-204 | View Publications on Google Scholar

The subpolar North Atlantic is the only extensive region of the world ocean where the sea surface temperature (SST) has cooled since 1900 (IPCC 2013). This cooling occurs in the context of strong multi-decadal variability in SST. The causes of these features of the subpolar North Atlantic are currently debated, and include external forcing as well as externally forced and internally generated variability in the atmosphere and ocean circulation. Understanding the causes of past changes is critical for predicting how the ocean will evolve in the near-term future, as well as in long-term projections. Here we will address the causes of the cooling trend (the so-called ‘warming hole’) in the context of a highly variable subpolar North Atlantic. As a first step, we will dive into the causes of the trend itself, and evaluate a previously untested mechanism: that the cooling is the ocean response to a northward shifted North Atlantic jet. We will employ a climate model hierarchy that includes: a fully coupled model, a slab ocean model, a slab ocean with Ekman included, as well as forced ocean experiments. The mechanisms revealed in this hierarchy will also be tested in CMIP5/6 model experiments and observations. As a second step, we will examine fluctuations about this trend, and how the forced component can impact internally driven ocean and atmosphere variability. To do this we will develop a new framework that can applied across models for evaluating the time-evolving ocean conditions that can lead to predictability. We will also use a large-archive of CESM and CMIP5/6 simulations to evaluate the relationship between the ensemble-mean forced signal and the spread due to internally-generated variability. Our proposed work responds to the NOAA Climate Program Office’s "CVP - Decadal Climate Variability and Predictability" competition. In particular, we address the priority area of ‘Investigation of mechanisms that govern variability of the coupled climate system and its predictability on the interannual to multi-decadal timescales within longterm observation data and/or model data (such as, CMIP6), with a focus on either the Atlantic or Pacific Ocean region.’ By using a combination of the CESM model hierarchy, CMIP5/6 data, and observations, we will gain a process-level understanding of the ocean and atmosphere on the interannual to multi-decadal timescales, which will lead to greater confidence in our ability to predict future changes in the North Atlantic and their impacts.

Program (s): Climate Variability and Predictability

Competition: Improved Understanding of Tropical Pacific Processes, Biases, and Climatology

Award Number: NA14OAR4310275 | View Publications on Google Scholar

Previous studies have shown that extra-tropical influences from the North Pacific can impact the tropical Pacific climate, offering some additional degree of predictability of El Nino/Southern Oscillation (ENSO). More recent work has also shown that similar physical mechanisms operate in the Southern hemisphere, and further suggest that the connection between the southern extra-tropics and the equatorial climate is even more direct than the north. The reason is that southeasterly trade winds cross the equator and allow signals to propagate from the south deep into the tropics, while in the northern hemisphere signals are essentially ‘blocked’ by the convergence of winds in the northern ITCZ. Simulating this connection from the southern hemisphere is problematic in coupled GCMS in which an erroneous southern ITCZ can potentially block extra-tropical signals originating in the south. Thus, in this project we will test the hypothesis:

That Pacific ITCZ biases in climate models weaken the southern hemisphere influence on the equator and diminish a potential source of ENSO predictability.

We will test this hypothesis using a collection of climate model simulations that offer multiple realizations of the mean state biases in the Pacific ITCZ. We will perform diagnostic studies using CMIP5 model simulations to test the model-dependence of the mean state and its influence on variability propagating from the southern hemisphere to the equator. We will perform a large number of experiments with climate models in which the ITCZ position is altered in two ways: First by externally imposed perturbations to the energy budget of the model, and second by altering the strength of regional radiative feedbacks using a novel approach that we have developed in prior work (funded by NOAA). This dual method approach of altering ITCZ builds upon recent work that suggests that in addition to local processes in the Pacific that can lead to biases in the ITCZ, the mean ITCZ position is also influenced by processes outside the tropics that alter the radiative balance of the planet. This experimental approach has the advantage of being able to (1) test which regions of the globe are key for the simulation of the Pacific ITCZ, and (2) examine mean state interactions with variability in a consistent framework. Further we will use a hierarchy of models including aqua planet models, AGCM-slab ocean mixed layer models, and fully coupled models, which will allow us to identify the fundamental mechanisms which control the position of the ITCZ and impact Pacific climate variability. The ultimate goal of this work is to design an experimental framework in which we can test how potential sources of predictability, particularly from the southern hemisphere, are affected by Pacific ITCZ biases.

This work will contribute to the goals of the ‘ESS - Climate Variability and Predictability (CVP): Improving Understanding of Tropical Pacific’ by identifying processes and regions outside the tropical Pacific that exert a remote influence on the mean climate. This will help to provide a more complete and global context for understanding tropical Pacific biases, and suggest novel ways in which models can be developed to reducing this external influence on these biases. A unique aspect of this work is that we are focused on additional (an perhaps unexploited) sources of predictability in the Pacific climate system, namely signals or precursors from the southern hemisphere. By testing the central hypothesis or this project, our work will enhance NOAA’s core capabilities in both ‘Modeling and understanding’ and in ‘Predictions’ of the Pacific climate system.

Program (s): Climate Variability and Predictability

Competition:

Award Number: | View Publications on Google Scholar

The World Climate Research Program’s Working Group on Coupled Modeling will be carrying out a coordinated set of model experiments that includes, for the first time, simulations of decadal climate prediction. The ultimate goal of these simulations will be to provide policymakers with information on decadal timescales to assess possible consequences of climate change. To what extent these experiments will be useful to stakeholders and policymakers will depend upon whether there is a predictable signal of climate change and to what extent this signal varies on regional scales. In this proposed research we will focus on systematic errors in the predictable signal forced by sea surface temperature (SST) biases in the coupled model’s response to external forcing. In addition, we will investigate how these model biases limit predictability by impacting the spatial and temporal structure of natural variability. An active hypothesis is that the predictable signal of climate change comes from low-frequency ocean variability and it’s forcing of the atmosphere. We will explore this hypothesis by studying how systematic biases in Pacific Ocean SSTs impact the decadal predictability of the hydrological cycle over North America, focused primarily on the following two questions: 

1. To what extent is the predictable decadal signal over North America related to the spatial pattern of SST anomalies in the Indian and Pacific Ocean basins? 

2. Do systematic biases in Indian and Pacific Ocean SSTs impact potential predictability over North America by forcing regional variations in the climate signal, as well as, biases in the spatial and temporal structure of natural variability? 

Based on the results of previous studies, we will use model output from coupled climate model simulations of the 20th Century as unassimilated decadal climate predictions. We will then use AGCM model studies forced by SSTs output from these simulations to determine how biases in the models’ response to radiative forcing (through lowfrequency ocean variability) impact the decadal predictability of the hydrological cycle over North America. We will focus on identifying physical mechanisms that cause biases in predictability over North America, such as biases in the structure of the PDO. We will study the decadal predictability of the hydrological cycle by focusing our analysis on the variability of rainfall, surface temperature, and circulation patterns over North America. We will investigate strategies to correct for model biases in SSTs thereby improving probabilistic projections of decadal climate forecasts.

Program (s): Climate Variability and Predictability

Competition: Understanding Arctic Sea Ice Mechanisms and Predictability

Award Number: NA15OAR4310172 | View Publications on Google Scholar

The annual cycle of sea ice in the Arctic and marginal ice zones is strongly affected by the flux of heat from the ocean to sea ice. This flux is mediated by a number of processes:

1. During the summer, solar radiation can penetrate below the seasonal mixed layer. This is mediated by colored dissolved organic material, whose concentration is thought to be increasing in the Arctic, and by the presence of clear ice-melt layers. Neither of these processes is well-represented in the current generation of GFDL coupled climate models.
2. During the winter, this heat can be returned to the mixed layer by mixing, and additional heat is added from Atlantic waters entering the Arctic. The ease with which this occurs depends on the amount of freshwater stored in the mixed layer and the depth to which this water is mixed. Understanding the evolution of freshwater anomalies within the Arctic may therefore be important for predicting the future of Arctic sea ice.
3. Once it reaches the mixed layer, the heat must be transferred to the sea ice by turbulent exchange. The current version of the GFDL model parameterizes this exchange in a relatively crude fashion, using a heat transfer coefficient that is independent of the friction velocity.
Our proposal will carry on work currently being done at Johns Hopkins to look at all three processes.

Graduate student Grace Kim has recently developed a new parameterization of solar absorption which includes colored dissolved material (CDM), an important absorber of light in the open-ocean Arctic waters, and implemented this within the GFDL CM2Mc model (Galbraith et al., 2011). She finds that the inclusion of CDM produces significant regional changes in Arctic ice cover with a small overall increase. It is thus possible that increasing CDM in Arctic rivers and increasing chlorophyll in the Arctic interior will serve as a negative feedback on sea ice loss. We propose to expand this work to higher resolution GFDL models, particularly ESM2G, which has a very different mixed layer scheme.

Tom Haine's group has worked extensively on the processes maintaining the freshwater anomalies in the Arctic. A particular question we wish to examine is whether changes in Arctic freshwater storage will modulate both the seasonal storage of heat and the supply of heat from warm Atlantic waters. This work will be done using high-resolution models of the Arctic previously used in Haine's group. Postdoctoral support will be requested for this task.

Graduate student Eshwan Ramadu in Gnanadesikan's group has been examining the impact of relaxing the assumption of a constant (relatively large) heat transfer coefficient within the GFDL ice model. This work has found that increasing this coefficient (as might be expected to occur with thinner sea ice) leads to a build up of freshwater in the Beaufort Sea and a reduction of sea ice in the marginal ice zone. We have also started simulations where the ice-ocean heat transfer coefficient is replaced with one that is dependent on the friction velocity. Support is requested to continue this work.

We anticipate that all of these projects will be conducted in collaboration with Gnanadesikan's former colleagues at the Geophysical Fluid Dynamics Lab, particularly Bob Hallberg and John Dunne.

Program (s): Climate Variability & Predictability

Competition: COM and CVP: Innovative Ocean Dataset/Product Analysis and Development for support of the NOAA Observing and Climate Modeling Communities

Award Number: NA21OAR4310256 | View Publications on Google Scholar

While Earth System Models often fail to reproduce biological fields like phytoplankton biomass and chlorophyll, the reasons behind such failures are complex. Because phytoplankton growth rates depend on environmental conditions like nutrients and light, and these in turn depend on the rates of mixing and upwelling, physical biases in models can produce biases in circulation such that a “perfect” biological model will still give imperfect results. For example, an Earth System Model in which the relationship between macronutrients and biomass matches that found in the real North Atlantic will still produce a spatial bias in the distribution of nutrients if the path of the North Atlantic Current is poorly simulated. If we knew that the model had the correct relationship between biomass and nutrients, we could unambiguously tie such an error to model physics. However, the actual apparent relationships (those seen in the real world between environmental drivers and phytoplankton biomass) are far from simple and may deviate from intrinsic relationships based on bench science which are often coded into models. For example, low phytoplankton biomass may be associated with low levels of nutrients in the presence of high levels of light, or high levels of nutrients in the presence of low levels of light. Simply plotting biomass against nutrients will then result in a maximum biomass concentration at intermediate levels of nutrient, capturing the asymptote of biomass at high levels of nutrient may require careful extrapolation. Better constraining the drivers of phytoplankton change and variability is essential to NOAA’s mission to improve the prediction of the Earth System in order to build resilience to changes. The proposal directly addresses the call within the competition to “examine biases in observed and modeled data/products and advance understanding of the causes for large differences between observed and modeled ocean data/products.” We aim to build on recent work showing that machine learning methods (in particular, Neural Network Ensembles) can be used to extract biologically reasonable complex relationships from ESMs and also used to compare the similarity of the biological codes across models. We propose to examine whether such methods can find robust relationships between biomass and observed environmental parameters on regional and global scales, and use the resulting relationships as metrics for evaluating Earth System Model output. We will do this using combinations of remotely sensed data (chlorophyll, carbon biovolume) and in-situ data (phytoplankton biomass, nutrients, Ekman upwelling, light, mixed layer depth). We will also develop a toolkit whereby Earth System Models that are part of the current IPCC process can be compared with observational relationships and to each other.

Program (s): Climate Variability and Predictability

Competition: AMOC-Climate Linkages in NA/SA

Award Number: NA16OAR4310174 | View Publications on Google Scholar

Transient tracers offer a unique and important window into the Atlantic Meridional Overturning Circulation (AMOC). While they have been used to estimate the total rate of formation of NADW and to trace the pathways by which watermasses spread, less attention has been paid to the ways in which tracers can tell us about how the overturning is changing. Increasing amounts of observational data and model simulations with transient tracers offer new opportunities to understand the relationship between transient tracers and the large-scale ocean circulation. This proposal has three parts:
1. Analysis of coupled climate models: As demonstrated in this proposal, transient tracers in climate models can be better correlated with long-period variability in the overturning than spot measurements of the overturning itself. However, the fingerprint of overturning variability in tracers such as ideal age and oxygen has a complex three-dimensional spatial structure, with different responses at different latitudes. We propose to conduct a cross-model comparison of different models developed at NOAA GFDL as well as Earth System Models that are part of the IPCC AR5 model intercomparison. We will examine how robust the correlations with overturning are across models and evaluate how much data is required to extract them from other modes of variability. Anand Gnanadesikan will lead this part of the project.
2. Observational data analysis: We will analyze repeat hydrographic sections (Line W between Cape Cod and Bermuda, and elsewhere) to examine whether the patterns of changes in ventilation age and oxygen seen in the models also show up in the observations. We will also explore whether transient tracers in the observations are linked to changes in the stratification, as they appear to be in the models. If found, robust relationships between oxygen, stratification and age would enable a much broader mapping of changes in ventilation pathways within the North Atlantic. Darryn Waugh will take the lead in this part of the project.
3. High-resolution modeling. In order to better understanding the sources of the tracer fingerprints we see in the coupled models, we propose to conduct some high-resolution (1/10 degree) regional simulations of the North Atlantic in which different forcings are applied to change the overturning circulation and the resulting fingerprints on the tracer fields are computed. Insofar as we see the same results as the lower-resolution simulations, these results will help establish the robustness of tracer signatures of overturning change. If they show very different results, this will point to the importance of eddy processes in setting up tracer anomalies. Thomas Haine will lead this part of the project.
Key products that will emerge from this work are maps of mean age change over the North Atlantic over time and hopefully, identification of fingerprints of AMOC variability on the age and oxygen fields. This could allow extension of our estimates of AMOC variability back in time, which would also help to constrain mechanisms for AMOC variability.