In this thesis I present the results of a comprehensive assessment of the Pacific-North American (PNA) teleconnection pattern in general circulation models (GCMs) and a regional climate model (RCM). The PNA teleconnection pattern is a quasi-stationary wave field over the North Pacific and North America that has long been recognized as a robust feature of Northern Hemisphere atmospheric circulation, and directly affects the interannual variability of North American temperature and precipitation. The teleconnection is evaluated under present (1950-2000) and future (2050-2100) climate in a coupled GCM (MPI/ECHAM5) and a high-resolution regional climate model (RegCM3). I further assess the PNA in 27 atmosphere-ocean GCMs and earth system models (ESMs) from the ongoing fifth phase of the Coupled Model Intercomparison Project (CMIP5). The National Centers for Environmental Prediction and Atmospheric Research (NCEP/NCAR) Reanalysis serves a quasi-observational baseline against which the models are evaluated. For each analysis, changes in the spatial and temporal patterns of the PNA spatial are assessed for both the present and future climates, and these changes are then related to changes in climate and surface hydrology in North America. Coupling the NCEP and ECHAM5 GCMs with RegCM3 is very successful in that the PNA is resolved in both models with little loss of information between the GCMs and RegCM3, thereby allowing an assessment of high-resolution climate with an inherent skill comparable to that of the global models. The value of the PNA index is generally independent of the method used to calculate it: three- and four-point modified linear pointwise calculations for both the RegCM3 and ECHAM5 model simulations produce very similar indices compared with each other, and compared with those extracted from a rotated principle component analysis (RPCA) which is also used to determine the PNA spatial pattern. The spatial pattern of the PNA teleconnection emerges as a leading mode of variability from the RPCA, although the strength of the teleconnections are consistently weaker than NCEP as defined by four main "centers of action". This discrepancy translates into the strength of the controls of the PNA on surface climate. Maps of the correlations between the GCM PNA indices and RCM surface climate variables are compared to the results from the NCEP/NCAR Reanalysis. I find that correlation patterns with temperature and precipitation are directly related to the positioning of the Aleutian low and Canadian high, the two main drivers of upper-atmospheric circulation in the PNA sector. The CMIP5 models vary significantly in their ability to simulate the quasi-observed features of the PNA teleconnections. The behavior of the models relative to NCEP is more definite than the trends within the models. Most models are unable to resolve the temporal variability of NCEP; however, on the other hand most of the models are able to capture the PNA as a low-frequency quasi-oscillation. Many of the models are unable to simulate the barotropic instability that initiates wave energy propagation through the 500-hPa geopotential height field, thereby leading to phase-locking and thus the positive and negative modes of PNA are indistinguishable. The behavior and the spatial patterns of the PNA throughout the 21st century are consistent with other projections of future climate change in that most models exhibit a lengthening of the eddy length scale and a poleward shift of the mid-latitude jet stream associated with polar amplification of greenhouse-gas driven global warming. Finally, my analyses underscore the robustness of multi-model means, suggesting that the cumulative results of multiple climate models outperform the results from individual models because ensemble means effectively cancel discrepancies and hereby expose only the most robust common features of the model runs. While ensembles provide better representation of the average climate, they potentially mask climate dynamics associated with inter-annual and longer time scales. Relying on ensemble means to limit model spread and uncertainties remains a necessity in using models to project future climate. / Graduation date: 2013

The El Niño Southern Oscillation (ENSO), which refers to a coupling between equatorial Pacific Ocean and atmosphere anomalies, is a major source of interannual climate variability. Although it is fundamentally a tropical Pacific phenomena, both warm (El Niño) and cold (La Niña) events alter atmospheric circulations -- and subsequently temperature and precipitation patterns -- well into the mid- latitudes. Furthermore, both El Niño and La Niña have characteristic multi-year life cycles of sea surface temperature and zonal wind anomalies. The research in this thesis focuses on understanding whether the global teleconnections and multi-year evolution of El Niño and La Niña imposes a risk of synchronous or sequential crop failures relevant to global food production. In the first chapter, which focuses on maize, wheat and soy in the Americas, we analyze the dynamics underlying ENSO life cycles to illustrate which aspects of the system are most important for agriculture. In North America, the same-season teleconnections affecting soybean and maize have been well studied, but we demonstrate the importance of lagged soil moisture teleconnections for wheat in the southern Great Plains. In South America, peak ENSO sea surface temperature (SST) teleconnections are concurrent with, and therefore critical for, wheat and maize growing seasons while soil moisture memory in Argentina plays an important role during the soybean growing season In the second chapter we show how the teleconnections from chapter one lead to correlated crop production anomalies in North and South America. We estimate the magnitude of ENSO-induced Pan-American production anomalies and discuss how increasing crop harvesting frequency may affect Pan-American production variability. We find that ENSO-induced production anomalies are greatest for maize, with median anomalies of about 5% of Pan-American production. After broadly characterizing ENSO-induced production anomalies, we demonstrate that they are not static in time. Increasing crop harvesting frequency in Brazil has affected the correlated risks posed by ENSO to soybeans and maize. In the third chapter we expand our analysis of agriculturally relevant teleconnections to the greater Pacific Basin region, and move beyond observations into model simulations. In this chapter we propose a coherent framework for understanding how trans-Pacific ENSO teleconnections pose a correlated risk to crop yields in major agricultural belts of the Americas, Australia and China over the course of an ENSO life cycle. The potential for consecutive ENSO-induced yield anomalies is of particular interest in these major food producing areas, where modest changes in yield have significant effects on global markets. We demonstrate that ENSO teleconnections relevant for crop flowering seasons are the result of a single trans-Pacific circulation anomaly that develops in boreal summer and persists through the following spring. These trans-Pacific ENSO teleconnections are often (but not always) offsetting between major producing regions in the Americas and those in northern China or Australia. Multi-year La Niñas, however, only tend to force multi-year growing season anomalies in Argentina and Australia. In our final chapter we estimate of the relative contribution of major modes of climate variability to crop yield variability at the global scale. We consider the influence of not only ENSO, but also the Indian Ocean Dipole (IOD), tropical Atlantic variability (TAV) and the North Atlantic Oscillation (NAO). We find that modes of climate variability account for 18.4%, 7.4% and 5.4% of globally aggregated maize, soy and wheat production variability, respectively. All modes of variability are important in at least one region studied, but only ENSO has a significant influence on global production. The low fractions of global-scale soy and wheat production variability attributable to climate is a result of significant but offsetting ENSO-induced yield anomalies in major production regions. Our findings represent an observationally-derived limit on the importance of climate variability to crop production stability that is not dependent on the fidelity of present generation of climate or crop models. In terms of synchronous crop failures within a single harvest year, we find that ENSO poses a significant correlated risk to maize yields but that it has a much smaller effect on global wheat and soy production. ENSO-forced maize production anomalies offset less than wheat and soy at the global scale because production is concentrated in regions with same-sign yield anomalies, notably the United States and Southeast Africa. To illustrate this point, we show that ENSO is largely responsible for the largest synchronous maize failure in the post-1960 historical record. These results demonstrate how the distribution of global cropland in relation to ENSO teleconnections contributes significantly to the presence for maize or absence for wheat and soy of synchronous global crop failures

Recent studies have found that equatorial Atlantic sea surface temperature (SST) variability may be influencing tropical Indian Ocean climate (Kucharski 2009, Wang 2009). Due to the economic and social impact of tropical cyclones, it is important to investigate how an Atlantic-Indian Ocean connection may be affecting tropical cyclone behavior in the southern Indian Ocean. In this study, the International Best Track Archive for Climate Stewardship (IBTrAC) tropical cyclone database is used to derive metrics of tropical cyclone behavior, which are then compared with indices of tropical Atlantic SST variability representing Atlantic Niño, and Benguela Niño events. Changes in tropical Atlantic SSTs are found to coincide with significant differences in tropical cyclone activity for portions of the southern Indian Ocean. In addition, for these same regions, tropical Atlantic SST variability is associated with changes in large-scale atmospheric conditions, including steering flow, low level vorticity, and humidity, typically associated with tropical cyclogenesis, and tropical cyclone track. The changes in steering flow related to both indices of Atlantic SST anomaly are reproduced by an atmospheric model. The changes in steering flow are also found to be linked to changes in TC translational velocity, and TC tracking. These findings indicate a possible link between tropical Atlantic conditions and cyclone activity in the Indian Ocean mediated through a teleconnection between tropical Atlantic SSTs and large scale atmospheric conditions over the southern Indian Ocean. The teleconnection related to the Benguela Niño region of SST variability was found to consist of a Rossby wave initiated off the coast of South America, and propagating into the Indian Ocean, thereby influencing several atmospheric variables, including steering flow. The teleconnection related to the Atlantic Niño region of SST variability was not well defined, although there was some evidence of a Walker circulation anomaly extending from the equatorial Atlantic over the continent of Africa, and influencing SIO steering flow. / Graduation date: 2012

Indiana University-Purdue University Indianapolis (IUPUI) / Precipitation trends and their driving mechanisms are examined over a variety of spatial and temporal scales using a multi-proxy, decadally-resolved sediment record from Martin Lake that spans the last 2300 years. This unique archive from a northern Indiana kettle lake documents significant climate variability during the last 2 millennia and shows that the Midwest has experienced a wide range of precipitation regimes in the late Holocene. Three independent proxies (i.e., oxygen and carbon isotopes of authigenic carbonate and %lithics) record variations in synoptic, in-lake and watershed processes related to hydroclimate forcing, respectively. Together, these proxies reveal enhanced summer conditions, with a long period of water column stratification and enhanced summer rainfall from 450 to 1200 CE, a period of time that includes the so-called Medieval Climate Anomaly (950-1300 CE). During the Little Ice Age, from 1260 to 1800 CE, the three proxy records all indicate drought, with decreased summer rainfall and storm events along with decreased lake stratification. The Martin Lake multi-proxy record tracks other Midwest climate records that record water table levels and is out-of-phase with hydroclimate records of warm season precipitation from the High Plains and western United States. This reveals a potential warm season precipitation dipole between the Midwest and western United States that accounts for the spatial pattern of late Holocene drought variability (i.e., when the Midwest is dry, the High Plains and the western United States are wet, and vice versa). The spatiotemporal patterns of late Holocene North American droughts are consistent with hydroclimate anomalies associated with mean state changes in the Pacific North American teleconnection (PNA). Close associations between late Holocene North American hydroclimate and records of Northern Hemisphere temperatures and the Pacific Ocean-atmosphere system suggests a mechanistic linkage between these components of the global climate system that is in line with observational data and climate models. Based on our results, predominantly –PNA conditions and enhanced Midwestern summer precipitation events are likely to result from continued warming of the climate system. In the western United States, current drought conditions could represent the new mean hydroclimate state.