Climate variability has large impacts on humans and their agricultural systems. Farmers are at the center of this agricultural network, but it is often agricultural planners—regional planners, extension agents, commodity groups and cooperatives—that translate climate information for users. Global climate models (GCMs) are a leading tool for understanding and predicting climate and climate change. Armed with climate projections and forecasts, agricultural planners adapt their decision-making to optimize outcomes. This thesis explores what GCMs can, and cannot, tell us about climate variability and change at regional scales. The question is important, since high-quality regional climate projections could assist farmers and regional planners in key management decisions, contributing to better agricultural outcomes. To answer these questions, climate variability and its regional impacts are explored in observations and models for the current and future climate. The goals are to identify impacts of observed variability, assess model simulation of variability, and explore how climate variability and its impacts may change under enhanced greenhouse warming.
Chapter One explores how well Goddard Institute for Space Studies (GISS) atmospheric models, forced by historical sea surface temperatures (SST), simulate climatology and large-scale features during the exceptionally strong 1997–1999 El Niño Southern Oscillation (ENSO) cycle. Reasonable performance in this 'proof of concept' test is considered a minimum requirement for further study of variability in models. All model versions produce appropriate local changes with ENSO, indicating that with correct ocean temperatures these versions are capable of simulating the large-scale effects of ENSO around the globe. A high vertical resolution model (VHR) provides the best simulation. Evidence is also presented that SST anomalies outside the tropical Pacific may play a key role in generating remote teleconnections even during El Niño events.
Based on the results from Chapter One, the analysis is expanded in several ways in Chapter Two. To gain a more complete and statistically meaningful understanding of ENSO, a 25 year time period is used instead of a single event. To gain a fuller understanding of climate variability, additional patterns are analyzed. Finally analysis is conducted at the regional scales that are of interest to farmers and agricultural planners. Key findings are that GISS ModelE can reproduce: (1) the spatial pattern associated with two additional related modes, the Arctic Oscillation (AO) and the North Atlantic Oscillation (NAO); (2) rainfall patterns in Indonesia; and (3) dynamical features such as sea level pressure (SLP) gradients and wind in the study regions. When run in coupled mode, the same model reproduces similar modes spatially but with reduced variance and weak teleconnections.
Since Chapter Two identified Western Indonesia as the region where GCMs hold the most promise for agricultural applications, in Chapter Three a finer spatial and temporal scale analysis of ENSO's effects is presented. Agricultural decision-making is also linked to ENSO's climate effects. Early rainy season precipitation and circulation, and same-season planting and harvesting dates, are shown to be sensitive to ENSO. The locus of ENSO convergence and rainfall anomalies is shown to be near the axis of rainy season establishment, defined as the 6–8 mm/day isohyet, an approximate threshold for irrigated rice cultivation. As the axis tracks south and east between October and January, so do ENSO anomalies. Circulation anomalies associated with ENSO are shown to be similar to those associated with rainfall anomalies, suggesting that long lead-time ENSO forecasts may allow more adaptation than 'wait and see' methods, with little loss of forecast skill. Additional findings include: (1) rice and corn yields are lower (higher) during dry (wet) trimesters and El Niño (La Niña) years; and (2) a statistically significant negative relationship exists between malaria cases and ENSO.
The final chapter adds climate change to the climate variability story. Under high CO2, the model able to capture ENSO dynamics—an atmospheric model coupled to the Cane-Zebiak ocean model ('C4' here)—generates more El Niño-like mean conditions in the tropical Pacific. These changes produce a 4x larger increase in maximum precipitation with warming in C4 than an atmospheric model with a slab ocean (Q4), dramatically enhancing the Pacific Hadley and Walker circulations, and through positive feedbacks, increasing the global temperature. Near Nordeste warming alone (Q4) produces added rainfall, which in C4 is partially cancelled out by El Niño-like changes in the Walker Cell. Both Q4 and C4 produce small changes in Indonesia, although C4 generates large circulation and precipitation anomalies over the Western Indian Ocean. C4 changes in the midlatitudes produce a very strong Pacific North American pattern (PNA) response that dominates a small positive AO change associated with Q4. These PNA changes produce increased rainfall over the Southeastern United States (SEUS) in C4. AO and NAO-like variability are also found to increase with enhanced CO2.
This thesis highlights how climate variability influences regional climate variability, with an emphasis on four regions: Nordeste, Brazil, Western Indonesia, the Southeastern United States (SEUS), and the Mediterranean. It links El Niño-driven delay in the onset of rainy season drivers in Western Indonesia to decision-making about when to plant the year's largest crop. In a coupled configuration, the GISS GCM produces strong El Niño-like changes with global warming. This result suggests that the impacts—climatological and agricultural—of climate change may ultimately exceed the impacts of current variability. Somewhat paradoxically, these results indicate that one of the central manifestations of climate change is likely to be changes in patterns of climate variability and their regional impacts.