Diagnosing the ENSO teleconnection to precipitation over tropical land regions in observations, reanalysis, and climate models

The climate of the Tropics is characterized by complex interactions among the atmosphere, ocean, and land surface. On interannual timescales, the El Niño/Southern Oscillation (ENSO), the dominant mode of tropical interannual variability, substantially impacts tropical climate. Such remote effects are referred to as ENSO teleconnections. In this dissertation we explore how ENSO affects land region precipitation in the Tropics. A key objective here is to advance mechanistic understanding of the ENSO-tropical land region rainfall teleconnection in both observations and current-generation global climate models. The canonical basis for interpreting ENSO tropical teleconnections is the anomalous Walker circulation (Lau and Nath 1996; Klein et al. 1999), which is often formulated in terms of the dry shallow water equation model of Gill (1980). While this canonical view of the ENSO tropical teleconnection may account for its broad features, e.g., the widespread reduction in rainfall outside of the equatorial Pacific source region of ENSO during El Niño events, it is largely qualitative and descriptive. The quantitative analysis of the ENSO tropical teleconnection, e.g., how much is rainfall reduced for a given a given amount of anomalous sea surface temperature (SST) forcing in the ENSO source region, ultimately requires a more rigorous, process-based approach. This is especially true on regional scales for which the teleconnected response may be quite variable. As a first step toward a quantitative and mechanistic understanding of the ENSO rainfall teleconnection over tropical land, in Chapter 2, we build on prior work by Lyon [2004] and Lyon and Barnston [2005] demonstrating how the spatial extent of drought areas over tropical land responds to ENSO. Using available observations and a categorical approach for drought quantification, these earlier studies found that during ENSO warm phase (El Niño) conditions, spatially coherent and nearly simultaneous droughts develop over tropical land regions. Here, we apply an analogous approach to several ensembles of CMIP5 models in order to assess global climate model fidelity in capturing the time evolution of bulk tropical drought area and its scaling relationship with ENSO, as well as the projected behavior of drought area-ENSO relationships in a future warming climate. Considering first the ensemble of prescribed SST forcing (or “Atmospheric Model Intercomparison” [AMIP]-style) simulations, the observed and simulated tropical land region drought areal extents are found to reflect temporal evolution similar to the observations, with comparable sensitivity in the increase of tropical land area experiencing drought as the strength of El Niño increases. Turning to climate model projections of the El Niño-drought area scaling relationship in the future, we document an apparent decrease in future ENSO-tropical drought area sensitivity. At least some of this apparent decrease may be related to atmosphere-ocean coupling, as coupled model simulations over the recent few decades also exhibit lower sensitivity compared to the observations. The results in Chapter 3 directly extend the analysis of Chapter 2 to tropical continental-scale subdomains and to a comparison of drought behavior with the extent of anomalously wet, or pluvial, conditions. By studying the teleconnection of ENSO to droughts (and pluvials) at regional scales, we aim to quantify the extent to which behavior varies across the entire tropics. Over the Tropical Americas subdomain, both observed and simulated El Niño events are frequently associated with increasing drought area, with widespread agreement across and among the observations and models analyzed. On the other hand, the behavior over both Tropical Africa and Tropical Australasia subdomains is far less consistent: although some drought and pluvial occurrences over these regions correspond to ENSO events, there are also instances in which there is no apparent ENSO connection, pointing to other sources of variability, but also less agreement among observations and models. Moreover, we find that while the response to La Niña implies an increase in the areal extent of pluvials over the Tropics similar to the response to El Niño, there is an evident asymmetry in the way droughts and pluvials respond to El Niño and La Niña forcing in the different tropical regions. Given the pronounced signature of the observed ENSO rainfall teleconnection over the tropical Americas, we appeal in Chapter 4 to some simple theoretical frameworks as a means for further diagnosis over this region. Specifically, tropospheric temperature (TT) has previously been suggested as the mechanistic bridge between the Pacific source region of ENSO and the rest of the tropics. According to this “TT mechanism,” during an El Niño event, positive SST anomalies in the equatorial central and eastern Pacific drive co-located convection anomalies that warm the troposphere, with efficient wave dynamics acting to propagate this warming around the tropics, particularly in the upper troposphere. With respect to rainfall over the remote tropics (including the tropical Americas), the preferential warming of the upper troposphere is considered to stabilize the troposphere to deep convection, thereby reducing rainfall. Our analysis in Chapter 4 begins by examining monthly lead-lag relationships among NINO3.4 (an SST-based index of ENSO), upper-level TT, and rainfall in observations, ERA5 reanalysis, and the AMIP-style CMIP5 models. Although the lead/lag results show consistency among NINO3.4, TT, and rainfall, the relative phasing among these suggests a nuance, as peak NINO3.4-rainfall correlations in fact lead peak NINO3.4-TT correlations, i.e., the relationship between ENSO forcing (via TT) and the response (rainfall) is ambiguous. We show, however, that consideration of tropospheric moisture, which is also changing in response to ENSO, can yield a more consistent picture of the regional rainfall teleconnection over the tropical Americas. Moreover, in the context of the CMIP5 models, the intermodel variation in the strength of the simulated ENSO-rainfall correlation over the tropical Americas appears to bear some relationship to the sensitivity of regional moistening with ENSO. In order to elucidate the interplay of the ENSO forcing with the land surface and the potential role of land surface feedbacks in modulating the strength of the rainfall teleconnection, we perform an analysis of the anomalous surface energy budget. This analysis demonstrates a strong covariation of regional rainfall anomalies with downward surface shortwave radiation, consistent with decreased cloudiness associated with reduced rainfall during El Niño conditions enhancing net downwelling shortwave, which in light of the zero net surface flux constraint over land, is largely compensated by increased sensible heat flux. Relative to the ERA5 reanalysis, however, many of the models analyzed exhibit a much stronger shift of the turbulent surface flux partitioning toward sensible heat flux during El Niño, which tends to amplify the simulated rainfall reduction via positive feedback. Chapter 5 summarizes the principal findings of this dissertation and presents some directions for future work.Ph.D.Includes bibliographical reference