An Examination of El Ni~No and La Ni~Na Teleconnections to Sahel

Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2017 An Examination of El Niño and La Niña Teleconnections to Sahel and Guinea Coast Rainfall in the Context of the 1968 Rainfall Regime Change Thomas Ashley Vaughan

Follow this and additional works at the DigiNole: FSU's Digital Repository. For more information, please contact [email protected] FLORIDA STATE UNIVERSITY

COLLEGE OF ARTS AND SCIENCES

AN EXAMINATION OF EL NIÑO AND LA NIÑA TELECONNECTIONS TO SAHEL AND

GUINEA COAST RAINFALL IN THE CONTEXT OF THE 1968 RAINFALL REGIME

CHANGE

THOMAS ASHLEY VAUGHAN

A Thesis submitted to the Department of Earth, Ocean, and Atmospheric Science in partial fulﬁllment of the requirements for the degree of Master of Science

2017

Sharon Nicholson Professor Directing Thesis

Philip Sura Committee Member

Guosheng Liu Committee Member

The Graduate School has veriﬁed and approved the above-named committee members, and certiﬁes that the thesis has been approved in accordance with university requirements.

ii ACKNOWLEDGMENTS

First, I would like to express my deepest gratitude to Dr. Sharon Nicholson for affording me the opportunity to study under her guidance and for all of the hard work and time she has dedicated to me and all of her students in the Nicholson Climatology Lab over the years. When I first entered the master’s program at Florida State, I never thought I would end up researching rainfall in Africa. However, as fate would have it, I ended up in the Nicholson Lab studying African climate and it could not have been a more positive experience. I would also like to thank Douglas Klotter, who served as an invaluable resource when I needed data or had technical issues (of which there were many). In addition to Dr. Nicholson and Doug, many other faculty members have mentored and advised me during my graduate school career. Gratitude is owed to my committee members, Dr. Philip Sura and Dr. Guosheng Liu, for all of their input, guidance, and advice on this thesis and in classes. Additionally, I am grateful to Dr. Robert Hart and Dr. Jeff Chagnon for their friendship and unwavering support of me and all of my fellow graduate students in the department. I also want to thank all of my present and former colleagues in the Nicholson Lab (Adam Hartman, Sid King, and Amana Hosten) for the many discussions and for providing ideas for my project. Thank you to Aaron Swearingen and Enoch Jo for the many hours they spent helping me with computer programming and fine-tuning my Matlab codes. Thank you to my family in North Carolina and all the new friends I have made at Florida State for their loving support and patience with me throughout this process. Finally, special gratitude is owed to Matthew Dawson and Thomas McKenzie, two of my fellow graduate students whom have impacted my life more than they will ever know. I doubt I would have finished my master’s degree had it not been for Matt and Tom, and I will be forever grateful for their life advice, support, friendship, and brotherhood.

iii TABLE OF CONTENTS

List of Tables ...... vi List of Figures ...... vii Abstract ...... xii

1 INTRODUCTION 1 1.1 The West African Sahel and the Guinea Coast ...... 1 1.1.1 Geographical Overview ...... 1 1.1.2 Climatological Overview ...... 3 1.2 Motivation ...... 5

2 PRIOR RESEARCH 10 2.1 Characteristics of Sahel Rainfall ...... 10 2.1.1 The West African Monsoon ...... 10 2.1.2 Atmospheric Circulations and Jets ...... 11 2.1.3 Seasonal Rainfall Cycle and Intraseasonal Variability ...... 15 2.1.4 Spatial Variability and the Dipole Pattern ...... 17 2.1.5 Rainfall Trends in the Twentieth Century ...... 18 2.1.6 The 1968 Rainfall Regime Change ...... 19 2.1.7 Global Teleconnections ...... 20 2.2 El Niño/Southern Oscillation (ENSO) and West African Rainfall ...... 22 2.2.1 Introduction to ENSO ...... 23 2.2.2 El Niño and African Rainfall Teleconnections ...... 24 2.2.3 La Niña and African Rainfall Teleconnections ...... 28 2.3 Research Questions and Objectives ...... 28

3 METHODOLOGY 41 3.1 Time Frame of Study ...... 41 3.2 Rainfall Data ...... 42 3.2.1 Standard Departures ...... 42 3.2.2 Rainfall Regions ...... 43 3.2.3 Data Adjustments ...... 44 3.3 El Niño and La Niña Events ...... 45 3.3.1 Historical Deﬁnitions of ENSO ...... 45 3.3.2 Climate Prediction Center (CPC) Methodology ...... 48 3.3.3 ENSO Events in the Twentieth Century ...... 49 3.4 Rainfall Calculations ...... 51 3.5 An Examination of Atmospheric Circulations ...... 51

iv 4 RESULTS AND DISCUSSION 60 4.1 Rainfall Anomalies During El Niño and La Niña Years ...... 60 4.1.1 Sahel Regions ...... 60 4.1.2 Guinea Coast Region ...... 65 4.1.3 Sahel vs. Guinea Coast: Question of the Dipole ...... 67 4.2 Consistency of the ENSO Signal and Further Discussion ...... 67 4.3 Atmospheric Circulation Patterns and ENSO ...... 71

5 CONCLUDING REMARKS 92 5.1 Conclusions ...... 92 5.2 Implications ...... 95 5.3 Future Work ...... 95

References ...... 97 Biographical Sketch ...... 105

v LIST OF TABLES

1.1 Adopted from Nicholson (1981). Approximate climatic characteristics of four vegetation zones in West Africa; mean annual rainfall (mm), coefﬁcient of variation (CV%), length of the rainy season (months)...... 6

2.1 Adopted from Diatta and Fink (2014). Correlation coefficients between remote indices of climate variability and the West Sahel (WS), Central Sahel (CS), and Guinea Coast (GC). See text for further information on abbreviations. Correlation coefficients in bold are significant at the 95% significance level, while those in bold and with an asterisk are significant at the 99% level according to the F-test using the standard degree of freedom. Unless otherwise indicated, the time period 1921- 2009 was analyzed...... 34

3.1 A list of rainfall regions utilized in this study, the number of gauge stations in each region, and the coordinates and country of the geographic center point of each region. The color labels correspond to the color of the combined region in the bottom portion of Figure 3.1 that contains that individual region...... 53

3.2 Calculated ONI values before 1968 (1921-1967). Values of 0.5 or greater are highlighted in red and values of -0.5 or less are highlighted in blue. El Niño episodes are defined when there are 5 red ONI values in a row, while La Niña events are classified as years with five consecutive blue values. In cases when an episode spans multiple years, the first year (year of onset) is considered the ENSO year...... 57

3.3 Same as Table 3.2, except for the years after 1968 (1968-2012)...... 58

3.4 List of El Niño and La Niña Years from 1921 to 2012 that are used in this study. These years are determined based on the Oceanic Niño Index (ONI) values that are shown in Tables 3.2 and 3.3. Note that all analyses conducted in this study will consider an ENSO event as the 24-month period commencing in the July-September season of the year prior to its onset (JAS -1) and continuing until the April-June season of the year following the ﬁrst year of onset (AMJ +1). For this reason, El Niño and La Niña events spanning more than 2 years have been excluded to maintain consistency...... 59

4.1 Seasonal evolution of standard departures during El Niño years. Negative values are highlighted in red; positive values are highlighted in blue...... 80

4.2 Seasonal evolution of standard departures during La Niña years. Negative values are highlighted in red; positive values are highlighted in blue...... 81

vi LIST OF FIGURES

1.1 Adopted from Nicholson (2013). Annual mean precipitation over Northern Africa (in mm). The location of the Sahel is labeled...... 6

1.2 Adopted from Nicholson (2011). Vegetation zonation of West Africa (ﬁve zones): the Sahelo-Sahara is desert steppe with widely spaced grass clusters, the Sahel is semi-desert grassland, the Soudan is a savanna grassland, the Soudano-Guinean zone is woodland, and the Guinean zone is primarily forest...... 7

1.3 Adopted from CILSS (2016). The Sahel landscape is made up of scattered small trees, shrubs, and grasses. Vegetation cover varies from season-to-season and from year-to-year based on the amount of rainfall the area receives...... 7

1.4 Adopted from CILSS (2016). The Guinea Coast landscape is primarily made up of forest with a discontinuous upper canopy and dense lower canopy. The region receives the most annual rainfall of all of West Africa...... 8

1.5 Adopted from Nicholson (2011). Length of the rainy season (months) and month of maximum rainfall across Africa...... 8

1.6 Adopted from Nicholson (2011). Temperature (°C) and rainfall (mm) at typical stations over West Africa. The adjacent map gives the location of these stations in the context of the African continent as a whole...... 9

2.1 Adopted from Nicholson (2009). Classic picture of the ITCZ over Africa. In recent years, a revised view of the ITCZ has been developed...... 30

2.2 Adopted from Nicholson (2009). Revised view of the West African Monsoon. . . . . 30

2.3 Adopted from Nicholson et al. (1988). Schematic of boreal winter (top) and boreal summer (bottom) atmospheric circulation patterns (including winds and surface pressure) over the African continent. Arrows depict relative wind speed and direction, thin dashed circles depict various jet streams, the dotted line indicates the location of the ITCZ, and solid lines indicate surface pressure contours (in mb). . . . 31

2.4 Adopted from Thorncroft et al. (2011). Schematic showing the four main phases of the annual cycle of the West African monsoon. Each phase shows the location of the main tropical rain belt (indicated by the location of clouds and rainfall with peak values indicated by darker shades), the location of the SHL (yellow, orange, and red shading at the surface with darker orange/red indicating stronger intensity), At- lantic SSTs and mixed layer depth (indicated by colors red-green-blue in order of decreasing SSTs), moisture ﬂux convergence (solid contours) and divergence (dashed contours), and deep and shallow meridional circulations (blue and red lines with ar-

vii rows). Where there is uncertainty surrounding the extent to which the circulation return ﬂow penetrates the tropical rain belt, the lines have been dotted...... 32

2.5 Adopted from Flohn (1964). The annual migration of the tropical rain belt across the African continent and the associated patterns of rainfall seasonality as a function of latitude...... 33

2.6 Adopted from Nicholson (2008). Schematic showing the four most common rainfall anomaly patterns over West Africa. Light shading indicates below normal rainfall, while dark shading indicates above normal rainfall...... 33

2.7 Adopted from Mitchell (2016). Rainfall anomalies in cm/month from 1901 to the present. Notice the periodic shift between wet years and dry years throughout the twentieth century, as well as a prolonged drought during the 1970s and 80s...... 34

2.8 Adopted from Nicholson et al. (in press). Standard departures of Sahel and Guinea Coast rainfall. This time series is based on an updated dataset produced by Nichol- son (see Nicholson 1986; Nicholson et al. 2012) of gauge data that is considered the longest and most complete dataset of Sahel and Guinea Coast rainfall available. The green bars at the bottom of each plot indicate the total number of stations. Red indicators are placed on years where a factor is applied to adjust for the changes of variance when the number of stations is relatively small, following the methodology of Nicholson (1986)...... 35

2.9 Adopted from Nicholson et al. (in press). Seasonal standard departures of Sahel and Guinea Coast rainfall. A dashed line has been drawn at 1968 to mark the major change in the rainfall regime...... 36

2.10 Adopted from Nicholson et al. (in press). Correlations between rainfall in the Sahel (left) and Guinea Coast (right) and global SSTs for the years 1886 - 1967 for all four phases of the West African monsoon...... 37

2.11 Adopted from Nicholson et al. (in press). Correlations between rainfall in the Sahel (left) and Guinea Coast (right) and global SSTs for the years 1969 - 2013 for all four phases of the West African monsoon...... 38

2.12 Adopted from Lutgens and Tarbuck (2010). Simplified illustration of the seasaw pattern of atmospheric pressure between the eastern and western Pacific, called the Southern Oscillation. (a) During average years, high pressure over the eastern Pa- cific causes surface winds and warm equatorial waters to flow westward. The result is a pileup of warm water in the western Pacific, which promotes the lowering of pressure. (b) An El Niño event begins as surface pressure increases in the western Pacific and decreases in the eastern Pacific. This air pressure reversal weakens, or may even reverse the trade winds, and results in an eastward movement of the warm waters that had previously accumulated in the western Pacific...... 39

viii 2.13 Adopted from Vaughan (2014). Composites of equatorial Paciﬁc SST anomalies (°C) during OND of (top) strong El Niño years and (bottom) strong La Niña years. These composites demonstrate classic ENSO equatorial Paciﬁc SST signatures during El Niño and La Niña events...... 40

3.1 Individual rainfall regions across the entire African continent deﬁned in previous studies by Nicholson (top) and the combined regions utilized in this study (bottom). . 54

3.2 Adopted from the National Oceanic and Atmospheric Administration. Summary of the Niño regions in the equatorial Paciﬁc Ocean. SST anomalies within the Niño 3.4 region (5°N to 5°S; 120°W to 170°W) are generally used to determine the phase of ENSO...... 55

3.3 Adopted from Huang et al. 2015. Monthly and globally averaged ERSST v4 anomalies from 1854-2014. The data becomes much more consistent around 1880, with the most reliable data available after the 1940s, as evidenced by the shrinking uncertainty spread...... 55

3.4 Adopted from the National Oceanic and Atmospheric Administration. Graph of base periods used by CPC for calculating Niño 3.4 SST anomalies. The base period shifts every 5 years to accommodate for a long-term warming trend over time. . . . . 56

4.1 Seasonal evolution of standard departures during (top) El Niño years before and after 1968 and (bottom) La Niña years before and after 1968 for combined Region 9/13/18. El Niño years include 1923, 1925, 1930, 1945, 1951, 1953, 1963, 1965, 1968, 1969, 1972, 1976, 1982, 1991, 1994, 1997, 2002, 2004, and 2012. La Niña years include 1924, 1933, 1938, 1942, 1949, 1964, 1967, 1970, 1973, 1975, 1984, 1988, 1995, 2007, 2010, and 2011...... 73

4.2 Same as Figure 4.1, but for combined Region 14/19...... 74

4.3 Same as Figure 4.1, but for combined Region 15/20...... 75

4.4 Same as Figure 4.1, but for combined Region 16/17/21...... 76

4.5 Same as Figure 4.1, but for Region 22...... 77

4.6 Same as Figure 4.1, but for Region 23...... 78

4.7 Same as Figure 4.1, but for combined Region 24/28/29...... 79

4.8 Percent of JAS (-1) standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign

ix for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012...... 82

4.9 Percent of OND (-1) standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012...... 83

4.10 Percent of AMJ standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012...... 84

4.11 Percent of JAS standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012...... 85

4.12 Percent of OND standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012...... 86

4.13 Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the JAS (-1) season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis...... 87

4.14 Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the OND (-1) season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis...... 88

4.15 Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the AMJ season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis...... 89

x 4.16 Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the JAS season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis...... 90

4.17 Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the OND season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis...... 91

xi ABSTRACT

The Sahel and Guinea Coast regions of Africa have long been the subject of studies on interannual and intraseasonal rainfall variability. The unique geography, monsoon circulation regime, and a variety of climatic teleconnections produce large variations in year-to-year rainfall across the region. These large fluctuations in rainfall can have devastating effects on the inhabitants of West Africa, who rely on the rainfall for both agriculture and human consumption. Thus, a better understanding of the nature of rainfall variability in the area is warranted. The El Niño/Southern Oscillation (ENSO), one of the most studied climate phenomena, is known to have far-reaching impacts on weather across the globe. This study provides one of the most comprehensive and complete analyses of the relationship between ENSO and rainfall across the Sahel and Guinea Coast to date. Several previous studies have found little connection between Sahel rainfall and ENSO phase, while others have suggested that ENSO can result in changes within the monsoon circulation and cause a reduction in Sahel rainfall during El Niño years. By utilizing the largest and longest dataset of rainfall gauge data available, this study provides an analysis of rainfall anomalies experienced during El Niño and La Niña years from 1921-2012 in the context of a major shift in the rainfall regime that occurred around the year 1968. This research finds that before 1968, rainfall during the peak Sahel rainy season in El Niño years was below normal, but above normal in the Guinea Coast. The same is observed after 1968, but the anomalies are of stronger magnitude than before 1968, suggesting an increased ENSO- Sahel rainfall teleconnection after 1968. Similar intensifications of the El Niño signal are observed in other seasons as well. In general, opposite rainfall anomalies were observed during La Niña years when compared to El Niño years. An increase in La Niña influence in more recent years is also detected. An analysis of the consistency of the ENSO signal suggests that the ENSO rainfall response is most consistent in areas of the Sahel during the JAS (-1), OND (-1), JAS, and OND seasons. Evidence also suggests that there was a weakening of the Sahel/Guinea Coast dipole after 1968. Finally, an analysis of upper air circulations shows few differences in zonal winds during El Niño and La Niña years versus non-ENSO years, suggesting the relationship between ENSO

xii and Sahel rainfall may be fairly weak. There are some subtle differences seen, however, when comparing years before 1968 to years afterwards that were consistent with the observed rainfall anomalies in certain seasons. This study concludes that the rainfall response to El Niño and La Niña events in the Sahel and Guinea Coast as a whole is relatively inconsistent, but there was some meaningful connection found between ENSO and rainfall in the Sahel during certain seasons outlined above. This relationship intensiﬁed after the 1968 rainfall regime change, consistent with ﬁndings from previous studies.

xiii CHAPTER 1

INTRODUCTION

A geographical and climatological overview of this project’s area of study, West Africa (including the Sahel and Guinea Coast regions), is discussed in this chapter. The geography and overall climatology of the Sahel and Guinea Coast are intricately linked to the regions’ rainfall regimes, making it important to highlight some general characteristics of the regions in order to build a foundation for the basis of this project. Chapter 1 concludes by providing the motivation behind the present study.

1.1 The West African Sahel and the Guinea Coast

West Africa has long been the subject of studies on interannual and intraseasonal rainfall variability. This is owed, in large part, to the unstable nature of precipitation patterns from year to year and the subsequent consequences that the ﬁckle rainfall regime can have on the social and economic aspects of life in the region (Okonkwo and Demoz 2014).

1.1.1 Geographical Overview

Historically, “Sahel” has been used as a general term referring to the large semi-arid region extending across the African continent covering an east-west extent of about 5000km, from the southern boundary of the Sahara Desert to about 10°N latitude (Nicholson 2013). However, the name “Sahel” more accurately refers to the smaller region between about 14°N to 18°N latitude (Nicholson 2013) and 15°W to 30°E longitude (Nicholson 2008) including the countries of Mau- ritania, Senegal, Mali, Niger, Chad, the Sudan, and northern portions of Burkina Faso and Nigeria (see Figure 1.1). Most studies of Sahel climatology omit areas to the east of 30°E including eastern Sudan and the country of Ethiopia due to complex topography’s inﬂuence on large scale weather patterns in this region (Nicholson 2013).

1 Despite considerable variation among different authors’ definitions and spatial extents at- tributed to the region dubbed the “Sahel” (e.g. Sanogo et al. 2015, Okonkwo and Demoz 2014, Nicholson 2008, CILSS 2016), the widely accepted practice is to define the region (and others in West Africa) based on long-term rainfall averages and geographical features. Most sources cate- gorize the entire West African region into five geobotanical zones, shown in Figure 1.2 (Nicholson 2011). From north to south these regions consist of: the Sahelo-Sahara zone, a desert steppe with widely spaced grass clusters; the semi-arid grassland known as the Sahel; the Soudan zone consisting of savannah grasslands; the woodland Soudano-Guinean zone marking a transition between the savanna and forest, and finally the Guinean zone made up primarily of forest (Nicholson 2011). Various definitions of these individual zones are defined by different authors (for example, CILSS (2016) defines five slightly different bioclimatic zones) but the general conscience tends to be similar to that of Figure 1.2. These regions serve to illustrate the close relationship between each region’s climate and vegetation cover, and a gradual transition of geography based on latitude. Despite a strong north-south gradient of climate and vegetation conditions throughout West Africa as evidenced by Figures 1.1 and 1.2, there is general uniformity in the east-west direction. For the purposes of this study, “Sahel” will roughly coincide with the region outlined in Figure 1.1 and the Sahel zone outlined in Figure 1.2, while “Guinea Coast" will coincide with the Guinean Zone of Figure 1.2. More detail about other specific rainfall regions in West Africa will be described in the next chapter. The Sahel region outlined above is characterized by the transition between the arid Sahara Desert to the north and the tropical Guinea Coast to the south. The region is considered semi-arid, with a mainly flat topography consisting of grasses, shrubs, and trees (Nicholson 2013) as shown in Figure 1.3 (CILSS 2016). Annual grass fires are common in regions where grasses are prevalent, but the region is also home to to numerous small wetland areas as well (CILSS 2016). The Sahel supports a mostly rural and agricultural society, but it is also home to major cities such as Dakar, Niamey, Bamako, and Khartoum. The diversity in geography and land use across West Africa creates a wide range of lifestyles within the region, from urban dwellers to farmers. Recently, a focus has been placed on studying land use changes within the Sahel. These noticeable changes largely result from factors such as

2 increasing human population, institutional factors and regional policies, and processes of glob- alization. However, natural factors have contributed to changes in the Sahel landscape as well. On decadal time scales, the presence of droughts (to be discussed in detail in later chapters) has produced soils that are less productive, bodies of water that are a fraction of their original size, stressed vegetation, and increased erosion (CILSS 2016). The Guinean Zone of Figure 1.2 (henceforth “Guinea Coast") consists of a drastically different geographical landscape to the Sahel. This region receives much more rainfall per year, making way for woodland and forest landscapes that can be fairly dense in some places (CILSS 2016). Despite some decline in forested area over the years due to human activity, the region is still home to the richest ﬂora in West Africa. The majority of the forest landscape consists of a discontinuous upper canopy towering over a lower, dense forest canopy. Figure 1.4 (CILSS 2016) shows vegetation similar to that of the Guinea Coast. The Soudan and Soudano Guinean Zones (Figure 1.2) serve primarily as transition regions between the Sahel and Guinea Coast. The landscapes of these regions range from savannah to forrest, depending on latitude and physical location. Rainfall within these transition regions will also be analyzed within the scope of this project.

1.1.2 Climatological Overview

As previously mentioned, the geographical diversity of the geobotanical zones of West Africa outlined in Figure 1.2 are heavily linked to rainfall variability across the region. The Sa- hara desert is produced from dominance of a high-pressure regime, subsidence, and an absence of rain-bearing disturbances, all of which also result in the general semi-arid nature of sub-Saharan Africa. The further away from the Sahara you get, the less these conditions dominate the regime. Therefore, as a general rule, the further south one travels in West Africa, the longer the rainy season. This is shown in Table 1.1 (Nicholson 1981), which provides approximate rainfall amounts, coefﬁcient of variation, and length of the rainy season for each zone. Note that the exception to this rule is an area of the Guinea Coast that remains relatively dry, likely due to the inﬂuence of high pressure, cool sea-surface temperatures nearby, and parallel winds (Nicholson 2011). Fig- ure 1.5 compliments Table 1.1 by visually showing the length of the rainy season and month of

3 maximum rainfall across across the African continent. One should be careful not to assign precise limits to the zones in Figure 1.2. Their boundaries should be thought of as ﬂuid transitions rather than sharp contrasts. A sense of the seasonal progression of rainfall amounts and temperatures at several stations within these zones is shown in Figure 1.6 (Nicholson 2011). Rainfall throughout West Africa, and in particular the Sahel, is generally produced by squall lines and mesoscale convective systems which propagate across the east-west extent of the region (Nicholson 2011). These systems are generally located between the cores of the African Easterly Jet (AEJ) and the Tropical Easterly Jet (TEJ), both of which are discussed in Chapter 2.1.2. The occurrence of these systems is also heavily inﬂuenced by the West African Monsoon (Thorncroft et al. 2011). As a result, it is rare for areas in West Africa to experience rainfall events outside the normal rainy season. In the Sahel, the vast majority of rainfall is limited to the boreal summer months of June, July, August, and September, with peak rainfall occurring in August. The Sahel generally receives between 100 and 600 mm of rainfall annually, with the lowest rainfall totals being in the north near the Sahara and highest totals to the south (Nicholson 2013). However, there is much variability in timing, amount of rainfall, and prevailing circulation systems from year to year (CILSS 2016; Ali and Lebel 2009; Nicholson and Grist 2003), with several studies indicating contrasts throughout the Sahel region (Lebel and Ali 2009, Nicholson and Palao 1993, and others). According to Nesbitt et al. 2006, as few as 2% of rainfall features during the rainy season produce 60-90% of annual rainfall over the Sahel. These intense features, known as organized convective systems, include at least one cloud cluster with temperatures of 213K or colder and a size of at least 5000 km2, and last at least 3 hours with mean speed of at least 10m/s (Mathon et al. 2002). Although these organized convective systems comprise only 12% of mesoscale convective systems over West Africa, they can produce 80% of cloud cover and upwards of 90% of rainfall in some areas of the Sahel. The rest of Sahel rainfall is mainly produced as a result of localized thunderstorms, with more storms occurring in the south versus the north, but still concentrated in the rainy season outlined above (Nicholson 2011). During transition and winter months, it is not unheard of for areas of West Africa to receive occasional precipitation from diagonal cloud bands known as tropical plumes (Knippertz et al. 2003). Although they rarely produce more than 25mm of rainfall, these tropical plumes can greatly enhance moisture content in the northwestern

4 extremes of the Sahel, and can even occur during winter months in parts of the Guinea Coast (Knippertz and Fink 2008). The yearly cycle of the West African monsoon and its inﬂuence on rainfall across West Africa will be further discussed in Chapter 2.

1.2 Motivation

It should now be evident that both the geography and climate of West Africa are extremely complex systems, posing unique challenges to meteorologists and climatologists. In a region where water is an important resource for both human consumption and agriculture, studying variability in the rainfall and knowing the causes and implications that certain meteorological phenomenon have on the amount of rainfall received becomes paramount to its inhabitants. Drought has had significant impacts on the lives of those who live in the Sahel, particularly during the latter portion of the twentieth century. Although it is well documented that the area naturally experiences interannual variability in its rainfall amounts, there was a noticeable change in the rainfall regime of the Sahel around the year 1968 that persisted for quite some time (and some authors would suggest continue to persist today). Long-term droughts place significant strain on a society that heavily relies on agriculture, and numerous studies have analyzed the societal impacts that drought caused in the Sahel during the 1970s and 1980s. As shown in these studies and beyond, the effects of long-term climatic changes, particularly fairly sudden changes, can be far reaching. It is therefore the job of meteorologists and climatologists to attempt to piece together the puzzle and figure out how to more accurately predict when and where these dry spells will occur. Certain climate features, such as the El Niño/Southern Oscillation (ENSO), have been documented to have impacts on weather patterns across the globe. Numerous studies (Nicholson and Entekhabi 1986; Nicholson and Kim 1997; Nicholson et al. 2000; Nicholson and Selato 2000; Janicot et a. 2001; Rowell 2001; Losada et al. 2012; Rodríguez-Fonseca et al. 2013; Parhi et al. 2016; Nicholson et. al in press) have analyzed both El Niño and La Niña’s role in African rainfall variability. ENSO, like severe droughts, can have devastating impacts on a society by al- tering the state of the atmosphere. As just one example of how ENSO can affect society indirectly,

5 Okonokwo and Demoz (2014) examined how El Niño events can affect cereal production in the Sahel. This demonstrates the far-reaching consequences that ENSO can have on not only weather patterns but also our daily lives. The goal of this project is to build upon the existing literature regarding rainfall variability in the Sahel and Guinea Coast. Speciﬁcally, this project will analyze the impacts of El Niño and La Niña on rainfall both before and after the aforementioned 1968 rainfall regime change. By reconstructing and documenting the inﬂuence of El Niño and La Niña during both rainfall regimes (wet and dry) of the twentieth century, it will provide a historical context for evaluating the nature of ENSO’s impacts on rainfall across the region in the future.

Figure 1.1: Adopted from Nicholson (2013). Annual mean precipitation over Northern Africa (in mm). The location of the Sahel is labeled.

Table 1.1: Adopted from Nicholson (1981). Approximate climatic characteristics of four vegetation zones in West Africa; mean annual rainfall (mm), coefﬁcient of variation (CV%), length of the rainy season (months).

Vegetation Zone Rainfall (mm) CV (%) Season (months) Sahelo-Sahara 50-100 50 1-2 Sahel 100-400 30-50 2-3 Soudan 400-1200 20-30 3-5 Soudano-Guinean 1200-1600 15-20 5-8

6 Figure 1.2: Adopted from Nicholson (2011). Vegetation zonation of West Africa (ﬁve zones): the Sahelo-Sahara is desert steppe with widely spaced grass clusters, the Sahel is semi-desert grassland, the Soudan is a savanna grassland, the Soudano-Guinean zone is woodland, and the Guinean zone is primarily forest.

Figure 1.3: Adopted from CILSS (2016). The Sahel landscape is made up of scattered small trees, shrubs, and grasses. Vegetation cover varies from season-to-season and from year-to-year based on the amount of rainfall the area receives.

7 Figure 1.4: Adopted from CILSS (2016). The Guinea Coast landscape is primarily made up of forest with a discontinuous upper canopy and dense lower canopy. The region receives the most annual rainfall of all of West Africa.

Figure 1.5: Adopted from Nicholson (2011). Length of the rainy season (months) and month of maximum rainfall across Africa.

8 Figure 1.6: Adopted from Nicholson (2011). Temperature (°C) and rainfall (mm) at typical stations over West Africa. The adjacent map gives the location of these stations in the context of the African continent as a whole.

9 CHAPTER 2

PRIOR RESEARCH

2.1 Characteristics of Sahel Rainfall 2.1.1 The West African Monsoon

Rainfall in the Sahel and throughout West Africa is strongly linked to the West African monsoon circulation. The classical view of the monsoon, as shown in Figure 2.1, depicts an In- tertropical Convergence Zone (ITCZ) that is defined as the convergence of northeasterly Harmattan winds originating from the Sahara in the north and the southwest monsoon flow originating from over the Atlantic. In this classical view, the ITCZ moves northward during boreal summer and southward during austral summer bringing rain with it as a result of local thermal instability and low level convergence. There are several drawbacks to this classical view of the West African Monsoon. First, the classical view is based on the Hadley Cell, which was not intended to model individual regions but instead was developed to describe the overall global mean state of atmospheric circulations. Additionally, it is well understood that the absence of trade winds make the Hadley Cell concept a poor one over continents, while it is mostly valid over oceans. Another drawback of the classical view of the West African monsoon is that it was originally developed when it was believed that tropical rainfall was of local origin, but studies during the latter part of the twentieth century have suggested that tropical rainfall is instead initiated by large-scale disturbances as a result of wave motion. Finally, there has been much scientific debate over the meaning of the term "Intertropical Convergence Zone (ITCZ)," with various authors assigning a variety of definitions to the term. Nicholson 2008 points out that the term ITCZ is commonly, though mistakenly, used to refer to the zone of maximum rainfall. However, these are two separate phenomena. Nicholson 2008 argues that the ITCZ is a surface feature that should be defined as the zone of wind convergence, not the zone of maximum rainfall since the locations of surface convergence and the location of maximum

10 rainfall do not always coincide. Therefore, in this paper, the term "tropical rain belt" will be used to describe the zone of maximum rainfall. In order to reconcile these shortcomings, a revised view of the West African monsoon has been developed. The main adjustments to the classical view include a decreased reliance on the ITCZ and the inclusion of additional elements such as jet streams, upper level circulations, African Easterly Waves, the Saharan Heat Low, and the influence of mesoscale convective systems (Nichol- son 2013). Despite these adjustments to the classical view, several overall characteristics of the monsoon system remain consistent. In short, thermodynamic contrasts between the Sahara and the Atlantic ocean cause shifts in winds. A cyclonic flow, called the Saharan Heat Low, develops over the Western Sahara during summer months. The cyclonic pattern of convergence between northeasterly Harmattan winds with southwesterly monsoon winds transporting moisture from the cold tongue of the Atlantic over the continent results in a circulation pattern with great influence on the modulation of rainfall during boreal summer. In the following sections, atmospheric circulations and rainfall characteristics associated with the West African Monsoon are discussed, including the contrasting dipole pattern between the Sahel and Guinea Coast. This is followed by an overview of rainfall throughout the regions during the twentieth century, recent changes in the rainfall regime, and various global teleconnections to rainfall variability in West Africa.

2.1.2 Atmospheric Circulations and Jets

In order to fully understand the nature of rainfall patterns throughout the Sahel and Guinea Coast, it is important to first consider the wind regimes and jet streams that dominate over northern Africa. These circulations and jets have large influences on the intensity, amount, and spatial distribution of rainfall and the migration of the tropical rain belt. The major mid and upper level jet streams with a large influence on the dynamics of West African rainfall are the African Easterly Jet (AEJ) and the Tropical Easterly Jet (TEJ). Recently, the importance of low-level jets such as the African Westerly Jet (AWJ), West African Westerly Jet (WAWJ), the Bodélé Jet, and a nocturnal low-level jet has also been highlighted.

11 Due to it’s influence in producing African Easterly Waves, the AEJ is at the forefront of studies regarding jet streams in northern Africa. The AEJ is defined as a thermally induced jet produced by the strong temperature contrasts between the the Sahara and Atlantic Ocean, fueled by moisture from the south and dry convection in the north (Nicholson and Grist 2003). Although usually studied as a single feature, the AEJ consist of two sectors: an eastern core that tends to be prominent in years of low rainfall and a western core that is prominent during wet years (Nicholson 2013). Nicholson and Grist (2003) describe the AEJ as having a core located around 650-700mb that is typically strongest in May and June. They found mean core wind speeds are on the order of around 10-12 meters per second, with the highest wind speeds in the western sector and occurring just before the maximum of the Sahel rainy season. In general, slower wind speeds are indicative of a somewhat wetter monsoon season, while faster speeds are associated with a drier rainy season (Nicholson and Grist 2001). Additionally, the location of mesoscale convective systems is related to latitudinal shifts of the AEJ. Intraseasonal shifts in the jet core occur, causing a split and the formation of a secondary jet to the north of the MCS, displacing the original jet southward (Wang and Elsberry 2010). The TEJ is an upper level jet located between 100 and 200mb in height. The TEJ, like the AEJ, is also a thermally induced jet that forms as a result of the strong temperature contrast between the Himalayan Plateau and the Indian Ocean. Maintained by tropical divergent circulations associated with the Hadley (north-south) and Walker (east-west) circulations, the TEJ is strongest in boreal summer when speeds average 18 meters per second over the eastern Sahel, but it is also strong during January through March when its core is located in the Southern Hemisphere (Nicholson and Grist 2003). Unlike the AEJ, the TEJ does not vary much latitudinally, but instead fluctuates in speed and east-west extent (Grist and Nicholson 2001; Nicholson and Grist 2001; Nicholson 2008; Nicholson 2013). Numerous studies by Nicholson and Grist have also shown that there is a consistent relationship between the strength of the TEJ and rainfall over West Africa. Upper-level divergence resulting from strong meridional components of the TEJ seems to be the primary link between the TEJ and rainfall in the Sahel, which is consistent with the idea that a strong TEJ is linked to a more intense tropical rain belt (Nicholson 2008, 2009). In other words, there is a link between a stronger (weaker) TEJ and wetter (drier) conditions in the Sahel. Nichol-

12 son and Grist (2001, 2003) and Grist and Nicholson (2001) have shown instances when the TEJ reached mean speeds of around 30 m/s during some wet years and extended across the African continent. On the other hand, during some dry years, the TEJ has reduced to speeds of less than 10m/s and cover only eastern portions of the region. They have also suggested a link between drier conditions in the Sahel during the latter part of the twentieth century and a diminished TEJ speed and spatial extent (see section 2.1.6 Rainfall Regime Change for more). In addition to the AEJ and TEJ, there are also various low level jets that have been documented to have a presence in the Sahelian circulation regime and a a relationship with West African rainfall. First described by Washington and Todd (2005), the Bodélé Low-Level Jet, is present throughout the year with the exception of August. With a core near 18°N at 925mb, it is strongest in January and weakest in July with a mean core speed of 8m/s – just enough to generate tremendous amounts of dust in the Bodélé Depression north of Lake Chad. It has a strong diurnal cycle with strongest winds in the evening and weakest during the day (Washington and Todd 2006). Orography plays an important role in the jet’s formation, as evidenced by its absence from relatively ﬂat areas further east. A separate entity from the monsoon circulation itself, the African Westerly Jet (AWJ) is another low-level jet that only appears during wet years in the Sahel region with speeds on the order of 10m/s during late summer. First identiﬁed by Grist and Nicholson (2001), the AWJ extends well into the mid-troposphere during its peak, but nearly disappears completely during dry years. It is believed to form from inertial instability which develops as a result of a pressure gradient between 20°S and 20°N during the height of the Sahel rainy season (Nicholson 2013). Not to be confused with the AWJ, the West African Westerly Jet (WAWJ) is a low-level marine jet that is best developed near the location where trade winds converge, around 10°N. Al- though it does not extend into the Sahel, there is a strong correlation between the strength of the WAWJ and rainfall in the Western Sahel, suggesting that could be an important factor in transporting moisture from the Atlantic to the continent even when the monsoon is weak (Grodsky et al. 2003). It also introduces strong relative vorticity gradients into the region, working to stabilize the regional vorticity balance (Nicholson 2013).

13 Finally, the nocturnal low-level jet (NLLJ) lies just 200-400m above the surface. It typically reaches its maximum during the onset of the monsoon and dissipates when the tropical rain belt migrates southward late in the monsoon season (Bain et al. 2010; Peyrille and Lafore 2007). By bringing moisture into the monsoon, particularly around its onset, the NLLJ helps sustain deep convection and is credited with forming the stratus cloud decks that are common south of the Sahel extending into the Sudan and Guinea Coast zones (see Figure 1.2) (Schrage and Fink 2012). In addition to the jets that have an influence on rainfall variability in the Sahel, there are two other noteworthy features that dominate the atmospheric circulations of the region: The Sharan Heat Low (SHL) and the Saharan Air Layer (SAL). The SHL is characterized as a shallow area of low surface pressure and generally high temperatures. Usually located below 700mb, the SHL originates in southern Sudan and migrates northwestward, arriving in the western Sahara during boreal summer about five days before the onset of the West African monsoon (Lavayesse et al. 2009). Chauvin et al. (2010) suggests that the SHL connects the West African monsoon with the midlatitudes by consisting of two phases: west and east. The west phase, proceeded by large-scale functions in the midlatitudes such as Rossby waves, occurs when maximum surface temperatures associated with the SHL occur over Morocco, Western Sahara, and Mauritania coasts with a temperature minimum between Libya and Sicily. The east phase of the SHL has the opposite structure. Parker et al (2005) concludes that changes in the intensity of the SHL can affect intensity of the monsoon circulation further south. In addition, it has been shown that the SHL is influenced the amount of dust in the atmosphere, rainfall amounts, and inflow of cold, stable air from the Atlantic (Lavayesse et al. 2011). Dust has been shown to have important impacts on the atmospheric circulation regime of the Sahel and Guinea Coast regions, with impacts on the strength of the SHL, the AEJ, and the development of African easterly waves. The major dust feature affecting dynamics in the region is called the Saharan Air Layer (SAL). The SAL is a well mixed layer of mineral dust originating from the Sahel and Sahara located between 500mb and 900mb in height. The SAL’s makeup is generally uniform in the sense that it consists of warm temperatures and low humidity. Its east- west extend can vary greatly, but it can cover an area as large as 5000km. As discussed in a recent review (Nicholson 2013), the SAL’s radiative effect, thermal character, and dryness can

14 impact many different meteorological developments in West Africa, including impacts on the AEJ, African Easterly Waves, and therefore the development of Atlantic tropical cyclones. Figure 2.3 (Nicholson et al. 1988) shows a summary schematic of circulation patterns across the African continent in both the winter and at the height of the monsoon in July/August. The various features discussed in this section are shown, including the TEJ, AEJ, SHL, and the migration of the ITCZ (tropical rain belt).

2.1.3 Seasonal Rainfall Cycle and Intraseasonal Variability

The annual precipitation received across Northern Africa (in mm) is shown in Figure 1.1. However, to best describe the seasonal cycle of this rainfall and its temporal variability, Thorncroft et al. (2011) deﬁned four distinct phases of the West African monsoon based on the location of peak rainfall: the oceanic, coastal, transitional, and Sahelaian phase. The conclusions of Thorncroft et al. (2011) are summarized in Figure 2.4 (Thorncroft et al. 2011) and described in detail in this section. In addition, Figure 2.5 shows the annual migration of the tropical rain belt over the African continent (Flohn 1964). During the oceanic phase, which lasts from November to mid-April, the main rain band (i.e. “tropical rain belt") covers a broad area near the Equator, where SSTs exceed 28°C. As a result, West Africa receives very little rainfall during this phase, as the rainfall maximum is well to the south. When equatorial sea-surface temperatures (SSTs) begin to cool as a result of poleward migration of column moisture ﬂux convergence and increased importance of the SHL, a cold tongue develops in the equatorial Atlantic and the tropical rain belt begins its northward migration. Once equatorial SSTs drop below 28°C, peak rainfall migrates to the coastal region around 4°N where warmer SSTs remain (although peak rainfall lies over the ocean but near the coast). This occurs between April and May, and marks the beginning of the coastal phase, lasting until the end of June. It is during this phase of the monsoon that the Guinea Coast receives the majority of its yearly rainfall. This is also the wettest phase of the monsoon, despite most studies focusing on the later Sahelian phase (Nguyen et al. 2011).

15 Near the end of June until mid-July, coastal moisture flux convergence remains strong but coastal peak rainfall noticeably decreases. This is consistent with the observed decrease in SSTs in the coastal region and stabilization by advection of warm, dry air as a result of the return flow of the SHL. Finally, there is an abrupt shift in the latitude of maximum rainfall that occurs in mid-July and lasts until September (this can be seen in Figure 2.5). During this time, the latitude of maximum rainfall shifts from about 5°N to 10°N, and is sometimes referred to as the “monsoon" jump. This final phase of the monsoon is referred to as the continental phase (Lebel et al. 2003) or Sahe- lian phase (Thorncroft et al. 2011), because this is when the Sahel receives the vast majority of its yearly precipitation. As discussed in Nicholson (2013), there is much debate among researchers on the exact cause of this this abrupt change. Some possible explanations point to the influence of westward-propagating disturbances (Sultan and Janicot 2000), intensification and shift of the SHL (Sijikumar et al. 2006; Ramel et al. 2006), a northward shift in the AEJ and associated horizon- tal and vertical shear zones (Gu and Adler 2004), the interaction between the Atlantic equatorial cold tongue and the monsoon (Okumura and Xie 2004), the interactions among convection, AEJ dynamics, and local topography (Sultan et al. 2003), and the release of potential instability with subsequent inertial instability causing the sudden shift to 10°N (Hagos and Cook 2007). Nichol- son (2013) points out that all of these possible explanations are consequences of a northward shift temperature and pressure gradients in West Africa. In the Sahel, a large focus has been on studying variability in rainfall on both intraseasonal (10 to 90 days) and interannual (year-to-year) time scales. One of the primary objectives of the present study is to examine the intraseasonal and interannual variability as it relates to the El Niño/Southern Oscillation (to be discussed in the following sections). However, it is important to keep in mind that there are many other factors determining rainfall variability across the region as well. On the intraseasonal time scale, the Madden-Julian Oscillation (MJO) has been found to have an impact on rainfall across West Africa, although recent research (Maloney and Shaman 2008) has suggested that the influence in the Sahel proper may be small, with the greatest impacts located further south. Originating in the Pacific, the MJO consists of an eastward-propagating

16 Kelvin wave and westward-propagating Rossby wave. When these waves meet over Africa during boreal summer, they can enhance convection and inﬂuence the atmospheric circulation patterns, African easterly waves, and moisture transport (Janicot et al. 2009; 2010). In addition to the MJO, intraseasonal variability related to the SHL on time scales of 10 to 25 days has also been linked to intraseasonal variability (Roehring et al. 2011). Interannual variability of rainfall in the Sahel and Guinea Coast is a large focus of the present study, and is discussed at length in sections 2.1.4 through 2.1.7.

2.1.4 Spatial Variability and the Dipole Pattern

As discussed in the previous section, West Africa receives two distinct rainy seasons: the ﬁrst in boreal spring (during the coastal phase of the monsoon) near the Guinea Coast, and the second in boreal summer (Sahelian phase of the monsoon) further inland over the Sahel (Nguyen et al. 2011). However, there are also spatial variations on interannual time scales that differentiate between abnormally wet and dry years. There are two main patterns that describe most of the variability of rainfall in West Africa (Nicholson and Grist 2001; Nicholson 1981, 1986, Nicholson and Palao 1993, among others). The most common pattern is the so-called “dipole," referring to rainfall anomalies of opposite sign between the Guinea Coast versus the Sahel. As evidenced in the bottom row of Figure 2.6 (Nicholson 2008), the node of the dipole lies near 10°N, marking an abrupt transition where nearly all station anomalies north or south of this latitude are the same sign. This pattern is the result of a change in latitudinal displacement of the tropical rain belt (Nicholson and Grist 2001). The second pattern of variability variability is characterized by anomalies of the same sign across both the Sahel and Guinea Coast (see the top row of Figure 2.6), usually as a result of changes in intensity of the tropical rain belt (Nicholson and Grist 2001; Nicholson 2008). As mentioned, the latitudinal location of the tropical rain belt accounts for the difference among wet and dry years in the dipole pattern, as shown in Figure 2.6. During years when the rain belt is located further north (around 12°N to 13°N), the pattern is similar to the 1950 case shown in the ﬁgure. During years when the rain belt is located further souther (from around 7°N to 8°N), the rainfall dipole takes the form of the 1984 example in Figure 2.6. Although this

17 accounts for the differences seen between the two dipole scenarios, it does not account for the non- dipole cases. Nicholson (2008) found that the non-dipole scenarios can instead be explained by uniform changes in intensity of the rain belt. For example, during the “wet" year of 1955, rainfall exceeded 200mm/month. However, in the “dry" year of 1983 rainfall only reached 100mm/month. Nicholson (2008) also found a large decrease in the extent of the tropical rain belt during 1983 versus 1955, indicating that its latitudinal extent can also play a role. Recent studies (Losada et al. 2012) have suggested that the dipole pattern appears to have disappeared in recent decades, but was no doubt notable during the twentieth century.

2.1.5 Rainfall Trends in the Twentieth Century

Rainfall in the Sahel during the 1900s has long been the subject of many researchers at- tempting to understand the long-term variations of rainfall in the region. The twentieth century was marked by periods of both excess rainfall and drought, as shown by the time series of rainfall anomalies in the Sahel (defined here as 20°N to 10 °N, 20°W to 10°E) in Figure 2.7 (Mitchell 2016). This figure shows the average number of centimeters per month that rainfall was either either above or below normal (average) for each year from 1901 to 2016. Early studies conducted on Sahel rainfall variability analyzed the drought periods in the 1910s and again in the 1940s, both of which can be seen in Figure 2.7. During the early 1970s, when rainfall once again fell below normal, the idea of a thirty-year drought cycle came to light. However, this theory was later dispelled by the persistence of the drought into at least the 1980s and 1990s. Additionally, the severity of the drought of the 1970s and 1980s highlighted the fact that the earlier twentieth century droughts were relatively weak, particularly considering that rainfall in the 1980s dropped to around 60% of the long term mean (Nicholson 2005). In stark contrast to the periods of drought in the 1910s, 1940s, and finally the 1970s and 1980s, there also were relatively wet periods throughout the twentieth century as well. Perhaps the wettest of these was in the 1950s and early 1960s, when positive rainfall anomalies persisted for nearly a twenty year period. Other wet periods evidenced by Figure 2.7 include the 1920s to late 1930s, as well as prior to 1910.

18 Despite overall agreement about the major rainfall peaks and declines during the 1900s, there has been some controversy regarding the rainfall regime since the 1990s to the present day. Some studies have suggested that since the intense drought during the 1970s and 1980s Sahel rainfall has largely recovered back to near-normal amounts (Mahè and Paturel 2009; Ali and Lebel 2009; Lebel and Ali 2009; Sanogo et al 2015). Others (Nicholson et al. 2000; L’Hôte et al. 2002; L’Hôte et al. 2003) have suggested that while this might be the case in most recent years, drought conditions persisted throughout the 1990s. Still more studies have suggested some combination of the two, proposing that the drought persisted despite a few very wet years during an otherwise dry period (Nicholson 2005; Lebel and Ali 2009; Diatta and Fink 2014). Finally, some (Ozer et al. 2003) have suggested that the Sahel is actually entering a more humid period, citing recent ﬂood events.

2.1.6 The 1968 Rainfall Regime Change

In an attempt to finally provide a definitive answer to the question of the rainfall regime recovery, most recent work by Nicholson et al. (in press) utilized the longest and most comprehensive dataset of 742 gauge stations throughout the Sahel and Guinea Coast over a time span of 171 years. Their analysis produced three very important conclusions that serve as the basic foundation of this project: (1) a full recovery from the extended drought has not occurred, (2) a major rainfall regime change occurred around the year 1968, and (3) the influence of large-scale teleconnections to Sahel and Guinea Coast rainfall have noticeably changed since the 1968 rainfall regime change. Based on the complete time series of rainfall standard departures shown in Figure 2.8 (to be explained in further detail in Chapter 3), Nicholson et al. (in press) found regime shifts that were statistically significant at the 5% level at 1876, 1897, 1927, 1950, 1968, 1982, and 1994. Further- more, they found that the 1968 regime change was by far the the largest and most abrupt. Figure 2.8 demonstrates this by indicating that there were very few years with rainfall above the long-term mean after 1968 in both the Sahel and Guinea Coast. Figure 2.9 shows the seasonal anomalies, broken up by April-May, June, July-August-September, and October-November to roughly coincide with the phases of the West African monsoon outlined in Figure 2.4 by Thorncroft et al. (2011). The seasonal anomalies also show an abrupt change at the year 1968, indicated by a

19 dashed line in the figure. The change in the rainfall regime in the Sahel was notably greatest during the July-August-September period, while changes in June and October-November are greatest for the Guinea Coast. Nicholson et al. (in press) also noted important east-west contrasts within the Sahel. Prior to 1968, they found that rainfall anomalies in the west were generally wetter, while anomalies in the east were generally drier than average. After 1968, however, the pattern reversed, leading them to conclude that the persistence of drought after 1968 was linked to a reduction in the east-west gradient of rainfall. In regards to the Sahel/Guinea Coast dipole, they found that the dipole pattern was less frequent after 1968, and that after 1968 the monsoon was weaker and associated with a tropical rain belt that was located further to the south. Overall, the results of Nicholson et al. (in press) provide important conclusions about the nature of the long-discussed drought across West Africa during the latter portion of the twentieth century. Their findings can shed light on seasonal prediction as well as the future rainfall characteristics in the Sahel and Guinea Coast. By establishing concrete evidence of the previously suspected 1968 change in the rainfall regime, their results are groundbreaking in that they have opened up many new questions regarding the overall conditions that could have caused such an abrupt and long-lasting change. One such category of questions undoubtedly surrounds the influence that various well-established global teleconnections have on Sahel and Guinea Coast rainfall both before and after this sudden regime change. This will be the starting point of the present study, but first a brief overview of these global teleconnections will be summarized in the next section.

2.1.7 Global Teleconnections

Diatta and Fink (2014) published an article that outlined the statistical relationship between remote climate indices and West African monsoon variability. This study, perhaps the most comprehensive of its type, provides an excellent overview of the inﬂuence of remote variations of the atmosphere-ocean systems to rainfall variability in West Africa via so-called teleconnections. Di- atta and Fink (2014) divided West Africa into three regions, West Sahel, Central Sahel, and the Guinea Coast in order to test the relationship between the region’s rainfall and various indices, including the Atlantic Multidecadal Oscillation (AMO), the Atlantic Meridional Mode (AMM),

20 the Indian Ocean Dipole (IOD), the Southern Oscillation Index (SOI), the Pacific Decadal Oscil- lation (PDO) index, sea-surface temperatures (SSTs) in the Niño 3.4 region, the Oceanic Niño Index (ONI), Indian Ocean (IO) SSTs, Eastern Mediterranean Sea (EMS) SSTs, Atlantic 3 region (0°- 20°W and 3°S - 3°N) SSTs, and the All Indian Rainfall (AIR) series. The results of these correlations are results are shown in Table 2.1. A summary of their findings include: (1) a positive correlation between AMO and AMM and Sahel rainfall, (2) a robust and statistically significant correlation between ENSO indices (including N3.4 and ONI) and Sahel rainfall, (3) N3.4 and ONI are better suited as West African monsoon predictands than the SOI, (4) strong connec- tions between West African monsoon rainfall and SSTs in adjacent oceans, (5) strong connection between ATL3 SSTs and Guinea Coast rainfall, (6) opposing links between the Western Sahel and Guinea Coast rainfall to the AIR, (7) West African monsoon rainfall can be fairly accurately predicted when June-September combinations of adequate climate teleconnections indices are utilized. Many of these findings, particularly (2), (3), and (7), are fundamental to the design of the present study and will be discussed in further detail in following sections and chapters. In addition to the Diatta and Fink (2014) study, many other studies have focused on remote teleconnections to Sahel rainfall, and the role of sea-surface temperatures in particular. A review of the role that global sea-surface temperature anomalies play in association with interannual variability of rainfall across the region was completed by Rodrìguez-Fonseca (2015), which concludes that on interannual time scales, warming in the equatorial Atlantic and Pacific/Indian Oceans often results in a reduction of rainfall over the Sahel, while positive SST anomalies in the Mediterranean often have the opposite effect– increasing Sahel rainfall. Previously, Losada et al. (2012) suggested a change in the relationship between Sahel rainfall and SSTs after the 1970s. However, more recent work (Nicholson et al. in press) has instead suggested that the change actually took place a little earlier– around 1968, the year of the rainfall regime change discussed in the previous section. Their findings are shown in Figures 2.10 and 2.11, which show the correlation between Sahel and Guinea Coast rainfall with global SSTs both before and after 1968. During the earlier period, 1886 to 1967, Figure 2.10 shows similarities between the two regions (although correlations remain relatively weak) for both April-May, where the strongest correlations are in the Atlantic, and June, where the strongest correlations are

21 in the Indian Ocean. During October-November, there is a fairly strong correlation between Sahel rainfall and equatorial SSTs around the globe. The Guinea Coast correlations in the Atlantic are similar, but notably opposite in the Pacific, where a weak negative correlation suggests a potential negative link between El Niño and Guinea Coast rainfall in October-November. The largest and most robust difference between the Sahel and Guinea Coast are during the JAS season (Sahelian phase of the monsoon). Here, JAS Sahel rainfall is negatively correlated with equatorial SSTs in the Pacific, suggesting a relationship with the El Niño/Southern Oscillation (ENSO). This is not observed for the Guinea Coast, where rainfall is instead strongly and positively correlated with SSTs in the Gulf of Guinea just off the coast of the African continent. SST correlations after 1968, shown in Figure 2.11 (Nicholson et al. in press), show a much different picture in many cases. For the Sahel, June teleconnections with Indian Ocean SSTs switch from negative to strongly positive. In JAS, the strong correlations with Pacific and Atlantic SSTs remain, and become even stronger. The positive correlation between the Sahel and SSTs in the Mediterranean becomes stronger after 1968 and supports the findings of Rodrìguez-Fonseca (2015). The correlations for the Sahel in April-May and October-November remain mostly un- changed. For the Guinea Coast, correlations in JAS are noticeably different after 1968 versus before. After 1968, a strong negative correlation appears with equatorial Pacific SSTs, indicating the emergence of a potential ENSO teleconnection with Guinea Coast rainfall after the rainfall regime change that was not present before. In addition, a stronger link between Atlantic and In- dian Ocean SSTs for both JAS and October-November are seen for the Guinea Coast after 1968. The findings of Nicholson et al. (in press), most of which agree with the findings of Diatta and Fink (2014) and Rodrìguez-Fonseca (2015), serve to provide further evidence not only of a shift in rainfall around 1968, but also in the various global SST teleconnections as well.

2.2 El Niño/Southern Oscillation (ENSO) and West African Rainfall

The results of Nicholson et al. (in press) and others seem to suggest a relationship between the El Niño/Southern Oscillation and rainfall in certain parts of West Africa and in certain seasons.

22 They suggest that this relationship changed around 1968, coinciding with the major shift in the rainfall regime. Therefore, this project serves to expand on this idea by providing a comprehensive analysis of rainfall in the Sahel and Guinea Coast during ENSO events both before and after 1968. In this section, a brief overview of ENSO is provided, in addition to other previous studies speciﬁcally regarding the link between ENSO events and African rainfall.

2.2.1 Introduction to ENSO

The El Niño/Southern Oscillation (ENSO) is a coupled atmosphere-ocean phenomenon that is characterized by periodic fluctuations in SSTs across the equatorial Pacific Ocean, as well as periodic oscillations in atmospheric pressure between Darwin, Australia and the island of Tahiti. There are two alternating phases of ENSO, referred to as El Niño and La Niña, which occur at ir- regular intervals approximately every 2-7 years. ENSO events are largely considered the dominant source of interannual climate variability across the globe (Gergis and Fowler 2009). They have far-reaching effects, causing extreme weather events such as drought, flooding, fires, and tropical cyclone activity in various locations worldwide, affecting the livelihoods of millions. During “normal" years (i.e. non-El Niño or La Niña years), large-scale atmospheric circulation in the tropical Pacific is dominated by high pressure over the eastern Pacific and the convergence of strong easterly trade winds near the Equator. These winds promote a strong equatorial current flowing westward from South America towards Australia, resulting in a lower air pressure and a pileup of warm water in the western Pacific. In addition, a strong equator-ward Peruvian current along the coast of South America promotes upwelling of cooler SSTs to the surface, resulting in cooler SSTs in the east. During El Niño years, the easterly trade winds weaken, and in some instances can even reverse. This causes a reduction in upwelling and allows warmer than normal SSTs to extend from the coast of South America westward across the equatorial Pacific. The result of El Niño is higher pressure, dry conditions, and cooler SSTs in the western equatorial Pacific, but lower pressure, wet conditions, and warmer SSTs in the central and eastern equatorial Pacific. As an El Niño event comes to an end, the air pressure switches back to low in the west and high in the east. This oscillation of atmospheric pressure is referred to as the Southern Oscillation, and is the

23 atmospheric component of ENSO. The differences between “normal" and El Niño years are shown in Figure 2.12 (Lutgens and Tarbuck 2010). The counterpart to El Niño is La Niña, which is when the opposite phenomenon occurs. During La Niña events, the trade winds become stronger than normal and promote more upwelling in the eastern Pacific. This causes cooler than normal SSTs to extend over the eastern and central Pacific, enhancing the low pressure and bringing more rainfall to the western Pacific. Figure 2.13 (Vaughan 2014) shows equatorial Pacific SST anomaly composites typical of strong El Niño and La Niña events for comparison. Typical ENSO events last between 18-24 months and generally reach a peak toward the end of the calendar year (Rasmusson and Carpenter 1983). It is important to note that despite the common characteristics of all ENSO events, they can vary drastically in terms of strength, timing of onset and maturity, duration, and spatial extent of maximum SST anomalies (Rasmusson and Carpenter 1983; Trenberth and Stepaniak 2001). El Niño and La Niña events can be quantified and their intensities measured in a variety of ways. One method is to use pressure differences between Darwin, Australia and Tahiti, defining ENSO based the Southern Oscillation component (Ropelewski and Halpert 1989; Halpert and Ropelewski 1992). In addition, SSTs can also be used to define ENSO events, and there are various metrics that can be used to do this (Rasmusson and Carpenter 1983; Fu et al. 1986; Halpert and Ropelewski 1992). For example, SSTs can be averaged across certain zones in the equatorial Pacific (i.e. the Niño 2 region, the Niño 3 region, the Niño 4 region, or the Niño 3.4 region). It is generally accepted that the best SST measure of ENSO is by analyzing anomalies across the Niño 3.4 region, which will be discussed further in the next chapter as it related directly to this work.

2.2.2 El Niño and African Rainfall Teleconnections

Despite being shown to have a great inﬂuence on weather patterns across the globe, ENSO’s impact on sub-Saharan Africa has been a subject of controversy. Various studies have shown conclusive relationships between rainfall in Southern Africa and ENSO (Nicholson and Entekhabi 1986; Nicholson and Entekhabi 1987; Nicholson and Kim 1997; Nicholson et al. 2001), but there is some disagreement on its link with rainfall across the Sahel and Guinea Coast. This section

24 aims to elaborate on studies specifically dealing with El Niño’s teleconnection (in addition to the previously discussed Nicholson et al. (in press) study) with rainfall across the African continent, with a special focus placed on the Sahel. Nicholson and Entekhabi (1986) was one of the first in-depth analyses of the ENSO-African rainfall teleconnection. They used spectral analysis to define the Southern Oscillation using a normalized series of Tahiti-Darwin station pressure differences from 1935 to 1973. They found that the influence of the Southern Oscillation on African rainfall was geographically limited by concluding that the Southern Oscillation exerts a significant influence on rainfall fluctuations throughout most of southern and equatorial Africa on scales of 2-3, 3-5, and 5-6 years, but that its influence north of about 10°N is likely minimal. Southern Africa exhibited the best association with the Southern Oscillation, where low rainfall values were associated with low Southern Oscillation Index (SOI) values, while low-index values corresponded with an increase in rainfall in equatorial regions in the 2.2-2.4 year range. Ropelewski and Halpert (1987, 1989) used a novel harmonic method to define the ENSO signal based on the Southern Oscillation, and found similar results to that of Nicholson and En- tekhabi (1986) as it relates to an ENSO-African rainfall teleconnection. They again found that a high index phase of the Southern Oscillation was associated with enhanced precipitation in south- eastern Africa but reduced rainfall in eastern equatorial Africa. The most comprehensive study to date on the teleconnection between El Niño and African rainfall is Nicholson and Kim (1997). In this study, a variant of Ropelewski and Halpert’s (1987, 1989) methodology is applied to a more complete and regionally averaged dataset, pointing out that data was quite sparse in many areas of Africa during the Ropelewski and Halpert studies. As in Halpert and Ropelewski (1992), Nicholson and Kim defined an El Niño episode as the “2-year period commencing in July prior to the low-index year of the Southern Oscillation (designated as July -1) and continuing until June of the year after the low-index year (designated as June +1)." This definition is important as it became the basis of most ENSO-African rainfall related studies hereafter. Their results indicated, like previous studies, that the strongest signals occurred in southern, eastern, and far northern Africa, as well as near Atlantic coastal regions near the Equator. Although they note a noticeable absence of a harmonic signal over the Sahel, they point out that

25 there are clearly preferential anomalies linked to phases of the ENSO signal in the region. This leads to the conclusion that ENSO may account for some of the previously discussed teleconnections demonstrated between the Sahel and Southern Africa. In addition to contrasting regional responses, they also found that the rainfall response to ENSO across the continent appears to be seasonally specific across different sectors of the continent. Like ENSO events themselves, the rainfall response to ENSO among the various sectors of Africa varies significantly in magnitude, timing, duration, and consistency from episode to episode. Nicholson et al. (2001) conducted an analysis of recent rainfall conditions across West Africa, and specifically examined the 1997 El Niño, which is considered one of the strongest to date. They noted that while dry conditions did persist into 1997, the year was not unusually dry compared to other years in the previous two decades, leading to the conclusion that the strong 1997 El Niño event did not have a large impact in the Sahel. As previously mentioned and hinted at by these El Niño - African rainfall teleconnection studies, the relationship between El Niño and West Africa, particularly the Sahel and Guinea Coast, is somewhat complicated and poorly understood. This is due, in part, by the fact that there is a large variation of Sahel annual rainfall and that the ENSO signal is not strongest during JAS, which is the season of maximum rainfall in the Sahel (Janicot et al. 2001). Despite the findings of Nicholson and Entekhabi (1986), Ropelewski and Halpert (1987, 1989), Nicholson and Kim (1997), and Nicholson et al. (2001), some studies (Hastenrath et al. 1987; Wolter 1989; Ward 1992; Palmer et al. 1992) have found contradictory evidence that suggests that ENSO can result in changes within the West African monsoon, and that El Niño specifically has been associated with reduction in Sahel rainfall. In addition, Rowell et al. (1995) found that SSTs in the eastern equatorial Pacific are significantly correlated with Sahel rainfall at interannual time scales for periods of less than 11 years, and as discussed in the previous section, equatorial Pacific SSTs are another measure of ENSO in addition to the Southern Oscillation Index which was used by Nicholson and Entekhabi, Ropelewski and Halpert, and Nicholson and Kim. Janicot et al. (1996) even found that there were higher correlations between El Niño and Pacific SSTs after 1970 versus before, which suggests that there was a possible shift in the ENSO-Sahel teleconnection after the 1968 rainfall regime change discussed by Nicholson et al. (in press).

26 Furthermore, Janicot et al. (2001) suggested that by weakening the monsoon trough and moisture advection over West Africa, El Niño events could impact Sahel rainfall and be linked to an interaction with global decadal-scale SST variability. Rowell (2001) and others have suggested that the link between equatorial Pacific SSTs and Sahel rainfall is indirect, acting via teleconnection changes to Atlantic SSTs, although there is also a direct atmospheric teleconnection. They also found that when when large-scale zonal gradient of SSTs from the west Pacific to east Indian Ocean is weakened, Sahel drought becomes more likely. Another important finding of Rowell (2001) is that during El Niño years, a Kelvin wave propagates across the Atlantic from east Pacific convective heating anomalies while an equatorial Rossby wave appears over the Indian Ocean. These can interact over Africa and enhance subsidence over the Sahel, possibly resulting in lower rainfall totals. According to Losada et al. (2012), when the Sahel - Guinea Coast dipole disappeared in the 1970s, the anti-correlation between Sahel rainfall and tropical Pacific SSTs strengthened. The most recent study regarding the teleconnection between El Niño and rainfall over Africa is that of Parhi et al. (2016). They point out that detecting the ENSO signal in rainfall variability over the continent is difficult due to multiple other climate factors that have been discussed here previously, namely the North American Oscillation, Atlantic Multidecadal Oscillation, Indian Ocean dipole, the AEJ, TEJ, and the SHL, since all have significant impacts on year-to-year variability of Sahel and Guinea Coast rain as has been summarized in this chapter. Parhi et al. (2016) found that the El Niño teleconnection is associated with the decrease in seasonal mean rainfall and a decrease in the number of “wet" days. They separate El Niño events into two phases, a growth phase and a mature phase, and note that the JAS rainy season over the western Sahel corresponds with the growth phase of El Niño evolution, and thus, produces the drier climatic response over the region. However, they also note that despite the relationship, a daily extreme rainfall event can still occur in the western Sahel. To add to the complication, social science studies have also been conducted on the relationship between El Niño and life in the Sahel. Once such study, Okonokwo and Demoz (2014), found that a statistically significant relationship exists between ENSO and cereal production in the Sahel. Their finings suggest that El Niño events can cause significant problems for millet crops, while having little or no impact on the growth of maize and sorghum, which they attribute to the

27 lower length of growing period compared to millet. This seems to suggest that El Niño’s impact on Sahelian climate can vary based intraseasonal time scales as well.

2.2.3 La Niña and African Rainfall Teleconnections

While many studies have analyzed El Niño events and the debate over their connection to rainfall in West Africa, very few studies have dealt with El Niño’s counterpart La Niña directly. Nicholson and Selato (2000) utilized a similar methodology to that of Nicholson and Kim (1997) to define La Niña events based on a harmonic method. In general, their results were the exact opposite of the El Niño study, with reduced rainfall over much of Africa during the first portion of the La Niña episode, and increased rainfall during the second half. They found that La Niña has the greatest influence on rainfall in southern Africa, and that wet periods tend to occur across the continent in the first few months of the year following the La Niña year. Like the equatorial-southern Africa contrast in the El Niño article, there also existed a contrast between eastern equatorial Africa and southern Africa during La Niña events as well, but it was somewhat weaker. Their conclusions also agree with that of Janicot et al. (2001) that ENSO’s impact on SSTs in the Atlantic and Indian Oceans are the primary influence on African rainfall. In addition to analyzing the 1997 El Niño episode, the previously discussed Nicholson et al. (2000) also looked at the strong La Niña year of 1998. They found that throughout most of the Sahel, rainfall throughout 1998 was still below the long-term mean, although Niger in the central Sahel did experience some localized flooding due to high rainfall in September. Despite being one of the wettest years in recent decades, 1998 Sahel rainfall was still considered to be well below normal overall, and only exceeded the long-term mean in a few sectors of West Africa.

2.3 Research Questions and Objectives

In light of the much debated relationship between El Niño and La Niña and West African rainfall, this study serves to add to the current literature by providing the most comprehensive and complete analysis of the relationship between ENSO and rainfall across the Sahel and Guinea Coast to date. This will be accomplished by utilizing the largest and longest dataset of rainfall gauge data across the Sahel and Guinea Coast (the same used by Nicholson et al. in press). In

28 addition, a new and updated list of El Niño and La Niña events dating back to 1921 will be created using a novel methodology based on that of the recently updated (2016) methodology used by the National Oceanic and Atmospheric Administration’s Climate Prediction Center. This updated and more complete list of ENSO events, in conjunction with an ever-increasing dataset of African rainfall gauge data, will provide a fresh look at the regional responses of rainfall across West and Northern Africa during El Niño and La Niña years during the 1900s and into twenty-ﬁrst century. The major focus of this study will be on analyzing Sahel and Guinea Coast rainfall during ENSO events in the context of the 1968 rainfall regime change discussed in section 2.1.6. This work is intended to build upon the work of Nicholson et al. (in press) discussed in 2.1.7 and shown in Figures 2.10 and 2.11. By utilizing combinations of rainfall regions that have been deﬁned in previous studies (to be discussed in more detail in the next chapter), various sectors of the Sahel and Guinea Coast will be analyzed separately in order to detect regional variations in the rainfall response. In short, this study aims to answer the following questions: What is the seasonal cycle of rainfall anomalies during different phases of El Niño across the various rainfall regions of the Sahel and Guinea Coast both before and after the 1968 rainfall regime change? What variations exist from region-to-region, season-to-season, and before and after 1968? How does the seasonal cycle of rainfall anomalies during different phases of La Niña events across the various regions differ from those of El Niño? What are the differences in La Niña year rainfall anomalies before 1968 vs. after? Do atmospheric circulation patterns, such as zonal and meridional winds, differ during ENSO events before and after 1968? How consistent is the ENSO signal among the various regions of the Sahel and Guinea Coast before and after 1968? This study will address all of these questions using a fairly straightforward but effective and consistent methodology. Not only will it provide new information about rainfall in West Africa during ENSO events, but by placing it in the context of the abrupt change in the rainfall regime, it will attempt to provide insight into potential mechanisms and contrasts observed in the ENSO response both before and after the documented regime change.

29 Figure 2.1: Adopted from Nicholson (2009). Classic picture of the ITCZ over Africa. In recent years, a revised view of the ITCZ has been developed.

Figure 2.2: Adopted from Nicholson (2009). Revised view of the West African Monsoon.

30 Figure 2.3: Adopted from Nicholson et al. (1988). Schematic of boreal winter (top) and boreal summer (bottom) atmospheric circulation patterns (including winds and surface pressure) over the African continent. Arrows depict relative wind speed and direction, thin dashed circles depict various jet streams, the dotted line indicates the location of the ITCZ, and solid lines indicate surface pressure contours (in mb).

31 Figure 2.4: Adopted from Thorncroft et al. (2011). Schematic showing the four main phases of the annual cycle of the West African monsoon. Each phase shows the location of the main tropical rain belt (indicated by the location of clouds and rainfall with peak values indicated by darker shades), the location of the SHL (yellow, orange, and red shading at the surface with darker orange/red indicating stronger intensity), Atlantic SSTs and mixed layer depth (indicated by colors red-green-blue in order of decreasing SSTs), moisture ﬂux convergence (solid contours) and divergence (dashed contours), and deep and shallow meridional circulations (blue and red lines with arrows). Where there is uncertainty surrounding the extent to which the circulation return ﬂow penetrates the tropical rain belt, the lines have been dotted.

32 Figure 2.5: Adopted from Flohn (1964). The annual migration of the tropical rain belt across the African continent and the associated patterns of rainfall seasonality as a function of latitude.

Figure 2.6: Adopted from Nicholson (2008). Schematic showing the four most common rainfall anomaly patterns over West Africa. Light shading indicates below normal rainfall, while dark shading indicates above normal rainfall.

33 Figure 2.7: Adopted from Mitchell (2016). Rainfall anomalies in cm/month from 1901 to the present. Notice the periodic shift between wet years and dry years throughout the twentieth century, as well as a prolonged drought during the 1970s and 80s.

Table 2.1: Adopted from Diatta and Fink (2014). Correlation coefficients between remote indices of climate variability and the West Sahel (WS), Central Sahel (CS), and Guinea Coast (GC). See text for further information on abbreviations. Correlation coefficients in bold are significant at the 95% significance level, while those in bold and with an asterisk are significant at the 99% level according to the F-test using the standard degree of freedom. Unless otherwise indicated, the time period 1921- 2009 was analyzed.

Climate Index WS CS GC AMO 0.281 0.287* -0.069 AMM (1948-2009) 0.470* 0.372* -0.116 IOD -0.300* -0.230 0.056 SOI 0.349* 0.283 0.016 PDO -0.300* -0.378* -0.154 N3.4 -0.352* -0.339* -0.001 ONI (1950-2009) -0.320 -0.353* -0.020 IO SST -0.483* -0.428* -0.100 EMS SST 0.263 0.447* -0.027 ATL3 SST -0.348* -0.250 0.524* AIR 0.316* 0.329* -0.166

34 Figure 2.8: Adopted from Nicholson et al. (in press). Standard departures of Sahel and Guinea Coast rainfall. This time series is based on an updated dataset produced by Nicholson (see Nichol- son 1986; Nicholson et al. 2012) of gauge data that is considered the longest and most complete dataset of Sahel and Guinea Coast rainfall available. The green bars at the bottom of each plot indicate the total number of stations. Red indicators are placed on years where a factor is applied to adjust for the changes of variance when the number of stations is relatively small, following the methodology of Nicholson (1986).

35 Figure 2.9: Adopted from Nicholson et al. (in press). Seasonal standard departures of Sahel and Guinea Coast rainfall. A dashed line has been drawn at 1968 to mark the major change in the rainfall regime.

36 Figure 2.10: Adopted from Nicholson et al. (in press). Correlations between rainfall in the Sahel (left) and Guinea Coast (right) and global SSTs for the years 1886 - 1967 for all four phases of the West African monsoon.

37 Figure 2.11: Adopted from Nicholson et al. (in press). Correlations between rainfall in the Sahel (left) and Guinea Coast (right) and global SSTs for the years 1969 - 2013 for all four phases of the West African monsoon.

38 Figure 2.12: Adopted from Lutgens and Tarbuck (2010). Simplified illustration of the seasaw pattern of atmospheric pressure between the eastern and western Pacific, called the Southern Os- cillation. (a) During average years, high pressure over the eastern Pacific causes surface winds and warm equatorial waters to flow westward. The result is a pileup of warm water in the western Pacific, which promotes the lowering of pressure. (b) An El Niño event begins as surface pressure increases in the western Pacific and decreases in the eastern Pacific. This air pressure reversal weakens, or may even reverse the trade winds, and results in an eastward movement of the warm waters that had previously accumulated in the western Pacific.

39 Figure 2.13: Adopted from Vaughan (2014). Composites of equatorial Paciﬁc SST anomalies (°C) during OND of (top) strong El Niño years and (bottom) strong La Niña years. These composites demonstrate classic ENSO equatorial Paciﬁc SST signatures during El Niño and La Niña events.

40 CHAPTER 3

METHODOLOGY

In order to achieve this project’s goal of examining rainfall both before and after the 1968 rainfall regime change during El Niño and La Niña years, several datasets and methodologies were employed. This chapter serves to describe the data, methods, and variables analyzed within the scope of this project. An effort has been made to remain as consistent as possible with previous studies and utilize practices common in the ﬁeld of climatology, particularly as it relates to Sahel and Guinea Coast rainfall studies.

3.1 Time Frame of Study

All analyses in this study, unless noted otherwise, cover the years 1921 to 2012, which were selected for several reasons. First, by spanning 91 years, it is long enough to provide a complete picture of rainfall both before and after the 1968 rainfall regime change. Additionally, although the rainfall dataset goes back as far as the 1860s for some locations, the number of stations across all regions becomes fairly robust beginning in the 1920s, adding conﬁdence in the data’s integrity around this time. Furthermore, beginning the analysis in year 1921 establishes some consistency among African rainfall teleconnection studies, since it was also selected as the ﬁrst year of the statistical analysis performed by Diatta and Fink (2014). Finally, although the dataset used to determine El Niño and La Niña years goes back as far as 1854, early years consisted of sparse data. After 1880, the signal strength becomes much more consistent, aligning nicely with the chosen 1921-2012 time frame of this study.

41 3.2 Rainfall Data 3.2.1 Standard Departures

The primary dataset used in this project is based on rainfall data collected from 876 rain gauge stations throughout the Sahel and Guinea Coast, derived from a much larger dataset consisting of over 1,000 stations across the entire African continent. This project utilized regionally averaged standard departures for each month in each region from January 1921 to December 2012. Standard departures have long been the standard for African rainfall studies, and it is has become common practice to use them in lieu of actual rainfall totals when studying a departure from mean rainfall (i.e. “wet" years and “dry" years). In previous studies by Nicholson and many others, station rainfall is expressed as a standard-

ized departure xi from the long term mean (i.e. the mean over the entire length of the record). The methodology for calculating monthly xi values for each region is described in Nicholson and Palao (1993) and reproduced here. For each station, the departures are calculated using the equation

r − ri xi = (3.1) σi

where r is the rainfall total in the month, ri is the mean monthly rainfall, and σ i is the standard deviation of rainfall totals from the mean. These departures are aerally integrated over a particular region or zone (described in further detail in the next section), resulting in one monthly rainfall departure for that month for that region, R, using the equation

−1 R = I ∑xi (3.2) i where I is the number of stations in the region and the summation is made over all I regions.

The standard departures are thus unit-less quantities that provide an idea of how far rainfall within that region is away from the mean. For example, a value of 1 would indicate an average regional anomaly of one standard deviation from the mean, and so on. Positive values indicate above normal rainfall, while negative values indicate below normal.

42 3.2.2 Rainfall Regions

As described above, standard departure values are calculated for individual stations and then averaged across various rainfall regions, shown in the top portion of Figure 3.1. This regionalization scheme is that of Nicholson et al. (2000), a slightly modiﬁed version of the regions used in the earlier studies of Nicholson (1985, 1994), Nicholson et al. (1988), and Nicholson and Palao (1993). Originally, the regions were based primarily on vegetation zonation as was the practice in many climatological studies following the severe drought in the early 1970s. However, later research showed that this practice was not adequate in some regions. Therefore, a revised regionalization scheme (the one shown in the top portion of Figure 3.1) was developed based on climatology and station/region correlations (Nicholson et al. 2000). For this study, only regions in the Sahel and Guinea Coast were used, namely regions 9, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 28, and 29. Table 3.1 shows a list of the regions used, the number of stations in each region, and the latitude, longitude, and country of each region’s geographic center point. As evidenced by the table, the number of rainfall stations varies greatly from region to region. For example, region 9 contains only eight stations, while region 19 contains 163. Additionally, the number of stations has changed over time, with a large increase in the number of stations during the mid-twentieth century. Corrections to the dataset have been made to account for these changes. In order to consolidate calculations and analysis of the regional rainfall response to ENSO across the Sahel and Guinea Coast, some of the aforementioned regions are combined in this study. The new, combined regions are outlined in the bottom portion of Figure 3.1. The various combinations were based on several factors, including initial results of the ENSO signal response among the individual regions as well as the correlation of rainfall anomalies between the regions. In general, regions with similar initial ENSO signal results and with standard departures in relatively high correlation with each other were combined. In cases where the ENSO signal appears fairly unique and there is little correlation with other regions, separation was maintained (as in regions 22 and 23). Hereafter, the seven combined regions analyzed in this study will be referred to by their respective labels in the bottom portion of Figure 3.1. In general, region 9/13/18 covers the western Sahel, regions 14/19 and 15/20 cover the central Sahel, region 16/17/21 covers the eastern Sahel,

43 region 22 includes the far east and is heavily inﬂuenced complex topography, region 23 serves as a transition zone between the western Sahel and Guinea Coast, and region 24/28/29 spans the Guinea Coast.

3.2.3 Data Adjustments

Now that the general methodology of defining standard departures and rainfall regions has been outlined, there are a couple of important modifications to the data utilized in this study that should be noted. First, as described in section 3.2.1, the standard practice has long been to calculate standard departures based on the long term mean, defined as the mean over the entire length of the record. However, since this study deals specifically with analyzing the ENSO signal in the rainfall both before and after 1968, an adjustment was made and this is not the case. The standard departures used in this study were instead calculated so that the years 1921-1967 were based on the mean and standard deviation from 1921-1967, while the departures for 1968-2012 were based on the mean and standard deviation from the 1968-2012 time period. This adjustment was made to mitigate the effect of the long-term drying signal from 1968-2012 in order to better show the seasonal ENSO impact on rainfall. Otherwise, rainfall departures during El Niño years after 1968 would almost unanimously be less than their pre-1968 counterparts, not necessarily because of a change in the rainfall response to ENSO, but because rainfall across almost all years after 1968 was less than than the long term mean. By recalculating the departures based on the mean and standard deviations for each time period (1921-1967 and 1968-2012), the effect of the long-term drought was essentially removed and any changes in the ENSO signal itself can be detected. Additionally, it should be noted that when the individual regions were combined, the monthly standard departure values were averaged over all stations within the new combined region. For example, the calculated monthly standard departures for combined region 9/13/18 are not simply an average of each of the three individual region’s standard departures. This cannot be done because each region has a different number of stations. Instead, the standard departures for the combined region were recalculated using all stations located in regions 9, 13, and 18 collectively. The known limitation of doing this, however, is that region 18 will have a much larger influence on the overall

44 combined region’s standard departure values, since it contains many more stations than regions 9 and 13.

3.3 El Niño and La Niña Events

One of the most complicated aspects of this study was not necessarily calculating the rainfall departures themselves, but instead determining which years to include as El Niño and La Niña years for analysis. This was a problem because nearly every article dealing with ENSO case studies outlined in Chapter 2 used a slightly different list of ENSO events based on a variety of different methodologies, each with its advantages and drawbacks. This section highlights a few of the methodologies of deﬁning ENSO episodes that were employed in previous studies, an overview of the current method used by the NOAA Climate Prediction Center (CPC), and ﬁnally the list of years this study utilized and a description of how they were determined.

3.3.1 Historical Deﬁnitions of ENSO

There are a variety of ways that ENSO can be defined, and previous studies have utilized a wide variety of methods and datasets. As previously mentioned, the use of SST anomalies in the equatorial Pacific Ocean is generally considered one of the best methods of defining ENSO events. Although there are several Niño regions, as shown in Figure 3.2, the Niño 3.4 region is by far the most commonly used region. Comprised partially of the Niño 3 region and partially of the Niño 4 region, it covers a spatial extent from 5°N to 5°S and 120°W to 170°W. Parhi et al. (2016) classified El Niño, La Niña, and neutral years based on the condition that the running average of July-December SST anomaly over the Niño 3.4 region be greater than 0.4°C, less than -0.4°C, or in between -0.4°and 0.4°C respectively. This methodology was based on the canonical definition in Trenberth (1997) and data obtained from Kaplan et al. (1998). Diatta and Fink (2014) also used SST anomalies obtained from the Niño 3.4 region to define the ENSO signal, but they obtained their SST values from the Hadley Centre Sea Ice and Sea Surface Temperature (HadISST) dataset and computed anomalies with respect to the 1950-1979 mean. They smoothed the anomalies with a 5-month running mean, and normalized the smoothed anomalies by their standard deviation over the period 1950-1990.

45 Regions other than Niño 3.4 have also been used in the past. For example, Fu et al. (1986) cataloged ENSO events based on SST anomalies from 0°to 10°S and from 160°W to the South American coast from 1927 to 1985 based on a reference mean period of 1950-1979. Note that, interestingly enough, this area does not align with any of the Niño regions shown in Figure 3.2. In contrast, Torrence and Webster (1999) obtained SST data from the U.K. Meteorological Of- fice (UKMO) Global sea-Ice and Sea Surface Temperature (GISST) 2.3-b dataset to define events based on anomalies in the Niño 3 region. This dataset uses empirical orthogonal functions (EOFs) determined from the 1960-1991 climatology of Parker et al. (1995) to fill data-void regions and years. Trenberth and Stepaniak (2001) also point to previous studies utilizing SST anomalies of +/- 0.5°C in the Niño 3 region. In addition to SST anomalies, the Southern Oscillation Index (SOI) has also been used to define ENSO years in many different studies. In general, the difference in pressure between Darwin, Australia and Tahiti is used in determining the SOI index. In most cases, so-called “high index years" correlate with La Niña events, while “low index years" correlate with El Niño events. Rasmusson and Carpenter (1983), Ropelewski and Jones (1987), Torrence and Webser (1999), and Diatta and Fink (2014) all used some variation of the SOI as well as SSTs to compile their lists of ENSO events, although their datasets and lists all differ slightly. Fu et al. (1986) highlights the variations among lists of ENSO years by comparing their list to those of Trenberth (1984), Quinn et al. (1978), van Loon and Madden (1981), and Rasmusson and Carpenter (1983). The major drawback of using only the SOI as a measure of ENSO is that it relies on data collected at just two points, which could be subject to a large degree of variability, thus compromising accuracy. For this reason, most studies that include SOI as an indicator of ENSO tend to also include SST anomalies in their analysis as well, defining their ENSO years based on a hybrid of the two main methodologies. In addition to using SST and atmospheric pressure indices, Gergis and Fowler (2009) used a number of percentile-based paleoclimate reconstructions to isolate signals of both the El Niño and La Niña phases of ENSO. They were able to compile a list of ENSO events from 2002 all the way back to 1525 and classify the events as weak, moderate, strong, very strong, or extreme. Their list included a total of 92 El Niño events and 82 La Niña events, and found that 43% of extreme and

46 28% of all prolonged ENSO events occurred in the twentieth century. Because the scope of this project includes years since 2002, the list compiled by Gergis and Fowler (2009) was considered inappropriate to use in this study, since a different methodology would be required to add-on the years since 2002. Finally, many studies on ENSO-rainfall teleconnections have defined their ENSO events based on previous work. For example, Nicholson and Kim (1997) simply used the years identified by Ropelewski and Jones (1987) as “low index years" as their list of El Niño years. Similarly, Nicholson and Selato (2000) used the list of years outlined in Ropelewski and Halpert (1989) as their list of La Niña years. Schonher and Nicholson (1989) took a consensus approach, choosing the years which were classified as ENSO by at least two other authors. Other studies have used similar consensus techniques to create their own lists based on previous work. The purpose of highlighting the above inconsistencies in defining El Niño and La Niña years is to provide a sampling of the different methodologies that have previously been used to define ENSO events. It should be evident based on the wide degree of variability and differences in the lists of the above authors, that defining El Niño and La Niña years has, and likely will continue to be, a point of controversy. None of the lists encountered during a review of literature included the entire time period 1921 to 2012, the length of this study. Since most classical ENSO methodology studies are quite old, studies which incorporated years dating back to 1921 ended prior to the turn of the century. On the other hand, many lists compiled using “new" or “updated" methodology tended to include most recent years, neglecting those prior to 1950 or so. For this reason, it was determined that a consensus approach like that of Nicholson and Kim (1997), Nicholson and Selato (2000), and Schonher and Nicholson (1989) would not be sufficient for this study. In an effort to establish some consistency, this study instead utilized the current widely accepted definition of ENSO events as defined by the National Oceanic and Atmospheric Administration’s (NOAA) Climate Prediction Center (CPC) to create its own list of years for analysis. This definition and methodology was recently updated by CPC in 2013 and is based on the most up-to-date data available. Additionally, it is accepted worldwide by various meteorological organizations and prediction centers as one of the best SST-based indicators of ENSO. The new CPC definition and methodology that was used in this study is described in detail in the next section.

47 3.3.2 Climate Prediction Center (CPC) Methodology

The NOAA CPC uses SST anomalies located in the Niño 3.4 region to create an Oceanic Niño Index (ONI) and define ENSO episodes. They utilize the Extended Reconstructed Sea Sur- face Temperature (ERSST) version 4 dataset (Huang et al. 2015), which is a global monthly SST dataset derived from the International Comprehensive Ocean-Atmosphere Dataset (ICOADS). Ranging from 1854 to the present, it is gridded at 2°x 2°resolution. Because of sparse data in early years, there is damping in the analyzed signal. However, after 1880 the signal is more consistent over time. Figure 3.3 shows globally averaged SST anomalies and uncertainty with the 95% confidence level (Huang et al. 2015) for the entire length of the dataset. Statistical methods have been applied to ERSST to enhance the spatial coverage of the data. Because of the long length of the record, ERSST is considered appropriate for studying long-term global and basin-wide studies, such as the one this project aims to achieve. With the SST data derived from ERSST, CPC calculates ONI values by using a three month running mean of SST anomalies averaged across the Niño 3.4 region. The anomalies are based on centered 30-year base periods updated every 5 years. As an example, ONI values during 1950-1955 are based on an average of base period 1936-1965; ONI values during 1956-1960 are based on the 1941-1970 base period, and so on. A graph of various base periods used during the twentieth century is shown in Figure 3.4. Note the warming trend over time, with the most recent base periods averaging around half a degree warmer each month than the first base period shown (1936- 1965). Note that this methodology differs from the previous (pre-2013) method of computing all ONI values based on fixed 30-year periods (i.e. 1971-2000 and the later 1981-2010). This update was made in 2013 in response to a significant warming trend in Niño 3.4 SSTs since 1950. ONI calculated based on fixed periods were incorporating a longer-term trend and not accurately reflecting interannual ENSO variability (NOAA). NOAA identifies that the two main advantages to the new strategy are that classification of El Niño and La Niña episodes will remain fixed over most of the historical record and the centered 30-year base periods allow for El Niño and La Niña episodes to be defined based on their contemporary climatology. Once ONI values are calculated using three month running means of SST anomalies based on sliding 5-year averages, El Niño and La Niña years can be defined. CPC defines a year as an

48 El Niño year if the three month anomaly in the Niño 3.4 region is above a threshold of 0.5°C for a minimum of five consecutive overlapping seasons. Likewise, La Niña years are identified if the anomalies are less than -0.5°C for a minimum of five overlapping seasons. A table of ONI values from 1950 to the present is published by CPC online, and can be used to identify El Niño and La Niña events from 1950 to 2017. The values can also be used to gauge the intensity of each ENSO event, since some events are much stronger than others. As it pertains to this study, the obvious issue with using years published by the CPC utilizing the methodology outlined above lies in the fact that, although the ERSST dataset contains a fairly reliable time series of SST data as far back as 1854, the ONI table published by CPC that is publicly available online only goes back to the year 1950. Since this study aims to analyze rainfall during El Niño and La Niña years beginning in 1921, the CPC methodology outlined above must be applied for the entire time period 1921-2012 to create a unique list of ENSO events for use in this project.

3.3.3 ENSO Events in the Twentieth Century

To calculate ONI based on the CPC Methodology described in the previous section, ERSST v4 data was obtained from the NOAA Office of Oceanic and Atmospheric Research’s Earth System Research Laboratory Physical Sciences Division, available online at https://www.esrl.noaa.gov/. The data, which consists of network common data form (netCDF) files, was read into Matrix Lab- oratory software (MatLab), a numerical programming software and language commonly used to manipulate large datasets, perform calculations, and create plots. MatLab was then used to calculate the three month running means of SST anomalies (in °C) based on the appropriate base periods for each five year period from December-February (DJF) through November-January (NDJ) of each year from 1921 to 2012. Once the calculations were complete, conditional formatting was applied to the resulting table using Microsoft Excel, and the final ONI tables, shown in Table 3.2 and Table 3.3, were produced. These tables were designed to mirror the format of the table published online by the CPC for the years after 1950 available at http://www.cpc.ncep.noaa.gov/. Table 3.2 contains ONI values for the years before 1968 (1921-1967), while Table 3.3 contains ONI values for the years after 1968 (1968-2012). Values greater than or equal to 0.5 are highlighted in red, while those less than or equal to -0.5 are highlighted in blue.

49 Finally, a list of El Niño years and La Niña years (Table 3.4) was compiled based on the results in Tables 3.2 and 3.3. When five or more consecutive “red" ONI values appear, that year was classified as an El Niño event. Likewise, La Niña years were defined as years consisting of five or more consecutive “blue" values. In cases when the El Niño or La Niña spans the course of more than one year, only the first year of the ENSO episode was considered the ENSO year. In keeping with the practice of Nicholson and Kim (1997) and Nicholson and Selato (2000), all analyses conducted in this study considered an ENSO event as the 24-month period commencing in the July-September season of the year prior to its onset (JAS -1) and continuing until the April- June season of the year following the first year of onset (AMJ +1). For this reason, El Niño and La Niña events spanning more than 2 years were excluded from this study and are therefore not listed in Table 3.4. Note that this excluded only a few events: the El Niño years of 1939-1942, 1957-1959, and 1986-1988 and the La Niña years of 1954-1956 and 1998-2001. The years included in Table 3.4 and in the calculations and analysis of this study are fairly consistent with those used in prior studies. Most notably, the ONI values that were calculated here are unanimously consistent with those that are made publicly available by the CPC for the years after 1950. This implies confidence in the calculations for the years pre-dating 1950 as well. When comparing El Niño years in Table 3.4 to those used by Ropelewski and Jones (1987) and Nicholson and Kim (1997), only slight differences were observed. Excluding the multi-year El Niño events of 1939-1942, 1957-1959, and 1986-1988, the only discrepancies are that the previous two studies also included 1932 and 1977 which are not included here, but this study included the years 1945, 1968, and 1976 which they excluded. When comparing La Niña years to those of Nicholson and Selato (2000), a few slight shifts in the timing of events were noted. For example, Nicholson and Selato (2000) included the years 1950, 1971, and 1974, whereas this study included 1949, 1970, and 1973. These slight differences are not too surprising given that their methodology was based on the Southern Oscillation Index and there is likely some degree of lag between the atmospheric response and SST response to ENSO. Regardless, the years in Table 3.4 are largely consistent with those used in various other studies, and a degree of confidence is placed in their accuracy.

50 3.4 Rainfall Calculations

Using the El Niño and La Niña years outlined in Table 3.4 and the adjusted rainfall data discussed in Section 3.2, MatLab was used to calculate three month running averages of seasonal standard departures from JAS (-1) to AMJ (+1) during (a) all years from 1921 to 2012, (b) all years before 1968, (c) all years after 1968 (d) El Niño years before 1968, (e) El Niño years after 1968, (f) La Niña years before 1968, and (g) La Niña years after 1968 for each of the combined regions shown in Figure 3.1. In addition, to easily show contrasts in the ENSO rainfall response before 1968 versus after, differences between them were computed by subtracting the earlier time frame from the later one. This way, negative values represent drier conditions after 1968, while positive values indicate an increase in rainfall. The resulting data tables were organized in Microsoft Excel to add conditional formatting and are displayed in the next chapter. In order to visually display the differences before 1968 versus after during El Niño and La Niña years, line graphs of the seasonal evolution of rainfall departures during ENSO events were created. This allowed for differences in the ENSO rainfall response both before and after 1968 to be easily detected and allowed for comparison among the different regions to detect any regional contrasts throughout the Sahel and Guinea Coast. The consistency of the ENSO signal throughout the Sahel and Guinea Coast was also calculated as a percentage of events crossing a standard departure threshold of +/- 0.1 during several seasons. The results of these calculations were plotted spatially on maps of the regions to easily compare the anomalies and their consistencies.

3.5 An Examination of Atmospheric Circulations

As evidenced in Chapter 2, atmospheric circulations play a large role in rainfall variability across the Sahel and Guinea Coast. Therefore, in addition to analyzing differences in rainfall departures during El Niño and La Niña years before and after the 1968 rainfall regime change in the various rainfall regions, a similar analysis was also performed of upper level circulation patterns. It was hypothesized that, if the rainfall regime shows any differences between ENSO events before 1968 versus after, then upper level winds likely would as well.

51 Grid Analysis and Display System (GrADS), a software program typically used for dis- playing environmental data and performing mathematical computations, was used to display cross sections of both zonal and meridional winds ranging in height from the surface up to 100mb along 0°longitude, chosen because it bisects many of the Sahel and Guinea Coast regions. Data was obtained from the NCEP/NCAR Reanalysis 1 dataset, which consists of wind data available 4 times a day (00Z, 06Z, 12Z, and 18Z) beginning January 1, 1948 to the present. Cross sections were produced for the seasons JAS (-1), OND (-1), AMJ, JAS, OND, JFM (+1), and AMJ (+1) for El Niño and La Niña years before 1968 (beginning here at 1948 instead of 1921 due to data limitations) and after 1968, in addition to non-ENSO years both before and after 1968 to be used for comparison. The plots were formatted in a similar manner to those of Nicholson and Grist (2001) for consistency, and the results are summarized in Chapter 4. In summary, the results chapter of this study presents graphs of the seasonal evolution of standard departures for the regions analyzed during ENSO years before and after 1968, consistency plots, and zonal wind cross sections for each season during both El Niño and La Niña events. The results are analyzed in the next chapter and several conclusions are outlined in Chapter 5, along with implications and suggestions for future work.

52 Table 3.1: A list of rainfall regions utilized in this study, the number of gauge stations in each region, and the coordinates and country of the geographic center point of each region. The color labels correspond to the color of the combined region in the bottom portion of Figure 3.1 that contains that individual region.

Region Number of Country of Latitude Center Longitude Center Number Stations Center Point

9 8 18.85°N 12.64°W Mauritania

13 25 16.17°N 14.95°W Senegal

14 43 15.56°N 1.72°W Mali

15 35 15.02°N 9.86°E Niger

16 21 13.44°N 22.58°E Sudan

17 81 13.59°N 32.42°E Sudan

18 124 13.79°N 11.92°W Mali/Senegal border

19 163 12.88°N 1.95°W Burkina Faso

20 85 12.97°N 9.12°E Niger/Nigeria border

21 36 10.19°N 19.61°E Chad

22 13 10.96°N 34.33°E Sudan/Ethiopia border

23 54 10.72°N 8.34°W Guinea/Mali border

24 48 10.49°N 2.1°W Ghana

28 120 6.92°N 1.84°W Ghana

29 20 6.76°N 6.33°E Nigeria

53 Figure 3.1: Individual rainfall regions across the entire African continent deﬁned in previous studies by Nicholson (top) and the combined regions utilized in this study (bottom).

54 Figure 3.2: Adopted from the National Oceanic and Atmospheric Administration. Summary of the Niño regions in the equatorial Paciﬁc Ocean. SST anomalies within the Niño 3.4 region (5°N to 5°S; 120°W to 170°W) are generally used to determine the phase of ENSO.

Figure 3.3: Adopted from Huang et al. 2015. Monthly and globally averaged ERSST v4 anomalies from 1854-2014. The data becomes much more consistent around 1880, with the most reliable data available after the 1940s, as evidenced by the shrinking uncertainty spread.

55 Figure 3.4: Adopted from the National Oceanic and Atmospheric Administration. Graph of base periods used by CPC for calculating Niño 3.4 SST anomalies. The base period shifts every 5 years to accommodate for a long-term warming trend over time.

56 Table 3.2: Calculated ONI values before 1968 (1921-1967). Values of 0.5 or greater are highlighted in red and values of -0.5 or less are highlighted in blue. El Niño episodes are defined when there are 5 red ONI values in a row, while La Niña events are classified as years with five consecutive blue values. In cases when an episode spans multiple years, the first year (year of onset) is considered the ENSO year.

57 Table 3.3: Same as Table 3.2, except for the years after 1968 (1968-2012).

58 Table 3.4: List of El Niño and La Niña Years from 1921 to 2012 that are used in this study. These years are determined based on the Oceanic Niño Index (ONI) values that are shown in Tables 3.2 and 3.3. Note that all analyses conducted in this study will consider an ENSO event as the 24- month period commencing in the July-September season of the year prior to its onset (JAS -1) and continuing until the April-June season of the year following the ﬁrst year of onset (AMJ +1). For this reason, El Niño and La Niña events spanning more than 2 years have been excluded to maintain consistency.

El Niño Years La Niña Years 1923 1924 1925 1933 1930 1938 1945 1942 1951 1949 1953 1964 1963 1967 1965 1970 1968 1973 1969 1975 1972 1984 1976 1988 1982 1995 1991 2007 1994 2010 1997 2011 2002 2004 2012

59 CHAPTER 4

RESULTS AND DISCUSSION

4.1 Rainfall Anomalies During El Niño and La Niña Years

This section discusses the results of the rainfall anomaly calculations. There are some inconsistencies among the regions as evidenced by Figures 4.1-4.6, but generalizations have been made based on the data shown in Tables 4.1 and 4.2. These tables show standard departures for each region calculated for El Niño and La Niña years before 1968, after 1968, and differences between the two time periods. This last panel of the table provides a mechanism for easily comparing the two time periods, with red (negative) values indicating drier conditions after 1968 and blue (positive) values indicating wetter conditions. For both the Sahel and Guinea Coast regions, a discussion is provided comparing El Niño years to La Niña years, El Niño years before 1968 to after 1968, and La Niña years before 1968 to after 1968. Section 4.1.1 includes analysis of the Sahel and transition regions 9/13/18, 14/19, 15/20, 16/17/21, 22, and 23, while section 4.1.2 provides a discussion of the Guinea Coast region 24/28/29. Finally, a comparison of the Sahel and Guinea Coast is made in Section 4.1.3, which addresses the question of the Sahel/Guinea Coast dipole in the context of changes in the ENSO signal before and after 1968.

4.1.1 Sahel Regions

The seasonal evolution of standard departures from JAS (-1) to AMJ (+1) during El Niño and La Niña years for Sahel regions is shown in Figures 4.1-4.6, with the calculated departure values displayed in Tables 4.1 and 4.2. For analysis in the Sahel, particular emphasis is placed on the JAS (-1) and JAS seasons since they correspond to the peak of the monsoon season as discussed in Chapter 2. Emphasis is also placed on the AMJ, OND (-1) and OND seasons, given that they showed interesting contrasts to JAS in most cases. OND is also the season when ENSO events commonly reach their maximum intensity. Rainfall anomalies were considered robust if the

60 standard departures were of +/- 0.1. The dry season, JFM, was largely ignored for Sahel regions due to the scarcity of rainfall that occurs during this time.

El Niño vs. La Niña Years. Several robust differences were observed in Sahel rainfall anomalies between the two main phases of ENSO. During the JAS (-1) season, positive anomalies were observed in most Sahel regions during El Niño years both before and after 1968. Positive anomalies were also observed during La Niña years after 1968, but anomalies before 1968 were uniformly negative. When comparing rainfall departures before 1968 during the OND (-1) season, El Niño years were largely positive while La Niña years were largely negative. However, both El Niño and La Niña anomalies changed sign after the 1968 rainfall regime change, resulting in negative anomalies during El Niño years but positive anomalies for La Niña, although there were some inconsistencies among the regions. In general, a comparison of AMJ standard departures yielded positive anomalies during El Niño years but a negative rainfall response for La Niña years. However, it should be noted that the signal was not uniform across the Sahel regions, particularly during La Niña years. Additionally, during El Niño years most Sahel regions experienced anomalies of opposite sign between AMJ and JAS, but this phenomenon was not as consistent during La Niña years. Rainfall responses to ENSO across the Sahel were most consistent during the JAS and OND seasons, particularly after 1968. During JAS, negative rainfall anomalies were prevalent across the Sahel during El Niño years, but positive anomalies were largely present during La Niña years. Generally speaking, for both phases of ENSO the strongest anomalies were observed in regions further east (16/17/21 and 22) while standard departures in regions near the west coast and in the western Sahel were weaker (14/19, 9/13/18, and 23). This gradient of increasing departures from west-to-east during JAS potentially suggests that some maritime inﬂuences may moderate and/or mitigate ENSO impacts on rainfall closer to the Atlantic during the peak rainfall season. An interesting note is that during the JAS season, anomalies were opposite in sign when compared to the JAS (-1) season, with a more widespread contrast seen during El Niño years. This phenomenon could be due, in part, to the fact that many El Niño events tend to be either preceded or followed by La Niña events.

61 During OND, the season following the rainy season, an interesting and abrupt change occurred. In nearly every Sahel region analyzed, El Niño OND standard departures were of opposite sign than JAS: negative during JAS but positive during OND. Therefore, during El Niño events, rainfall across the Sahel and in transition zones was largely below normal in the peak of the rainy season, but above normal during the subsequent peak intensity of the El Niño events themselves. For La Niña years, this contrasting sign was also seen between JAS and OND, with positive anomalies during JAS but negative during the subsequent OND. However, this is only applied to the years after 1968. Before 1968, mostly positive anomalies persisted from JAS into OND. In summary, during El Niño years the Sahel typically experienced below normal rainfall during the peak rainy season of JAS, but above normal rainfall in the JAS (-1), AMJ, and OND seasons. On the other hand, La Niña years experienced above normal standard departures during JAS, but considerable differences were observed in the rainfall response to La Niña during the OND and JAS (-1) when comparing years pre- and post-1968, which will be discussed in sections below.

El Niño Years Before 1968 vs. After 1968. Before 1968, El Niño year JAS (-1) rainfall was well above normal in most Sahel regions, particularly in the central and east Sahel, where departures were 0.2, 0.1, and 0.2 in regions 14/19, 15/20, and 16/17/21 respectively. These positive anomalies persisted after the regime change, though the anomalies were notably weaker after 1968 in every region except 22, as evidenced by the bottom panel of Table 4.1. In OND (-1), a mixed signal was observed across the Sahel during El Niño years before 1968, though most anomalies were fairly weak. After 1968, however, every region saw a reduction in rainfall during El Niño years with intensiﬁed negative anomalies, particularly in the central Sahel region 15/20. In AMJ, most Sahel regions experienced weak positive anomalies during El Niño years before 1968, with the exception of coastal regions 9/13/18 and 23. After 1968, rainfall across every region except 22 increased, with the highest positive anomaly (0.2) again occurring in the central Sahel region 15/20. Interestingly enough, rainfall during the peak rainy season, JAS, was negative across all Sahel regions during El Niño years both before and after 1968, although it should be noted that

62 anomalies were further below average after 1968. With the exception of the coastal regions of 9/13/18 and 23, peak season rainfall after 1968 was less than it was before 1968 during El Niño years. This is evidenced by the third panel in Table 4.1, where regions 14/19, 15/20, 16/17/21, and 22 all had more negative standard departures after 1968 than before. Keeping in mind that the standard departure values have already been adjusted to mitigate the effect of the long-term drying signal to better show the seasonal ENSO impact on rainfall, this suggests that, in addition to the long-term drying trend and overall less rainfall after 1968, there was still less rainfall during the JAS season of El Niño years after 1968 versus before. Both before and after 1968, a general west-to-east gradient of increasingly dry conditions was observed. The OND season was remarkably similar to AMJ, particularly after 1968 when every region observed robust positive rainfall anomalies with the exception of the far east and geographically complex region 22 and a negligible change for region 16/17/21. Therefore, like JAS, there was an intensiﬁcation of the El Niño signal in the OND season after 1968. To summarize, there was an overall strengthening of the rainfall response to El Niño in the Sahel after the 1968 rainfall regime change. This is evidenced by a fairly uniform strengthening of negative anomalies in OND (-1), positive anomalies in AMJ, negative anomalies in JAS, and positive anomalies in OND. Finally, the El Niño signal was generally strongest in the central Sahel, leading to the conclusion that this region is most closely linked to El Niño. Although changes did occur in regions closer to the coast, they were not as robust as locations further inland.

La Niña Years Before 1968 vs. After 1968. As previously mentioned, most anomalies observed during the El Niño years were of opposite sign during La Niña years. There were also several important contrasts that were evident when comparing rainfall across the Sahel in La Niña years before 1968 versus after 1968. In this respect, changes of sign were more common during La Niña years, but the overall rainfall response appears to be of weaker magnitude using the established threshold of +/- 0.1. In the JAS (-1) season, rainfall anomalies before 1968 were uniformly negative across the Sahel, with particularly low rainfall departures in regions 9/13/18 and 14/19 of 0.2 and -0.3, respectively. After 1968, the sign of these anomalies changed to positive with very large changes ob-

63 served in region 14/19 (+0.4). This shows intense weakening of negative JAS (-1) rainfall anomalies after the regime change. A similar change occurred during OND (-1), where mostly negative anomalies before 1968 switched to positive afterwards with the exception of the far eastern regions. The magnitudes of the departures were mixed, but the strongest positive anomalies after 1968 were in the central Sahel region 15/20. Similar to El Niño years, the AMJ season during La Niña years yielded a fairly inconsistent signal across the regions. Positive departures for AMJ in La Niña years before 1968 were observed in the coastal Sahel regions, while negative departures were observed in the central and eastern Sahel. This was also observed after 1968, but some regions became more strongly positive while others became more strongly negative. According to the top panel of Table 4.2, JAS standard departures during La Niña years before 1968 varied among the regions analyzed. For example, there was little signal at all for region 9/13/18 with a departure very near zero. Regions 14/19 and 15/20 in the Sahel experienced negative JAS rainfall during La Niña years, and it was surprisingly even lower than it was for El Niño years in these regions. Meanwhile, regions further east (16/17/21 and 22) experienced positive JAS departures during La Niña years, as did region 23. After 1968, all regions experienced positive departures during JAS. This means that a fairly substantial increase in rainfall was seen during the maximum rainy season from La Niña years before 1968 to after, as evidenced by the third panel of Table 4.2. In fact, every region except region 23 saw an increase in JAS departures after the rainfall regime change, though the changes in regions 16/17/21 and 22 were quite small. Interestingly, although JAS rainfall as a whole was greater during La Niña years after 1968 than before, OND standard departures switched sign in most regions and were reduced in all regions except 15/20. This shows a fairly strong emergence of a negative rainfall response across the Sahel after 1968. Overall, evidence presented in Figures 4.1-4.6 and Table 4.2 suggests that there were stark contrasts in the rainfall response to La Niña across the Sahel before 1968 versus afterwards. Sign changes observed in JAS (-1), OND (-1) and OND, combined with a strengthening of the anomalies in JAS all indicate an increased inﬂuence of La Niña on Sahel rainfall after 1968. However, the

64 magnitudes of the standard departures were not as robust during the peak rainy season of La Niña years when compared to El Niño, especially before the regime change. When comparing the two time periods, the inﬂuence of La Niña tended to increase the most during JAS (-1) and OND.

4.1.2 Guinea Coast Region

As mentioned in Chapter 2, the Guinea Coast has a much different rainfall regime than regions further north. Therefore, it is treated separately in this study. This section will discuss differences between El Niño and La Niña year departures in the Guinea Coast, as well as compare anomalies both before and after 1968. The results of the seasonal progression of these anomalies are shown in Figure 4.7, while the calculated departures are shown in Tables 4.1 and 4.2. The majority of anomalies in the Guinea Coast are of opposite sign from the Sahel, which was expected.

El Niño vs. La Niña Years. Similar to the Sahel, several robust differences were observed when comparing the El Niño and La Niña signals in the Guinea Coast. During the JAS (-1) season, opposite anomalies were observed between El Niño and La Niña years. Before 1968, departures during El Niño were negative, while little signal was observed in La Niña years. After 1968, opposite anomalies were also observed between the two ENSO phases, with positive anomalies during El Niño but negative during La Niña. A similar pattern was observed in OND (-1), where positive anomalies in El Niño years before 1968 corresponded with negative anomalies during La Niña years. The opposite effect occurred after 1968, where there was little-to-no signal in OND (-1) during El Niño years, but a weak positive signal during La Niña years. For the JFM season before 1968, there was no signal during El Niño but a fairly weak positive departure observed in La Niña years. After 1968 robust signals were detected of positive rainfall anomalies during El Niño years but negative during La Niña years. The JAS season also differed between El Niño and La Niña years, with positive anomalies occurring before 1968 during El Niño, but weak negative departures occurring during La Niña. Once again, the opposite was true after 1968, when strong negative anomalies occurred in JAS during El Niño but equally as strong positive anomalies occurred during La Niña.

65 Rainfall during OND in the Guinea coast did not show as strong of a contrast between El Niño and La Niña years as other seasons. The Guinea Coast experienced positive anomalies during El Niño and La Niña years during OND, except for El Niño years before 1968. Similarities were also seen in JFM (+1), but this season experienced positive anomalies in every case except for El Niño years after 1968. Thus a contrast emerged between El Niño and La Niña years after 1968 for JFM(+1), but was weakened during OND. Overall, there were signiﬁcant differences between El Niño and La Niña years, and these differences were largely opposite of those found in the Sahel across most seasons. As alluded to in the above discussion, Guinea Coast rainfall anomalies experienced notable sign changes after 1968, and the contrast between the two time periods is quite robust. These contrasts will be discussed in the next section for El Niño years.

El Niño Years Before 1968 vs. After 1968. The top panel of Figure 4.7 shows the progression of standard departures during El Niño years in the Guinea Coast region 24/28/29. In the Guinea Coast, there is a clear discernible difference between the El Niño departures before and after 1968. The JAS (-1), AMJ, JAS, OND, and JFM (+1) seasons are all of opposite sign before 1968 versus after, and the differences are generally much larger than were observed in the Sahel. Before 1968, El Niño years brought standard departures slightly below normal in the Guinea Coast for AMJ, well above normal in JAS, and below normal in OND. The opposite occurred after 1968, which produced notably strong positive departures in AMJ, negative departures in JAS, and positive again in OND. During El Niño years before 1968, OND (-1) was positive while OND was negative. After 1968, OND (-1) did not produce a discernible signal, but OND became positive. The signal of JFM was only meaningful after 1968 (positive) and was of opposite sign than JFM (+1) (negative). All of these differences indicate that, like in the Sahel, the 1968 rainfall regime change marked a major change in the rainfall response to El Niño in the Guinea Coast. The changes were generally more pronounced for the Guinea Coast when compared to the Sahel during El Niño years, with many sign changes taking place between the two time periods.

La Niña Years Before 1968 vs. After 1968. The seasonal evolution of standard departures during La Niña years for the Guinea Coast are shown in the bottom panel of Figure 4.7. Changes in

66 the La Niña signal before versus after 1968 were observed, but they are of much smaller magnitude than for El Niño years. Nonetheless, there was a noticeable contrast between La Niña year rainfall in the Guinea Coast between the two time periods, as discussed in this section. In contrast to the the large difference observed in departures during AMJ of El Niño years before and after 1968, no such change was seen in AMJ of La Niña years, and there was no meaningful La Niña signal detected in this season or in JAS (-1). The largest departure from normal was seen in JAS, which received a positive departure value near 0.15 after 1968, versus no detectable signal before with rainfall near normal. Additional changes in sign were also observed in OND (-1), JFM, and OND, but they were fairly small in magnitude compared to El Niño years. Overall, changes in the ENSO rainfall response after 1968 versus before were largest and most evident in the Guinea Coast region, particularly during El Niño years. Although opposing signals were detected in La Niña years as well, the overall La Niña inﬂuence on Guinea Coast rainfall was observed to be much weaker than El Niño.

4.1.3 Sahel vs. Guinea Coast: Question of the Dipole

A ﬁnal comparison was made between the Sahel and Guinea Coast in the context of the aforementioned “dipole," which is largely believed to have weakened in recent years. In this study, strong contrasts between the Guinea Coast and Sahel were observed JAS (-1), AMJ, JAS, and OND during El Niño years before 1968, but in no cases after 1968. Additionally, a strong dipole was also present during La Niña years before 1968 during JAS (-1), JFM (+1), and AMJ (+1). After 1968, however, only JAS (-1) and OND showed a dipole signal. Keeping in mind that the anomalies during JAS (-1) were fairly week in the Guinea Coast, this project supports the idea that the Guinea Coast dipole has “disappeared" in recent years and suggests that recent changes in the Guinea Coast rainfall response to El Niño could be a factor in this disappearance, or vice versa.

4.2 Consistency of the ENSO Signal and Further Discussion

In addition to the analysis of rainfall standard departures during different ENSO phases for different time periods, the consistency of the ENSO signal was also analyzed. The purpose of this analysis was to compare the “expected" rainfall anomaly to the percentage of El Niño and La Niña

67 events from 1921-2012 that actually produced the expected anomaly. The expected anomaly sign for each region was determined from the average of departures during El Niño or La Niña years for the entire time period, 1921-2012. A threshold of +/- 0.1 was selected, and the percentages were plotted on maps for the JAS (-1), OND (-1), AMJ, JAS, and OND seasons shown in Figures 4.8-4.12, respectively. This value, although arbitrary, was selected based on the anomalies seen in Figures 4.1 to 4.8 and was deemed most appropriate among the many thresholds that were considered. Figure 4.8 shows the percent of JAS (-1) standard departures occurring during El Niño and La Niña years that were greater than +0.1 in regions where the expected anomaly was positive, or less than -0.1 in regions where the expected anomaly was negative. For this season, positive anomalies were expected in every region analyzed for El Niño years, while negative anomalies were expected in every region for La Niña years. The strongest percent of expected standard departures occurring during JAS (-1) of El Niño years was region 16/17/21, with 45-60% of El Niño events resulting in more rainfall than normal. There was a uniform consistency across much of the Sahel with percentages ranging from 30% to 45%. The Guinea Coast was not consistent, with only 15-30% of El Niño events bringing more rainfall than normal. For La Niña years, negative anomalies were expected in every region analyzed. The consistency percentages were higher compared to El Niño years, with 45% to 60% of events producing below normal rainfall across much of the Sahel (and even higher percentages along the west coast). JAS (-1) standard departures for the Guinea Coast were slightly higher during La Niña years versus El Niño years, but the overall consistency remained lower than in the Sahel. Figure 4.9 is the same as the previous ﬁgure, but for the OND (-1) season. During El Niño years, a distinct dipole was seen between the Sahel and Guinea Coast. Anomalies across all Sahel regions were expected to be negative, while the Guinea Coast was positive. Additionally, the percentages were fairly consistent, with most Sahel regions experiencing a negative rainfall anomaly 45-60% of the time during El Niño OND (-1). Consistency was only slightly lower in the Guinea Coast, at 30-45%. In La Niña years, a dipole was also observed, but only between the central regions of the Sahel and the Guinea Coast. All regions were expected to experience a negative anomaly except the central Sahel regions 14/19 and 15/20. Consistency was lower in

68 Sahel regions, but stronger in the Guinea Coast region, which experienced negative anomalies 45-60% of the time. Figure 4.10 shows the consistencies for the AMJ season. In this case, the expected anomalies were fairly inconsistent from region-to-region and the percentages remain fairly low. For El Niño years, the coastal region 9/13/18 experienced its expected negative anomaly during roughly half of El Niño events. On the other hand, regions 14/19, 16/17/21, and 23 experienced their expected positive anomaly roughly half of the time as well. The remaining regions showed lower consistency among El Niño years, with 30-45% of events resulting in the expected anomalies. For La Niña years, a robust percentage of events showed negative rainfall in the central Sahel region of 15/20, with a slightly lower percentage in region 16/17/21 and 22. Rainfall anomalies in Sahel regions further east showed less consistency during AMJ of La Niña years, and the Guinea Coast showed very little relationship with La Niña. Figure 4.11 is among the most interesting among the spatial consistency plots, not only because it is the peak rainy season in the Sahel but also because it produces a uniform expected rainfall anomaly across all regions during El Niño years. The top panel of the ﬁgure shows that, in every region, a negative anomaly is expected during El Niño events, and this anomaly occurred quite often in the central 15/20 region, with a percentage of 60-75%. The remaining Sahel regions experienced negative rainfall anomalies during 45-60% of El NIño years. Transition region 23, the Guinea Coast, and far east region 22 showed the lowest percentages of expected anomaly outcomes. Note that this ﬁgure shows anomalies and percentages that counter those of the JAS (-1) season shown in Figure 4.8. In La Niña years, the expected anomalies are largely opposite from El Niño years, with the exception of regions 14/19 and 15/20. In fact, there is no difference between the percentage or anomaly expected in El Niño years and La Niña years for region 14/19. The percentages for La Niña are higher in the east, but overall the La Niña percentages are again not very robust. Like the El Niño results, the La Niña years shown in Figure 4.11 also largely counter those for the JAS (-1) season in Figure 4.8, with the exception of the central Sahel. Finally, the OND percentages for El Niño and La Niña are shown in Figure 4.12. During El Niño years, a clear distinction was evident compared to JAS, and OND (-1). Whereas the expected anomaly in JAS was negative for all regions, it was positive for all regions during OND, largely

69 opposite from the OND (-1) season in Figure 4.9 as well. Here, the percentages were lower in most regions during El Niño than with JAS, which was not too surprising given inconsistencies in the rainfall across much of the Sahel during this season. The highest occurrence of the expected anomaly during El Niño years was in region 14/19, which was in the 60-75% range. Neighboring regions were much lower, however, in the 30-45% range. For La Niña years, a robust occurrence of the expected anomaly occurred in both regions 14/19 and 15/20. In OND these regions experienced rainfall that was below normal during 60-75% of La Niña events. Regions 9/13/18, 16/17/21, and 22 also experienced negative rainfall, but at a slightly lower percentage, ranging from 45-60%. The Guinea Coast, which was expected to receive more rainfall than average during OND of La Niña years, experienced the anomaly roughly 30-45% of the time. When compared to the La Niña years of OND (-1) in Figure 4.9, the signs of expected anomalies is opposite for both the Guinea Coast and central Sahel, with the Sahel consistency higher during OND, but Guinea Coast consistency higher during OND (-1). There are several important conclusions that can be drawn from Figures 4.8-4.12. First, the ﬁgures show striking contrasts between the JAS (-1) and JAS seasons, as well as the OND (-1) and OND seasons. Additionally, there is very little consistency among the rainfall response to ENSO during AMJ seasons, which is not too surprising given the inconsistencies seen in the magnitude of the departures during this season as well. The strongest teleconnections between El Niño and rainfall were in the central Sahel during the peak rainy season, and a slightly lower but fairly consistent response was also seen during JAS (-1), OND (-1), and OND. La Niña showed the strongest relationship in the Sahel during JAS (-1) and OND, where a robust number of events produced the expected anomaly. These consistency results suggest that ENSO has only a marginal impact on Sahel and Guinea Coast rainfall since the expected anomalies during ENSO events generally only occurred a small percentage of the time. Additionally, the threshold of +/-0.1 further indicates a relatively weak relationship, since this is quite a small threshold. Finally, Figures 4.8- 12 imply that although there is a relatively inconsistent relationship between Sahel and Guinea Coast rainfall overall, any teleconnection between ENSO and West African rainfall is most likely to be observed in the Sahel (especially central Sahel) during the JAS (-1), OND (-1), JAS, and OND seasons. It is important to note that although a negative (positive) anomaly during El Niño

70 years and a positive (negative) anomaly during La Niña years in JAS (OND) was experienced a fairly high percentage of the time, the actual anomalies themselves were not so strong, as shown in Tables 4.1 and 4.2. The strongest anomalies among these seasons occurred in JAS (-1), suggesting that this season has the greatest potential to bring above (below) normal rainfall before El Niño (La Niña) years.

4.3 Atmospheric Circulation Patterns and ENSO

After the rainfall response to ENSO had been analyzed for the various regions throughout the Sahel and Guinea Coast, atmospheric circulation patterns were examined. Specifically, zonal and meridional wind cross sections during El Niño and La Niña years before and after 1968 were plotted and compared to those of non-ENSO years. This was done along zero degrees longitude, chosen because it crosses both a true Sahel region as well as the Guinea Coast. Specifically, it bisects regions 14/19 and 24/28/29. Figures 4.13-4.17 show the zonal wind cross sections before and after 1968 for both phases of ENSO and during non-ENSO years for the JAS (-1), OND (-1), AMJ, JAS, and OND seasons. Note that because of data limitations, the plots were made beginning at 1948 instead of 1921, which eliminated the El Niño years of 1923, 1925, 1930, and 1945 and the La Niña years of 1924, 1933, 1938, and 1942 from the analysis. Although few of the differences were pronounced, there were some notable observations from the zonal wind analysis. During the JAS(-1) season before 1968, the TEJ and the southwest monsoon flow were stronger in El Niño years than the mean or La Niña years. This is very consistent with an overall wet rainfall pattern in the Sahel. After 1968, however, no differences in the circulation regime were observed when compared to the mean or La Niña years. For OND (-1) before 1968 in El Niño years, the southwest monsoon flow was stronger than La Niña years but not stronger than average. After 1968, the the AEJ was stronger in El Niño years than both La Niña years and non-ENSO years. The AMJ season showed no discernible differences between any of the years both before and after 1968.

71 In JAS, the TEJ of El Niño years was weaker than average but was the same as average during La Niña years before 1968, which is consistent with the rainfall anomalies observed during this season. During post-1968 El Niño years, the AEJ was stronger than both La Nina years and the average. In this case, the TEJ was stronger in La Niña years than during El Niño, also consistent with the observed rainfall anomalies. Finally, in the OND season, El Niño years before 1968 saw a stronger than average TEJ but La Niña TEJ was weaker. This was, again, consistent with the observed rainfall anomalies during OND. Additionally, the southwest monsoon ﬂow was also somewhat weaker. After 1968, the AEJ was weaker than average for both El Niño and La Niña years. Overall, the differences between the zonal wind cross sections are quite small. One potential explanation for the lack of many differences between the El Niño years and the non-ENSO years could be that there were a large number of dry years that were not associated with ENSO. Fur- thermore, there were also few distinguishable differences between La Niña years and non-ENSO years, although magnitudes of the low-level westerly jet during La Niña years before 1968 tended to be lower. Despite these small differences between El Niño and La Niña circulation patterns compared to non-ENSO years, the largest differences were detected when comparing years before 1968 to after 1968. In nearly every case, low level westerly winds were of lower magnitude than prior to 1968. Because this change was also observed in the non-ENSO years, however, it suggests that the phenomenon was not unique to El Niño or La Niña. Finally, the results of the zonal wind analysis provide some consistency with the observed rainfall anomalies discussed in sections 4.1 and 4.2, particularly during JAS (-1), JAS, and OND (-1). A similar comparison was done using meridional winds, which also play a role in rainfall variability across the regions. These results, which are not included here, showed similar patterns to the zonal winds– very few differences between the El Niño and La Niña years when compared to non-ENSO years, most likely due to the large number of non-ENSO dry years. Like the zonal wind plots, there were some small discernible differences between the years before and after 1968.

72 El Niño Years: Region 9/13/18 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

La Niña Years: Region 9/13/18 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

Figure 4.1: Seasonal evolution of standard departures during (top) El Niño years before and after 1968 and (bottom) La Niña years before and after 1968 for combined Region 9/13/18. El Niño years include 1923, 1925, 1930, 1945, 1951, 1953, 1963, 1965, 1968, 1969, 1972, 1976, 1982, 1991, 1994, 1997, 2002, 2004, and 2012. La Niña years include 1924, 1933, 1938, 1942, 1949, 1964, 1967, 1970, 1973, 1975, 1984, 1988, 1995, 2007, 2010, and 2011.

73 El Niño Years: Region 14/19 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20 Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

La Niña Years: Region 14/19 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Stnadardized Departure RainfallStnadardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

Figure 4.2: Same as Figure 4.1, but for combined Region 14/19.

74 El Niño Years: Region 15/20 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

La Niña Years: Region 15/20 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

Figure 4.3: Same as Figure 4.1, but for combined Region 15/20.

75 El Niño Years: Region 16/17/21 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

La Niña Years: Region 16/17/21 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

Figure 4.4: Same as Figure 4.1, but for combined Region 16/17/21.

76 El Niño Years: Region 22 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20 Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

La Niña Years: Region 22 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

Figure 4.5: Same as Figure 4.1, but for Region 22.

77 El Niño Years: Region 23 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20 Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

La Niña Years: Region 23 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

Figure 4.6: Same as Figure 4.1, but for Region 23.

78 El Niño Years: Region 24/28/29 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20 Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

La Niña Years: Region 24/28/29 0.25 0.20 0.15 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20

Standardized Departure RainfallStandardized Departure -0.25 JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) Before 1968 After 1968

Figure 4.7: Same as Figure 4.1, but for combined Region 24/28/29.

79 Table 4.1: Seasonal evolution of standard departures during El Niño years. Negative values are highlighted in red; positive values are highlighted in blue.

Seasonal Standardized Rainfall Departures during El Niño Years Before 1968 El Niño Years Included: 1923, 1925, 1930, 1945, 1951, 1953, 1963, 1965

Region JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) 9/13/18 0.0985 -0.0055 0.0453 -0.1192 -0.0676 0.0840 0.1902 0.1016 14/19 0.1975 0.0697 -0.0685 0.1469 -0.0325 0.0997 0.0171 -0.0321 15/20 0.1392 0.0770 -0.0295 0.0428 -0.0993 0.1051 0.0228 0.1093 16/17/21 0.2457 -0.0188 -0.0275 0.0441 -0.1028 0.0985 0.0388 -0.0939 22 -0.0063 -0.0534 0.0036 0.0125 -0.1085 0.0705 0.0480 0.0559 23 0.1266 0.0408 0.1010 -0.0516 -0.0756 0.0350 0.0350 0.0898 24/28/29 -0.0594 0.0910 0.0145 -0.0528 0.1499 -0.0800 0.0325 0.0211

Seasonal Standardized Rainfall Departures during El Niño Years After 1968 El Niño Years Included: 1968, 1969, 1972, 1976, 1982, 1991, 1994, 1997, 2002, 2004, 2012

Region JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) 9/13/18 0.0855 -0.0888 0.0878 0.0137 -0.0351 0.1686 0.0513 0.0426 14/19 0.0471 -0.0786 0.1124 0.1765 -0.0735 0.2165 -0.0111 0.0428 15/20 -0.0431 -0.1070 0.0105 0.1983 -0.1445 0.1688 -0.0130 -0.0382 16/17/21 0.0116 -0.0600 -0.0052 0.0485 -0.1289 0.0972 0.0146 -0.0401 22 0.0912 -0.0737 0.0896 -0.0536 -0.1993 -0.0062 0.0617 -0.1276 23 0.0347 -0.0167 0.1680 0.1605 -0.0624 0.2268 0.0041 0.0285 24/28/29 0.1073 -0.0197 0.1965 0.1791 -0.1246 0.1513 -0.1445 -0.0489

Differences between Seasonal Standardized Rainfall Departures during El Niño Years After 1968 (1968-2012) minus Before 1968 (1921-1967)

Region JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) 9/13/18 -0.0130 -0.0833 0.0425 0.1329 0.0325 0.0846 -0.1389 -0.0590 14/19 -0.1505 -0.1483 0.1809 0.0296 -0.0410 0.1168 -0.0281 0.0749 15/20 -0.1823 -0.1840 0.0400 0.1556 -0.0452 0.0637 -0.0359 -0.1475 16/17/21 -0.2341 -0.0412 0.0223 0.0044 -0.0262 -0.0012 -0.0242 0.0539 22 0.0975 -0.0204 0.0860 -0.0661 -0.0908 -0.0768 0.0137 -0.1836 23 -0.0919 -0.0574 0.0670 0.2121 0.0132 0.1918 -0.0310 -0.0613 24/28/29 0.1667 -0.1107 0.1821 0.2318 -0.2745 0.2313 -0.1770 -0.0701

80 Table 4.2: Seasonal evolution of standard departures during La Niña years. Negative values are highlighted in red; positive values are highlighted in blue.

Seasonal Standardized Rainfall Departures during La Niña Years Before 1968 La Niña Years Included: 1924, 1933, 1938, 1942, 1949, 1964, 1967

Region JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) 9/13/18 -0.2153 -0.1063 0.2332 0.0105 0.0038 0.0642 0.0041 -0.0761 14/19 -0.2995 -0.0532 0.1460 0.0587 -0.1205 -0.0120 0.2031 -0.0650 15/20 -0.1944 0.0358 0.1048 -0.0753 -0.1415 -0.1564 0.0043 0.1236 16/17/21 -0.1368 -0.0357 0.0792 -0.0891 0.0683 0.0095 -0.0381 0.0959 22 -0.0570 -0.0859 0.2289 -0.0514 0.0836 0.0061 -0.1973 -0.2234 23 -0.0333 -0.1240 0.1436 0.0868 0.0833 0.1009 -0.0064 -0.1504 24/28/29 0.0183 -0.1314 0.0327 -0.0161 -0.0216 0.0946 0.0547 0.0570

Seasonal Standardized Rainfall Departures during La Niña Years After 1968 La Niña Years Included: 1970, 1973, 1975, 1984, 1988, 1995, 2007, 2010, 2011

Region JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) 9/13/18 0.0739 0.0038 -0.0999 0.0852 0.0883 -0.1439 0.0060 0.2163 14/19 0.1179 0.0780 -0.0969 -0.0452 0.0367 -0.1168 -0.0026 -0.2069 15/20 0.0163 0.0986 -0.0282 -0.0913 0.0155 -0.0765 0.0136 -0.0433 16/17/21 -0.0370 -0.0849 0.0055 -0.0129 0.0871 -0.0859 0.0217 0.0940 22 0.0197 -0.1253 0.0533 0.0053 0.0915 -0.1651 0.0087 0.1281 23 -0.0309 0.0366 -0.0792 0.0940 0.0306 -0.0802 0.0429 0.0350 24/28/29 -0.0384 0.0369 -0.0812 -0.0299 0.1467 0.0377 0.0801 -0.0228

Differences between Seasonal Standardized Rainfall Departures during La Niña Years After 1968 (1968-2012) minus Before 1968 (1921-1967)

Region JAS (-1) OND (-1) JFM AMJ JAS OND JFM (+1) AMJ (+1) 9/13/18 0.2892 0.1101 -0.3331 0.0747 0.0845 -0.2081 0.0019 0.2924 14/19 0.4174 0.1312 -0.2429 -0.1039 0.1572 -0.1049 -0.2057 -0.1419 15/20 0.2107 0.0628 -0.1330 -0.0159 0.1570 0.0799 0.0093 -0.1669 16/17/21 0.0998 -0.0492 -0.0737 0.0763 0.0187 -0.0954 0.0598 -0.0018 22 0.0768 -0.0394 -0.1756 0.0567 0.0079 -0.1713 0.2060 0.3515 23 0.0024 0.1607 -0.2228 0.0072 -0.0527 -0.1811 0.0493 0.1854 24/28/29 -0.0567 0.1682 -0.1139 -0.0137 0.1683 -0.0569 0.0254 -0.0798

81 El Niño Years: JAS (-1)

La Niña Years: JAS (-1)

75% 60% 45% 30% 15% 0% 0% 15% 30% 45% 60% 75% Negative Anomaly Expected Positive Anomaly Expected Percent of SDP < -0.1 Percent of SDP > 0.1

Figure 4.8: Percent of JAS (-1) standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012.

82 El Niño Years: OND (-1)

La Niña Years: OND (-1)

75% 60% 45% 30% 15% 0% 0% 15% 30% 45% 60% 75% Negative Anomaly Expected Positive Anomaly Expected Percent of SDP < -0.1 Percent of SDP > 0.1

Figure 4.9: Percent of OND (-1) standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012.

83 El Niño Years: AMJ

La Niña Years: AMJ

75% 60% 45% 30% 15% 0% 0% 15% 30% 45% 60% 75% Negative Anomaly Expected Positive Anomaly Expected Percent of SDP < -0.1 Percent of SDP > 0.1

Figure 4.10: Percent of AMJ standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012.

84 El Niño Years: JAS

La Niña Years: JAS

75% 60% 45% 30% 15% 0% 0% 15% 30% 45% 60% 75% Negative Anomaly Expected Positive Anomaly Expected Percent of SDP < -0.1 Percent of SDP > 0.1

Figure 4.11: Percent of JAS standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012.

85 El Niño Years: OND

La Niña Years: OND

75% 60% 45% 30% 15% 0% 0% 15% 30% 45% 60% 75% Negative Anomaly Expected Positive Anomaly Expected Percent of SDP < -0.1 Percent of SDP > 0.1

Figure 4.12: Percent of OND standard departures (SDP) occurring during El Niño years (top) and La Niña years (bottom) from 1921-2012 that are greater than +0.1 in regions where the expected anomaly is positive (blue color bar), or less than -0.1 in regions where the expected anomaly is negative (red color bar). The expected anomaly sign for each region was determined from the average of departures during all El Niño or La Niña years 1921-2012.

86 Average JAS (-1) Zonal Winds (m/s)

Figure 4.13: Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the JAS (-1) season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis. 87 Average OND (-1) Zonal Winds (m/s)

Figure 4.14: Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the OND (-1) season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis. 88 Average AMJ Zonal Winds (m/s)

Figure 4.15: Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the AMJ season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis. 89 Average JAS Zonal Winds (m/s)

Figure 4.16: Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the JAS season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis. 90 Average OND Zonal Winds (m/s)

Figure 4.17: Cross section composites of zonal winds (m/s) along the Prime Meridian during El Niño years (top row), La Niña years (middle row), and non-ENSO years (bottom row) before and after 1968 during the OND season. Note that the dataset begins at 1948, so years before this have been eliminated from the analysis. 91 CHAPTER 5

CONCLUDING REMARKS

5.1 Conclusions

The objective of this project was to shed light on the much debated relationship between the El Niño/Southern Oscillation climate cycle and rainfall variability in the West African regions of the Sahel and Guinea Coast. There have been several previous studies (Nicholson and Entekhabi 1986; Ropelewski and Halpert 1987, 1989; Nicholson and Kim 1997; Nicholson et al. 2001) which have found little connection between Sahel rainfall and ENSO phase, while others (such as Hastenrath et al. 1987; Wolter 1989; Ward 1992; and Palmer et al. 1992) have suggested that ENSO can, in fact, result in dynamical changes within the monsoon circulation and cause a reduction in Sahel rainfall during El Niño years. This project aimed to provide the most in-depth and comprehensive analysis to date of rainfall variability across the region during El Niño and La Niña events over a record of 91 years using data collected from an extensive network of gauge stations. The seasonal cycle of rainfall anomalies during different phases of El Niño and La Niña events in numerous rainfall regions both before and after the well-documented rainfall regime change that occurred in the Sahel around the year 1968 was analyzed. Additionally, comparisons were made between regions by analyzing differences between El Niño and La Niña years, as well as anomalies before and after 1968. The consistency of the ENSO signal among the various rainfall regions was examined to determine how often expected anomalies produced in the standard departure analysis actually occurred. Finally, zonal (and meridional) wind circulation cross sections were created to compare the atmospheric circulation regime of West Africa during El Niño, La Niña, and non-ENSO years both before and after 1968. In order to achieve the objectives outlined above, a list of El Niño and La Niña events from 1921-2012 was complied using SST anomalies in the Niño 3.4 region using the same methodology

92 currently employed by the NOAA Climate Prediction Center (CPC). This resulting list of years, shown in Table 3.4, was used as the basis for all analysis encompassing the scope of this study. The results of the rainfall analysis during El Niño and La Niña years both before and after 1968 are discussed in Chapter 4.1. Before 1968, the analysis showed that El Niño year rainfall was below normal during the peak rainy season (JAS) for all regions analyzed except the Guinea Coast, which experienced above normal rainfall. In AMJ and OND, rainfall anomalies were generally of opposite sign than during JAS. The departures observed in the Sahel increased in magnitude from west-to-east, with areas further from the coast being drier during the peak rainfall season of El Niña years. The JAS (-1) and OND (-1) seasons also produced robust signals, especially in the Sahel. After 1968, the negative anomalies observed in JAS season of the Sahel persisted, but were further below normal than before 1968 suggesting an increased response in Sahel rainfall to ENSO after the regime change. Coastal regions of the Sahel were not changed as much, and it is hypothesized that this is due to the addition of marine influences not seen in regions further east. The biggest post-1968 change was observed in the Guinea Coast, however, supporting evidence found in previous studies regarding the disappearance of the Guinea Coast “dipole" post-1968. During La Niña years, rainfall departures were more varied than during El Niño years, particularly before 1968. After 1968, all regions experienced positive anomalies during JAS, however, suggesting a potential increase in the La Niña influence post-1968 and a substantial increase in rainfall during the maximum rainy season following the change in the rainfall regime during La Niña events. There was an abrupt change in sign of JAS (-1), OND (-1), and OND rainfall anomalies during La Niña years before/after 1968. This finding is consistent with the SST analysis performed by Nicholson et al. (in press) and shown in Figures 2.10 and 2.11. The consistency analysis described in Chapter 4.2 provided further information regarding the rainfall response to ENSO. Very little consistency among the rainfall response to ENSO was found during the dry season, which was expected based on the erratic nature of rainfall events during those seasons. The strongest consistency of the El Niño signal was found to be in the central Sahel regions during the peak rainy season, suggesting that this region has the highest chance of receiving below normal rainfall during JAS of El Niño years when compared to other regions closer to the coast. Interestingly, the La Niña consistency was also highest in the central

93 Sahel, but during the OND season, when rainfall also tended to be below normal. The JAS (-1) and OND (-1) seasons also produced fairly consistent results of opposite sign when compared to JAS and OND. Altogether, this study concludes that the overall rainfall response to El Niño and La Niña events in the Sahel and Guinea Coast as a whole is relatively weak, but there is some connection between ENSO and rainfall in the Sahel during the JAS (-1), JAS, OND (-1), and OND seasons. This relationship intensified after the 1968 rainfall regime change, bringing stronger expected anomalies in those respective seasons. The analysis of upper level zonal and meridional winds further supports the idea of a weak relationship between ENSO and Sahel and Guinea Coast rainfall, but with a slight intensification seen after 1968. There were only a few differences found between El Niño, La Niña, and non- ENSO years, but changes were noted when comparing pre/post 1968. A possible explanation for this observation is the occurrence of many dry years that were not considered El Niño, particularly after 1968. The conclusions reached in this study agree with those of many previous studies. For example, they are consistent with findings by Losada et al. (2012) who suggested a change in the relationship between Sahel rainfall and SSTs after the 1970s. Additionally, it is consistent with finings by Janicot el al. (1996), which found higher correlations between Sahel rainfall and El Niño and Pacific SSTs after 1970 versus before. It is also supports the SST analysis performed by Nicholson et al. (in press) that found an increased relationship between central Pacific SSTs and Sahel rainfall after 1968. Perhaps most conclusively, this study agrees with those of Parhi et al. (2016), which found that El Niño has the potential to produce a drier climatic response over the Sahel, but the occasional extreme rainfall can still occur during El Niño years (and thus the overall relationship remains weak). To add to this point, this study identified specific seasons in which this is the case in addition to the seasons when opposite anomalies are likely to occur during La Niña. It also strongly agrees with their implication that detecting the ENSO signal in rainfall variability over West Africa is difficult, due to a multitude of other climatic factors a play, among them the North American Oscillation, the Atlantic Multidecadal Oscillation, the Indian Ocean dipole, the AEJ, the TEJ, and the SHL. Some studies have also hypothesized that the relationship between

94 ENSO and rainfall variability in the Sahel and Guinea Coast is an indirect one, which potentially explains the relatively low consistency percentages observed in Figures 4.8-4.12.the

5.2 Implications

The conclusions reached in this study suggest that ENSO phase alone has limited applications in making seasonal rainfall predictions in West Africa, but is most useful when predicting seasonal rainfall in certain regions of the Sahel during JAS (-1), OND (-1), JAS and OND. Due to the multitude of climatic factors at play, ENSO phase should be considered along with other factors when predicting Sahel and Guinea Coast rainfall on seasonal time scales. This study did conclude that there was some intensiﬁcation of the signal after 1968, which marked a change in the overall rainfall regime. This suggests that despite an overall weak relationship, there is a possible change in response to ENSO in Sahel and Guinea Coast rainfall during major regime changes, such as the one that caused a severe drought that persisted throughout the latter portion of the twentieth century.

5.3 Future Work

There are many remaining questions regarding the ENSO-Sahel/Guinea Coast teleconnection. Future work could expand on the atmospheric circulation analysis that was performed herein by analyzing other thermodynamic variables both before and after 1968, especially since few differences were observed in the zonal and meridional winds. Obviously, a major limitation to this type of analysis is the availability of reliable data before 1968. However, there are certain datasets, such as the NCEP/NCAR Reanalysis dataset utilized in this study, that do extend back prior to this time period. Another potentially enlightening expansion of this study could include the analysis of case studies. This project did not separate El Niño and La Niña events by intensity, instead averaging them all together. It is well-established that certain El Niño and La Niña events are stronger than others, and therefore it follows that certain events may have much more impact on rainfall in the Sahel and Guinea Coast than others as well. It would be interesting to look at the strongest

95 few ENSO events to see if there are large differences than when all ENSO events are considered together, or when compared to relatively weak events. This could provide some important distinctions between weak and strong ENSO events, and could be potentially useful when creating seasonal rainfall outlooks for the region. Additionally, differences between El Niño events and so- called El Niño Modoki events could be analyzed to detect any distinctions in the rainfall regime. El Niño Modoki events are characterized by a different SST pattern than regular El Niño, with anomalous warming in the central Pacific but cooling in the eastern and western tropical Pacific (Ashok et al. 2007). Finally, the consistency analysis conducted in this study was performed for all years, 1921- 2012. The findings from the rainfall departure calculations suggest that there was indeed an intensification of the ENSO-Sahel/Guinea Coast teleconnection after 1968, but this was not accounted for in Figures 4.8-4.12. Future work could separate the consistency percentages for years before and after 1968 to see if not only the standard departures were affected, but if there was a change in the consistency of the ENSO signal after 1968 as well. It is hypothesized here that a change in consistency would not be as obvious as one may think, due to the plethora of dry years after 1968 that were non-ENSO years. However, the analysis could be insightful nonetheless. Overall, this study provides useful information regarding the teleconnections between ENSO and rainfall variability throughout the Sahel and Guinea Coast. While no study is all-encompassing, this research does add a great deal of insight to regional responses to ENSO events, accomplishes its goals, and answers the questions outlined in Chapter 2. This work is but a small part of the much larger pursuit of a better understanding of the complexities of Sahel and Guinea Coast rainfall variability, which will continue in the form of future research.

96 REFERENCES

Ali, A. and Lebel, T., 2009. The Sahelian standardized rainfall index revisited. International Journal of Climatology, 29, 1705-1714.

Ashok, K., Behera, S.K., Rao, S.A., Weng, H., and Yamagata, T., 2007. El Niño Modoki and its possible teleconnection. Journal of Geophysical Research, 112, doi:10.1029/2006JC003798.

Bain, C.L., Parker, D.J., Taylor, C.M., Kergoat, L., and Guichard, F., 2010. Observations of the nocturnal boundary layer associated with the West African monsoon. Monthly Weather Re- view, 138, 3142-3156.

Chauvin, F., Roehrig, R., and Lafore, J., 2010. Intraseasonal variability of the Saharan Heat Low and its links with midlatitudes. Journal of Climate, 23, 2544-2561.

Comité Inter-états de Lutte contre la Sécheresse dans le Sahel (CILSS), 2016. Landscapes of West Africa – A Window on a Changing World. U.S. Geological Survey Earth Resources Observation and Science Center, Garretson, SD, pp. 236.

Diatta, S. and Fink, A.H., 2014. Statistical relationship between remote climate indices and West African monsoon variability. International Journal of Climatology, 34, 3348-3367.

Flohn, H., 1964. Investigations on the Tropical Easterly Jet. Bonner Meteorologische Abhandlun- gen, 4, 83.

Fu, C., Diaz, H.F., and Fletcher, J.O., 1986. Characteristics of the response of sea surface temperature in the Central Paciﬁc associated with warm episodes of the Southern Oscillation. Monthly Weather Review, 114, 1716-1738.

Gergis, J.L. and Fowler, A.M., 2009. A history of ENSO events since A.D. 1525: implications for future climate change. Climatic Change, 92, 343-387.

Grist, J.P. and Nicholson, S.E., 2001. A study of the dynamic factors inﬂuencing rainfall variability in the West African Sahel. Journal of Climate, 14, 1337-1359.

Grodsky, S., Carton, J., and Nigam, S., 2003. Near surface westerly wind jet in the Atlantic ITCZ. Geophysical Research Letters, 30 OCE1.

Gu, G., and Adler, R.F., 2004. Seasonal evolution and variability associated with the West African monsoon system. Journal of Climate, 17, 3364-3377.

97 Hagos, S., and Cook, K., 2007. Dynamics of the West African monsoon jump. Journal of Climate, 20 5264-5284.

Halpert, M.S. and Ropelewski, C.F., 1992. Surface temperature patterns associated with the South- ern Oscillation. Journal of Climate, 5, 577-593.

Hastenrath, S., de Castro, L.C., and Aceituno, P., The Southern Oscillation in the tropical Atlantic sector. Contributions to Atmospheric Physics, 60, 477-462.

Huang, B., Banzon, V.F., Freeman, E., Lawrimore, J., Liu, W., Peterson, T.C., Smith, T.M., Thorne, P.W., Woodruff, S.D., and Zhang, H., 2015: Extended Reconstructed Sea Surface Temperature (ERSST), Version 4. NOAA National Centers for Environmental Information. doi:10.7289/V5KD1VVf

Janicot, S., Moron, V., and Fontaine, B., 1996. Sahel droughts and ENSO dynamics. Rowell, D.P., 2001. Geophysical Research Letters, 23, 515-518.

Janicot S., Mounier F., Gervois S., Sultan B., and Kiladis G., 2010. The dynamics of the West African monsoon. Part V: The role of convectively coupled Rossby waves. Journal of Cli- mate, 23, 4005-4024.

Janicot S., Mounier F., Hall N., Leroux S., Sultan B., Kiladis G., 2009. The West African monsoon dynamics. Part IV: Analysis of 25-90 day variability of convection and the role of the Indian monsoon. Journal of Climate, 22, 1541-1565.

Janicot, S., Trzaska, S., and Poccard, I., 2001. Summer Sahel-ENSO teleconnection and decadal time scale SST variations. Climate Dynamics, 18, 303-320.

Kaplan, A., Cane, M.A., Kushnir, Y., Clement, A.C., Blumenthal, M.B., and Rajagopalan, B., 1998. Analyses of global sea surface temperature 1856-1991. Journal of Geophysical Re- search, 103, 18567-18589.

Knippertz, P. and Fink, A., 2008. Dry-season precipitation in tropical West Africa and its relation to forcing from the extratropics. Monthly Weather Review, 136, 3579-3596.

Knippertz, P., Fink, A., Reiner, A., and Speth, P., 2003. Three late summer/early autumn cases of tropical-extratropical interactions causing precipitation in northwest Africa. Monthly Weather Review, 131, 116-135.

L’Hote, Y., Mahè, G., and Some, B., 2003. The 1990s rainfall in the Sahel: the third driest decade since the beginning of the century. Hydrological Sciences Journal, 48:3, 493-496.

98 L’Hote, Y., Mahè, G., Some, B., and Triboulet, J., 2002. Analysis of a Sahelian annual rainfall index from 1896 to 2000; the drought continues. Hydrological Sciences Journal, 47:4, 563- 572.

Lavaysse, C., Chaboureau, J., and Flamant, C., 2011. Dust impact on the West African heat low in summertime. Quarterly Journal of the Royal Meteorological Society, 137, 1277-1240.

Lavaysse, C., Flamant, C., Janicot, S., Parker, D.J., Lafore, J., Sultan, B., and Pelon, J., 2009. Seasonal evolution of the West African heat low: a climatological perspective. Climate Dynamics, 33 313-330.

Lebel, T. and Ali, A., 2009. Recent trends in the Central and Western Sahel rainfall regime (1990- 2007). Journal of Hydrology, 375, 52-64.

Lebel, T., Diedhiou, A., and Laurent, H., 2003. Seasonal cycle and interannual variability of the Sahelian rainfall at hydrological scales. Journal of Geophysical Research, 108, CIP14.

Lindsey, R., 2013. In watching for El Niño and La Niña, NOAA adapts to global warming. Accessed 12 January 2017. [Available online at https://www.climate.gov/news- features/understanding-climate/watching-el-niño-and-la-niña-noaa-adapts-global- warming].

Losada, T., Rodriguez-Fonseca, B., Mohino, E., Bader, J., Janicot, S., and Mechoso, C.R., 2012. Tropical SST and Sahel rainfall: A non-stationary relationship. Geophysical Research Let- ters, 39, L12705.

Lutgens, F.K. and Tarbuck, E.J., 2010. The Atmosphere: An Introduction to Meteorology, 11th edition. Prentice Hall.

Mahè, G. and Paturel, J., 2009. 1896-2006 Sahelian annual rainfall variability and runoff increase of Sahelian rivers. Comptes Rendus Geoscience, 341, 538-546.

Maloney, E.D., and Shaman, J., 2008. Journal of Climate, 21, 2898-2918.

Mathon, V., Laurent, H., and Lebel, T., 2002. A combined barotropic-baroclinic instability study of the upper tropospheric Tropical Easterly Jet. Journal of the Atmospheric Sciences, 40, 2708-2722.

Mitchell, T., 2017. Joint Institute for the Study of Atmosphere and Ocean. JISAO Data, Sahel Precipitation Index, 1900-2016. http://research.jiaso.washington.edu/data_sets/sahel/

99 National Oceanic and Atmospheric Administration (NOAA), 2015. Description of Changes to Ocean Niño Index (ONI). Accessed 12 January 2017. [Available online at [http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_change.shtml.]

National Oceanic and Atmospheric Administration (NOAA), 2015. Cold and Warm Episodes by Season. Accessed 12 January 2017. [Available online at [http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears2011.shtml.]

Nesbitt, S.W., and Zipser, E.J., 2003. The diurnal cycle of rainfall and convective intensity according to three years of TRMM measurements. Journal of Climate, 16; 10, 1456-1475.

Nguyen, H., Thorncroft, C.D., and Zhang, C., 2011. Guinean coastal rainfall of the West African monsoon. Quarterly Journal of the Royal Meteorological Society, 137 1828-1840.

Nicholson, S.E., 1981. Rainfall and atmospheric circulation during drought periods and wetter years in West Africa. Monthly Weather Review, 109, 2191-2208.

Nicholson, S. E., 1985. Sub-Saharan Rainfall 1981-84. Journal of Climate and Applied Meteorol- ogy, 24(12), 1388-1391.

Nicholson, S.E., 1986. The spatial coherence of African rainfall anomalies: interhemispheric teleconnections. Journal of Climate and Applied Meteorology, 25, 1365-1381.

Nicholson, S. E., 1994. Recent rainfall ﬂuctuations in Africa and their relationships to past conditions over the continent. The Holocene, 4, 121-131.

Nicholson, S.E., 2005. On the question of the “recovery” of the rains in the West African Sahel. Journal of Arid Environments, 63, 615-641.

Nicholson, S.E., 2008. The intensity, location, and structure of the tropical rainbelt over West Africa as factors in interannual variability. International Journal of Climatology, 28, 1775- 1785.

Nicholson, S.E., 2009. A revised picture of the structure of the “monsoon" and land ITCZ over West Africa. Climate Dynamics, 32, 1155-1171.

Nicholson, S.E., 2011. Dryland Climatology, Cambridge University Press, Cambridge.

Nicholson, S.E., 2013. The West African Sahel: A review of recent studies on the rainfall regime and its interannual variability. ISRN Meteorology, 2013, 1-32.

100 Nicholson, S.E. and Entekhabi, D., 1986. The quasi-periodic behavior of rainfall variability in Africa and its relationship to the Southern Oscillation. Archives for Meteorology, Geo- physics, and Bioclimatology, 34, 311-348.

Nicholson, S.E. and Entekhabi, D., 1987. Rainfall variability in Equatorial and Southern Africa: Relationships with sea surface temperatures along the Southwestern coast of Africa. Journal of Climate and Applied Meteorology, 26, 561-578.

Nicholson, S.E., Fink, A., and Funk, C., in press. Assessing recovery and regime change in West African rainfall from a 161-year record. Manuscript submitted for publication.

Nicholson, S.E. and Grist, J.P., 2001. A conceptual model for understanding rainfall variability in the West African Sahel on interannual and interdecadal timescales. International Journal of Climatology, 21, 1733-1757.

Nicholson, S.E. and Grist, J.P., 2003. The seasonal evolution of the atmospheric circulation over West Africa and Equatorial Africa. Journal of Climate, 16, 1013-1030.

Nicholson, S.E. and Kim, J., 1997. The relationship of the El Niño-Southern Oscillation to African rainfall. International Journal of Climatology, 17, 117-135.

Nicholson, S.E., Kim, J. and Hoopingarner, J., 1988. Atlas of African Rainfall and Its Interannual Variability. Florida State University, Tallahassee, FL. pp. 252.

Nicholson, S.E., Leposo, D., and Grist, J., 2001. The relationship between El Niño and drought over Botswana. Journal of Climate, 14, 323-335.

Nicholson, S.E. and Palao, I. M., 1993. A re-evaluation of rainfall variability in the Sahel part I: Characteristics of rainfall ﬂuctuations. International Journal of Climatology, 13, 371-389.

Nicholson, S.E. and Selato, J.C., 2000. The inﬂuence of La Niña on African Rainfall. International Journal of Climatology, 20, 1761-1776.

Nicholson, S.E., Some, B., and Kone, B., 2000. An analysis of recent rainfall conditions in West Africa, including the rainy seasons of the 1997 El Niño and the 1998 La Niña years. Journal of Climate, 13, 2628-2640.

Okonkwo, C. and Demoz, B., 2014. The relationship between El Niño Southern Oscillations and cereal production in Sahel. Environmental Hazards, 13, No. 4, 343-357.

Okumura, Y., and Xie, S., 2004. Interaction of the Atlantic equatorial cold tongue and the African monsoon. Journal of Climate, 17 3589-3602.

101 Ozer, P., Erpicum, M., Demarèe, G., and Vandiepenbeeck, M., 2003. The Sahelian drought may have ended during the 1990s. Hydrological Sciences Journal, 48:3, 489-492.

Palmer, T.N., Brankovic, C., Viterbo, P., and Miller, M.J., Modeling interannual variations of summer monsoons. Journal of Climate, 5, 399-417.

Parhi, P., Giannini, A., Gentine, P., and Lall, U., 2016. Resolving contrasting regional rainfall responses to El Niño over Tropical Africa. Journal of Climate, 29, 1461-1476.

Parker, D.E., Jackson, M., and Horton, E.B., 1995. The GISST2.2 sea surface temperature and sea-ice climatology. Hadley Centre for Climate Prediction and Research Technical Report CRTN63, pp. 35. [Available from Hadley Centre for Climate Prediction and Research, U.K. Meteorological Ofﬁce, London Road, Bracknell, Berkshire RG12 2SY, United Kingdom.]

Parker, D.J., Burton, R.R., Diongue-Niang, A., Ellis, R.J., Felton, M., Taylor, C.M., Thorncroft, C.D., Bessemoulin, P., and Tompkins, A.M., 2005. The diurnal cycle of the West African monsoon circulation. Quarterly Journal of the Roal Meteorological Society, 131, 2839- 2860.

Peyrille, P. and Lafore, J., 2007. An idealized two-dimensional framework to study the West African Monsoon. Part II: Large-scale advection and the diurnal cycle. Journal of the Atmo- spheric Sciences, 64 2783-2803.

Quinn, W.H., Zopf, D.O., Short, K.S., and Yang, R.J.W.K., 1978. Historical trends and statistics of the Southern Oscillation, El Niño, and Indonesian droughts. Fishery Bulletin. 76, 663-678.

Ramel, R., Gallée, H., and Messager, C., 2006. On the northward shift of the West African monsoon. Climate Dynamics, 26:4, 429-440.

Rasmusson, E.M. and Carpenter, T.H., 1983. The relationship between Eastern Equatorial Paciﬁc sea surface temperatures and rainfall over India and Sri Lanka. Monthly Weather Review, 111, 517-528.

Rodriguez-Fonseca, B., Mohino, E., Mechoso, C.R., Caminade, C., Biasutti, M., Gaetani, M., Garcia-Serrano, J., Vizy, E.K., Cook, K., Xue, Y., Polo, I., Losada, T., Druyan, L., Fontaine, B., Bader, J., Doblas-Reyes, F.J., Goddard, L., Janicot, S., Arribas, A., Lau, W., Colman, A., Vellinga, M., Rowell, D.P., Kucharski, F., and Voldoire, A., 2015. Variability and predictibil- ity of West African droughts: A review on the role of sea surface temperature anomalies. Journal of Climate, 28, 4034-4059.

Roehring, R., Chauvin, F., and Lafore, J., 2011. 10-25 day intraseasonal variability of convection over the Sahel: A role of the Saharan Heat Low and midlatitudes. Journal of Climate, 24, 5863-5878.

102 Ropelewski, C.F. and Jones, P.D., 1987. An extension of the Tahiti-Darwin Southern Oscillation Index. Notes and Correspondence, 115, 2161-2165.

Ropelewski, C.F. and Halpert, M.S., 1987. Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation. Monthly Weather Review, 115, 1606-1626.

Ropelewski, C.F. and Halpert, M.S., 1989. Precipitation patterns associated with the high index phase of the Southern Oscillation. Journal of Climate, 2, 268-284.

Rowell, D.P., 2001. Teleconnections between the tropical Paciﬁc and the Sahel. Quarterly Journal of the Royal Meteorological Society, 127, 1683-1706.

Rowell, D.P., Folland, C.K., Maskell, K., and Ward, M.N., 1995. Variability of summer rainfall over tropical North-Africa (1906-92) observations and modeling. Quarterly Journal of the Royal Meteorological Society, 121, 699-704.

Sanogo, S., Fink, A.H., Omotosho, J.A., Ba, A., Redl, R., and Ermert, V., 2015. Spatio-temporal characteristics of the recent rainfall recovery in West Africa. International Journal of Cli- matology, 35, 4589-4605.

Schonher, T. and Nicholson, S.E., 1989. The relationship between California rainfall and ENSO events. Journal of Climate, 2, 1258-1269.

Schrage, J.M. and Fink, A.H., 2012. Nocturnal continental low-level stratus over tropical West Africa: observations and possible mechanisms controlling its onset. Monthly Weather Re- view, 140 1794-1809.

Sijikumar, S., Roucou, P., and Fontaine, B., 2006. Monsoon onset over Sudan-Sahel: simulaiton by the regional scale model MM5. Geophysical Research Letters, 33 L03814.

Sultan, B., and Janicot, S., 2000. Abrupt shift of the ITCZ over West Africa and intra-seasonal variability. Geophysical Research Letters, 27; 20 2252-2256.

Sultan, B., Janicot, S., and Diedhiou, A., 2003. The West African monsoon dynamics. Part I: Documentation of intraseasonal variability. Journal of Climate, 16, 3407-4327.

Thorncroft, C.D., Nguyen, H., Zhang, C., and Peyrille, P., 2011. Annual cycle of the West African monsoon: regional circulations and associated water vapour transport. Quarterly Journal of the Royal Meteorological Society, 137, 129-147.

Torrence, C. and Webster, P.J., 1999. Interdecadal Changes in the ENSO-Monsoon System. Jour- nal of Climate, 12, 2679-2690.

103 Trenberth, K.E., 1984. Signal versus noise in the Southern Oscillation. Monthly Weather Review, 112, 326-332.

Trenberth, K.E., 1997. The deﬁnition of El Niño. Bulletin of the American Meteorological Society. 78, 2771-2777.

Trenberth, K.E. and Stepaniak, D.P., 2001. Indices of El Niño Evolution. Journal of Climate, 14, 1697-1701. van Loon, H. and Madden, R.A., 1981. The Southern Oscillation Part I: Global associations with pressure and temperature in northern winter. Monthly Weather Review, 109, 1150-1162.

Vaughan, T., 2014. Seasonal Rainfall Variability and Forecasts for the Hawaiian Islands (un- published). National Oceanic and Atmospheric Administration Paciﬁc ENSO Applications Climate Center.

Wang, Z., and Elsberry, R.L., 2010. Modulation of the African easterly jet by a mesoscale convective system. Atmospheric Science Letters, 11:3, 169-174.

Ward, M.N., 1992. Provisionally corrected surface wind data, worldwide ocean-atmosphere surface ﬁelds, and Sahelian rainfall variability. Journal of Climate, 5 454-475.

Washington, R., and Todd, M.C., 2005. Atmospheric controls on mineral dust emission from the Bodélé Depression, Chad: The role of the low level jet. Geophysical Research Letters, 32, L17701.

Washington, R., Todd, M.C., Lizcao, G., Tegen, I., Flamant, C., Koren, I., Ginoux, P., Engel- staedter, S., Bristow, C.S., Zender, C.S., Goudie, A.S., Warren, A., and Prospero, J.M., 2006. Links between topography, wind, deﬂation, lakes, and dust: the case of the Bodélé Depression, Chad. Geophysical Research Letters, 33, L09401.

Wolter, K., 1989. Modes of tropical circulation, Southern Oscillation, and Sahel rainfall anomalies. Journal of Climate, 2, 149-172.

104 BIOGRAPHICAL SKETCH

Originally from rural northeastern North Carolina, weather has always captivated me, and I knew from a very early age that I wanted to be a meteorologist. Pursuing a career that would incorporate my love of weather caught my attention in seventh grade, when I signed up to join my school’s Science Olympiad team. That year, I was assigned to compete in the meteorology com- petition, which is when I first learned about various topics related to the atmosphere and climate. From that moment, I knew that a career in meteorology would be perfect for me. After graduating high school as the valedictorian of my class in 2011, I began my undergrad- uate career as an Applied Atmospheric Science major at East Carolina University in Greenville, NC. At East Carolina, I was selected as one of only fifteen incoming students to join the EC Schol- ars Program, the most prestigious award program offered by the university providing a full ride scholarship and numerous additional benefits including rigorous coursework, research opportunities with faculty members across campus, and a stipend to fund a required Study Abroad experience. I was also an ECU Ambassador, serving as an official student representative of the university and completing over 400 hours of service to the campus and surrounding communities. In 2013, I was selected by the National Oceanic and Atmospheric Administration as a Hollings Scholar, which not only provided additional scholarship support but gave me the opportunity to spend a summer living in Honolulu, Hawaii studying seasonal rainfall variability across the Hawaiian Is- lands. Additionally, I created a seasonal statistical rainfall forecasting scheme which is still in operational use by the NOAA Pacific ENSO Applications Climate Center today. It was this project that first sparked my interest in studying ENSO, and was a precursor to my eventual masters thesis topic. In the spring of 2014, I was able to take a break from my rigorous science and math cur- riculum to study art, Italian language, and digital photography for a semester in Italy with ECU’s Italy Intensives program. Finally, as an Honors College student, I completed a senior honors thesis on the evolution of accumulated cyclone energy during the 2013 Atlantic Hurricane Season, which won the award for Best Natural Sciences Poster at ECU’s annual Research and Creative Achievement Week conference.

105 After graduating from ECU magna cum laude with my B.S. in Applied Atmospheric Science in 2015, I entered the M.S. in Meteorology program in the Department of Earth, Ocean, and At- mospheric Sciences (EOAS) at Florida State University. During my time at Florida State, I served as a teaching assistant for Introductory Meteorology Laboratory for four semesters, including two as the lead teaching assistant. I was fortunate to have been picked up by Dr. Sharon Nicholson as a research assistant beginning my second year of graduate school, when most of the work on this masters thesis commenced. In addition to my academic coursework, teaching, and research, I was also fortunate to have served as the meteorology intern at the Florida Division of Emergency Management (FDEM) for nine months while in graduate school. At FDEM, I gained valuable operational meteorology experience and learned a great deal about preparing and disseminating forecasts to emergency management personnel across the state of Florida. In recognition of my work at FDEM, I was selected as their 2015 intern of the year. Additionally, in the spring of 2016 I was appointed as President of the EOAS department’s meteorology honor society, Chi Epsilon Pi. I am excited to announce that I have achieved a lifelong dream and, following gradua- tion from Florida State, I will begin working as a Meteorologist at the National Weather Service Weather Forecast Ofﬁce in Wichita, KS. It will undoubtedly be a strange feeling to not be enrolled somewhere as a student as I have for the past 21 years of my life, but I look forward to all of the opportunities that lie ahead in Wichita and in my newly established career with the National Weather Service.

106