Interdecadal Changes in Ocean Teleconnections with the Sahel

The oceans have the capacity to accumulate heat in the surface layer. This energy is transferred to the atmosphere leading to changes in atmospheric circulation and consequent impacts in remote and nearby locations. Thus, the SST becomes a key variable to detect ocean sources of seasonal predictability from interannual to multidecadal time scales (e.g., Mohino et al. 2011a; Skinner et al. 2012; Rodríguez-Fonseca et al. 2015). Global spatial patterns of SST anomalies (SSTA) are organized in the so-called modes of variability, which are particular of each ocean basin and determine the principal directions in which the variability takes place. In this framework, the anomalous monsoonal rainfall is strongly linked to SST variability. Regarding the WAM, leading patterns of SSTA play a key role when tackling rainfall variability (Folland et al. 1986; Palmer 1986; Xue and Shukla 1998; Fontaine et al. 1998; Rodríguez-Fonseca et al. 2011, 2015; Losada et al. 2010a, b; Rowell 2013; Vera et al. 2013), becoming a key factor in the severe drought experienced in the Sahel from the early 1970s (Lu and Delworth 2005; Mohino et al. 2011a; Rodríguez-Fonseca et al. 2015). The West African Sahel is a narrow belt located around 15^oN between the Sahara desert to the north and the woody Sudanese savanna to the south, crossing Africa from the Atlantic Ocean to the Red Sea. Every year, the Sahel alternates a dry season, with total absence of precipitation, and a very wet season coinciding with the boreal summer months. Thus, the water resources available during the long dry season depend almost entirely on the intensity of precipitation during the rainy season.

General circulation refers to the global motion of the atmosphere. This atmospheric motion is driven by the uneven surface distribution of net solar incoming radiation, with a surplus in the tropics and a deficit in the polar regions which is translated into larger outgoing long-wave radiation (OLR) than the one that is absorbed (Fig. 2.1). To compensate this imbalance, atmospheric and oceanic transport processes distribute the energy around the globe. Atmospheric winds and ocean currents accomplish this transport. From a dynamical point of view, the oceans and the atmosphere must be considered together as components of the climate system.

As stated, the non-stationary nature of the interannual SST-forced teleconnections with the Sahel rainfall has been merely found out in observational studies (e.g., Janicot et al. 1996; Fontaine et al. 1998; Mohino et al. 2011; Rodríguez-Fonseca et al. 2011, 2015; Losada et al. 2012; Diatta and Fink 2014). In this way, the teleconnections are found to be strengthened or weakened depending on the sequence of decades under study. However, this evidence remains observational, and no physical explanation has been addressed so far, including the impact on predictability.

A monsoon is a circulation system with a set of well-defined features. During summer, winds in the low-level troposphere flow from the colder oceanic regions of the winter hemisphere toward heated continents. Conversely, winds flow from the summer to the winter hemisphere in the upper troposphere. Moreover, the so-called monsoon trough determines the occurrence of rainfall during summer (e.g., Webster et al. 1998, 2002; Webster and Fasullo 2003). This trough of low pressure locates in the surrounding regions of the heated continents and the adjacent oceans and seas, on which precipitation mostly takes place. Most summer rainfall is associated with synoptic disturbances that propagate through the regions aforementioned. These disturbances are referred to as “active monsoon periods,” being grouped in periods of disturbed weather and heavy rainfall lasting from 10 to 30 days. The intervening periods of disruption in this strong convective activity are referred to as “monsoon breaks.” The location of the monsoon trough and maximum monsoon precipitation is generally poleward of the position of the oceanic ITCZ, within which most of the tropical precipitation occurs. This location of maximum precipitation is the so-called tropical rain belt. As stated in the introduction (see Sect.​ 2.​2), each system is different in terms of intensity and atmospheric circulation features. Purely monsoon climates exhibit a single rainfall peak during the solstices, along which dry seasons occur for equatorial climates.

For an optimal interpretation of climate projections, a detailed examination of available data is essential. The knowledge of the main characteristics of the different types of data is important in order to correctly interpret the results obtained from them. The different studies conducted throughout this thesis have made use of different databases. These data can be classified into three fundamental groups:

The methodology applied along this thesis comes from a scientific method in which the available observational data are analyzed with the aim of posing preliminary working hypotheses to be tested with a dynamical model. This line of work starts with the preprocessing of data, a procedure that facilitates the subsequent application of the different methodologies.

As stated, SST is the key variable when tackling seasonal to decadal climate forecast. Dynamical models are unable to properly reproduce tropical climate variability, introducing biases that prevent a skillful predictability. Statistical methodologies emerge as an alternative to improve the predictability and reduce these biases. As a starting point for this thesis, a statistical model was designed and created to improve the predictability and investigate potential non-stationary teleconnections. This model has been named as the sea surface temperature-based statistical seasonal foreCAST model (S⁴CAST). The model is based on the MCA method and introduces the novelty of considering the non-stationary links between the predictor and predictand fields. The results presented in this section are focused on the model development, integrating the methodologies described in Chap. 6. The model has been published (Suárez-Moreno and Rodríguez-Fonseca 2015), and the code is freely available online (see Sect. 7.6). The S⁴CAST model has been created and used throughout this thesis to put forward a series of hypothesis. Most of the results collected in this section correspond to the publication aforementioned.

As stated, the potential causes of alterations in the seasonal cycle of the WAM are mainly related to anthropogenic and natural forcing (Giannini et al. 2008) (see Sect. 2.​2). As part of the results of this thesis, the present section focuses mainly on the latter, with the SST historically playing the dominant role (Folland et al. 1986; Palmer 1986; Xue and Shukla 1998; Fontaine et al. 1998; Ward 1998; Bader and Latif 2003; Giannini et al. 2003; Lu and Delworth 2005; Mohino et al. 2011a; Rodríguez-Fonseca et al. 2011, 2015; Losada et al. 2012; Rowell 2013).

The SST has been identified as driver of interannual to multidecadal Sahel rainfall variability (e.g., Bader and Latif 2003; Mohino et al. 2011), thus becoming a key factor in the predictability of West African droughts (Rodríguez-Fonseca et al. 2015). One of these drivers is the Mediterranean Sea. At multidecadal time scales, the wetting impact of anthropogenic Mediterranean warming has been recently presented by its dominant role on projected Sahel rainfall, prevailing over the tropical Atlantic and Indo-Pacific oceans, which historically were the main drivers of Sahel drought (Park et al. 2016). Otherwise, the impact of Mediterranean SSTA on the Sahel also comprises interannual time scales. Thereby, warm events in the Mediterranean enhance low-level moisture transport across the Sahara to the south, converging in the Sahel with the southwesterly monsoonal flow to increase precipitation. Conversely, cold Mediterranean events are associated to decreased rainfall. Although this teleconnection has been widely described (Rowell 2003; Jung et al. 2006; Fontaine et al. 2010, Gaetani et al. 2010), it has been suggested to be non-stationary over time, getting stronger in some decades compared to others (Fontaine et al. 2011; Rodríguez-Fonseca et al. 2011). Nevertheless, underlying causes for this instability have not yet been found and its clarification would be crucial to improve seasonal predictability of rainfall in the Sahel, with consequent socioeconomic benefits to the region.

The main conclusions of this thesis as well as the potential lines of future work are addressed in this section.