Annual precipitation over Central America and large areas of Mexico is typically characterised by its bimodal distribution, with a precipitation minimum in July to August that occurs between two separate maxima from May to July and August to October. Several theories have been proposed to explain this phenomenon, which is often termed the mid-summer drought (MSD), but most fail to address the different characteristics associated with individual MSD events. Here, a regression-based approach is used to detect and quantify the annual and climatological MSD signature over Central America and Mexico. This approach has been evaluated and shown to be robust for various datasets with different spatial resolutions. It was found that in the southeast of the Mexico/Central America region, MSDs start earlier and end later than elsewhere, and are thus longer in duration. However, the coast of the Gulf of Mexico, Cuba, and large areas of Central America, exhibit climatologically stronger MSDs. Changes in precipitation, brought about by the interaction between reversals ManuscriptClick here to access/download;Manuscript;full_manuscript_final.docx Click here to view linked References of the onshore/offshore winds and orographic forcing associated with the steep mountainous terrain, have also been shown to be significant factors in the timing of MSD occurrences, offering support for a combined theory of large-scale dynamics and regional forcing. Using self-organising maps (SOMs) as an analysis tool, it was found that MSD events over the domain display strong spatial variability. The MSDs over the domain also generate distinct signatures and may be forced by particular mechanisms. We found that El Niño-Southern Oscillation (ENSO) could be a potential classifier for the SOM identified atmospheric states, based on the correspondence of MSD occurrences with ENSO phases.

The Central American mid-summer drought (MSD) is the decline of precipitation during the middle of the wet season (July and August) over Central America and southern Mexico. It affects agriculture and favours the initiation of bushfires in Costa Rica's national parks, particularly during El Niño years. The MSD is a seasonal phenomenon that varies in intensity and timing inter-annually. The Madden-Julian oscillation (MJO) has been shown to influence Costa Rican rainfall on intra-seasonal time scales, and therefore may be important to the MSD. In this study we use rainfall data from seven stations in Costa Rica to analyse the MJO's influence on the timing of the onset and end of the MSD. We find that the MSD is more likely to start and end in MJO Phases 1 and 8, respectively. Our findings indicate enhanced MSD predictability on intra-seasonal time scales, which could be beneficial to agricultural planning in Costa Rica.

Decadal salinity variability is an important characteristic of the ocean. It characterizes differences in evaporative and precipitation fluxes at the surface, and in the subsurface it contributes to steric sea level change and freshwater/salt transports. In this paper, we identify and describe westward moving and decadally varying salinity anomalies within the thermocline of the tropical South Indian Ocean (SIO) based on ocean state estimates from the Simple Ocean Data Assimilation version 2.2.4 (SODA). This signature in the salinity anomalies is expressed at the depth of 20°C isotherm (D20). A two‐dimensional radon transform quantifies the westward speeds as being between 0.4 and 1.7 cm s−1. These speeds are slower than those of first baroclinic‐mode Rossby waves or mean advection speeds of the background flow in the same regions. The decadal salinity anomaly originates in the subtropical eastern SIO (∼ 39% of the variance explained) and merges with remote anomalies from the western tropical Pacific Ocean (WTPO) via the Indonesian Seas (∼ 11% of the variance explained). The eastern SIO displays both decadal (∼ 10–15 years) and interdecadal (∼ 15–30 years) variability influenced by the WTPO, whereas the decadal variability in the western SIO seems to be more influenced by signals originating in the subtropical eastern SIO. We conclude that these salinity anomalies are consistent with signatures of nonlinear baroclinic disturbances as explained in the recent literature, and possible interaction of higher order baroclinic‐mode Rossby waves with the mean flow.