The development of clouds has many causes, not all of those are examined. In consideration of rainfall behaviour and distribution knowledge of cloud formation, processes in the tropics are of particular importance. Clouds are part of the hydrological cycle, influencing water resources and the energy budget. The insight into unknown cloud generation processes is a great benefit in the developmental procedure of understanding the structure and functionality of an ecosystem and its biodiversity. The main objective of the presented study was to investigate an unidentified nocturnal cloud formation procedure in the eastern Andes of South Ecuador and the adjacent northern Peruvian Amazon. The central theses encompass the confluence of katabatic flows in highly complex terrain due to a concave configuration. This cold drainage of air induces a surface cold front in the foothills of the eastern Andes, which initiates moisture convection due to compressional lifting by the terrain; a nocturnal LLJ triggers the development of the MCS. For the evaluation of the hypotheses the numerical model ARPS was used to analyse the not fully understood highland - lowland interactions in the PBL. At first, simulations of an accurate katabatic flow and its behaviour in complex terrain were performed with optimal conditions and without location information. The main subject of the study was the confluence of the cold drainage of air as a result of concave-lined terrain. Simplified DEMs, inspired by the Andes, were used for this analysis, due to the very steep slopes and valleys of the real terrain. A stepwise increase in their complexity, beginning with a simple slope, enabled the examination of the impact of the terrain configuration on the flow’s dynamic behaviour. With the most complex terrain model, which represents a concave ridgeline interrupted by several valleys draining into a basin, the confluence of the downslope flows due to the geometry of the terrain was demonstrated. Thus, a representative, persistent, thermally driven flow was generated, creating a convergence line that was largest in the centre of the basin. Afterwards, a simulation of a katabatically induced surface cold front with subsequent convective cloud formation was performed with the same model framework, except for the atmospheric water vapour. The simulation showed the same confluence of the downslope flows with a convergence line inside the basin as before. The development of a katabatically induced cold front was identified based on characteristic attributes described in chapter 2 using a cross-section through this line. Furthermore, the results also showed a convergence line that was largest in the centre of the basin. Because of the initiation of moisture convection in this area, due to sufficient moisture in the atmosphere, it was evident that the terrain geometry was the triggering mechanism for cloud formation. The presence of an LLJ in the basin showed the intensification of the cloud formation process. However, the previous results showed that the cluster developed primarily due to compressional lifting by the terrain. This shows that the LLJ had marginal effects on the initiation of moisture convection, acting primarily as an enhancement of its occurrence. Finally, the spatial reference was enabled and the parametrisation set-up of the idealised studies was assigned and adjusted to a multi-nested, approximately realistic model setting, strengthening the evidence from the previous results. For the study, a specific situation, selected on the basis of GOES-E satellite data was used. The main subject was the demonstration of the development of an MCS in the foothills of the eastern Andes due to the presence of katabatic flows as the driving mechanism. The GOES-E brightness temperatures were used to compare the satellite-observed data with the ARPS data to verify the simulated cloud appearance. Due to the fact that the 4 km domain revealed no convective clouds, but a convective cloud cluster was generated on the 1 km domain, a scale dependency was determined. This was caused by two factors: first of all, the NBL processes were simulated more accurately due to the higher vertical resolution, thus more accurately representing the katabatic flows. Furthermore, the higher resolved domain represented a more structured terrain, resulting in stronger convergences of the downslope flows. The comparison of the satellite and the modelled data presented a good agreement concerning the orientation, the location as well as the cold tops of the cells. A closer look at the NBL revealed cold air drainage, nourishing the cell regeneration. The typical characteristics, discussed in the idealised study without location information, confirmed the occurrence of thermally induced downslope flows as the driving process behind convective initiation.