A multi-experiment ensemble is performed using the WRF (Weather Research and Forecasting) model at high spatial resolution (1 km) over the Antisana glacier region (Ecuador), during the year 2005. Our goal is to identify the best model configurations to simulate atmospheric processes at diurnal and seasonal scales. The model is able to reproduce the complex zonal gradient of precipitation between the wet Amazon and the drier inter-Andean region. The main precipitation biases are (i) an overestimation in the afternoon (up to 6 mm/day) in the Antisana region related to local surface circulation patterns and (ii) a nighttime overestimation (up to 20 mm/day) in the Andes-Amazon transition zone associated with the regional circulation. Changing the microphysics scheme and/or the cumulus scheme primarily affect nighttime processes, while changing the topography forcing and activating slope radiation and shading options mostly affects afternoon processes. An adequate choice of the model configuration allows a correct representation of the diurnal cycle of precipitation, and in particular: (i) the mid-level easterly regional flow, (ii) the local moisture transport along and across the valleys, and (iii) the orographic mountain waves on the Antisana summit. For this specific area and year, the best configuration retained defined as “dSRTM_LRad” shows nighttime (daytime) precipitation biases smaller than 2 mm/day (3 mm/day); it is based on non-smoothed SRTM digital elevation model (dSRTM), Lin Purdue microphysics (L), and slope and shading radiation options (Rad). This 1-km resolution configuration requires the activation of the cumulus scheme, that improves the regional nighttime convection induced by the easterly regional flow on the Amazon-Andes transition region. It allows also a realistic strengthening of the daytime upward moisture transport. This study demonstrates that in the Antisana region, 1 km is a resolution still too coarse to deactivate cumulus schemes for a correct representation of cloud convection.