The spatiotemporal evolution of the snow cover may help to obtain conclusions on the variability of the atmospheric agents in high mountain areas. That evolution is difficult to analyze due to the heterogeneity of the snow distribution on the ground. The use of terrestrial images which are treated to obtain snow detection, is an inexpensive and promising technique more capable of solving the drawbacks that other techniques presented. This work analyzes the spatiotemporal variability of the snow by using terrestrial photography and the effects of scale on its modelling in the river Trevélez valley, in southern Spain. Temporal series of images of the area were employed from September, 2011 up to May, 2013. By georeferencing the images, the snow pixels were identified, and the temporal variations in the snow cover with respect to its spatial distribution were determined. The maps obtained were used as a direct source of data assimilation in that model. Finally, the improvement in the global simulation of the snow model when this data source was incorporated was assessed by making a comparative study between the temporal series of the snow flow measured at the gauging point band the flow obtained in the simulation. As a result, a temporal series of snow maps of the area was made. In turn, the assimilation of the data improved the simulation by up to 9.74% for the equivalent of water. At a watershed scale, the simulation of the flow at the control point reproduced the trend observed. These results permit one to conclude that the methodology used is precise enough to find out the exact position of snow cover and to improve the efficiency of the model used.