Vegetal covers protect the soil against erosion in agricultural systems. In marginal olive farms, spontaneous grass covers are preferred due to their low cost, despite their high spatial and temporal heterogeneity which limits their efficiency. Although the impact of different managements, soil conditions or cover crops species has been evaluated and compared, there is currently insufficient information available on the spatial and temporal patterns of spontaneous grass covers in olive groves. This work focuses on the analysis of the spatial distribution patterns, temporal stability and protection capacity of spontaneous grass covers in olive orchards. First, the temporal variability of spontaneous grass cover was explored in relation to the meteorological conditions and the management system. Second, spatial analysis by distance indices (SADIE) was applied to explore the spatial patterns and persistence of the spontaneous cover. Finally, a descriptive analysis of events was made to illustrate the degree of influence of the cover on the hydrological response of the catchment. A total 12 surveys were carried out on a seasonal scale over 3 years (2011 − 2013) to measure the cover percentage of spontaneous grass through surface photograph analyses. A regular, permanent grid of 37 points (6 samples/ha) was used to define each sample-location. In order to evaluate the relations between the cover percentage and meteorological variables, multiple linear regressions were adjusted while the SADIE approach and correlation analysis were used to describe possible spatial aggregation patterns and its dependence on topographical and soil attributes such as aspect, slope, drainage area, height, richness of grass species and apparent electrical conductivity. The average annual percentage of cover ranged from 23% to 36%, with a coefficient of variation of 57% and 6%, respectively. On the seasonal scale, the cover varied between 0.2% and 50%. Accumulated precipitation during the precedent 15 days, the average of minimum temperature of the previous 60 days and the cumulated potential evapotranspiration of the same period showed a significant correlation with the mean cover percentage in the catchment. A simple multiple linear regression adjustment between the cumulated precipitation for the 15 previous days and the number of months since the vegetation was removed using herbicide or by summer conditions gave a determination coefficient of 77%. Moreover, a permanent spatial pattern was observed for periods characterized by abundant preceding rainfall. Only the apparent electrical conductivity of the topsoil (50 cm in depth) showed significant correlations with the spatial patterns of spontaneous grass. Finally, similar rainfall events taking place with different degrees of soil cover presented very different hydrological responses, which allowed us to quantify the contribution of spontaneous vegetation to sediment dynamics during humid periods. [ABSTRACT FROM AUTHOR]