Among the models used to assess water erosion, the RUSLE model is commonly used. Policy makers can act on cover (C-factor) and conservation practice (P-factor) to reduce erosion, with less costly action on soil surface characteristics. However, the widespread use of vegetation indices such as NDVI does not allow for a proper assessment of the C-factor in drylands where stones, crusted surfaces and litter strongly influence soil protection. Two sub-factors of C, canopy cover (CC) and soil cover (SC), can be assessed from phytoecological measurements that include gravel-pebbles cover, physical mulch, annual and perennial vegetation. This paper introduces a method to calculate the C-factor from phytoecological data and, in combination with remote sensing and a geographic information system (GIS), to map it over large areas. A supervised classification, based on field phytoecological data, is applied to radiometric data from Landsat-8/OLI satellite images. Then, a C-factor value, whose SC and CC subfactors are directly derived from the phytoecological measurements, is assigned to each land cover unit. This method and RUSLE are implemented on a pilot region of 3828 km² of the Saharan Atlas, composed of rangelands and steppe formations, and intended to become an observatory. The protective effect against erosion by gravel-pebbles (50%) is more than twice that of vegetation (23%). The C-factor derived from NDVI (0.67) is higher and more evenly distributed than that combining these two contributions (0.37 on average). Finally, priorities are proposed to decision-makers by crossing the synthetic map of erosion sensitivity and a decision matrix of management priorities.