Saturated hydraulic conductivity (K_s) mapping is an application tool in many fields of science, technology, and land management. Hydrologically, K_s is the main factor controlling aquifer recharge, but it is one of the more difficult hydraulic properties of the soil to predict. This work develops a novel methodology for topsoil K_s mapping at the catchment scale, based on auxiliary data that are quick to determine and low-cost. It includes a double-scale sampling of the K_s to account for small-scale variability in the spatial geostatistical interpolation. A supervised selection of variables through correlation analysis and hierarchical clustering of variables precedes the development of site-specific pedotransfer functions (PTFs) with machine learning techniques. The approach was applied to a heavily anthropized area subject to flooding on the island of Mallorca, Balearic Islands, Spain. The variable selection process filtered out four predictor variables from the initial pool of fourteen predictors. An artificial neural network (ANN) with one hidden layer and six input variables (latitude, longitude, silt, clay, medium sand, and land use) provided the best K_s prediction model. Latitude and longitude coordinates and land use are surrogates for other physical and environmental factors. This ANN was found to be much more accurate than other classical PTFs. The ANN was used to estimate new K_s values from a subsequent sampling of model covariates, which allowed doubling the input information for spatial interpolation using ordinary kriging (OK). The consistency of the resulting K_s map was evaluated by direct comparison with a geomorphological base map. Overall, the spatial distribution of K_s was consistent with the lithological variability as well as with other superimposed anthropic factors.