Evaluation of connectivity indices applied to a Mediterranean agricultural catchment

Connectivity analysis has proved a very valuable approach in understanding runoff and sediment fluxes in catchments. Many analyses use Connectivity Indices to model the patterns, but various versions of the Index have been developed. This paper tests the effects of these variants of the Index as applied to an agricultural-terraced catchment in the semi-arid Mediterranean region where occurrence of connectivity is relatively well documented to validate the modelling. This paper compares three versions of the Connectivity Index that use vegetation/land use (C), topography (RI) and roughness (n) as the weighting factors. Two different slope algorithms are also tested. The weightings are derived from a UAV (drone) survey, with simultaneous ground field mapping. The UAV imagery is found to provide a good, high resolution basis for determining detailed topography and vegetation cover. The slope limit present in some modelling is found to be unsuitable in this landscape where steep terrace embankments that are vegetated in places, and also partially vegetated uplands, occur. Subsequent modelling removed the slope limit. The major patterns of flow produced are similar between the models and are largely related to the overall topography and major drainage lines but considerably modified by the terracing. The model outcomes differ in detail, with the C factor model, incorporating vegetation, coming nearer to the field situation. The main deficiency in all the indices is the weakness in modelling ponding and sedimentation created by the terraces, and therefore the disconnectivity in the landscape. Comparison with field data confirms the variable connectivity in rainstorms, the effects of features such as tracks, and the effect of management actions. Evidence from field mapping during the last two decades indicates that the models probably represent the maximum possible functional connectivity