Atrial tachycardia (AT), flutter (AFL) and fibrillation (AF) are very common cardiac arrhythmias and are driven by localized sources that can be ablation targets. Noninvasive body surface potential mapping (BSPM) can be useful for early diagnosis and ablation planning. We aimed to automatically classify and locate the arrhythmic mechanisms behind AT, AFL and AF using BSPM features. 19 simulations of 567-lead BSPMs were used to obtain dominant frequency (DF) maps from which features reflecting the spatial distribution of DFs and the spectral organization were extracted. Rotational activity was tracked based on singularity points in phase maps; features were extracted to reflect its spatio-temporal stability. The torso was divided in 4 quadrants to assess the spatial distribution of the features. Random forest and least-square based algorithms were used to classify the arrhythmias and their mechanisms' location, respectively. The analyses were reproduced in different layouts (252 to 12 leads). The arrhythmic mechanisms and their locations were classified with 72.0% and 73.9% balanced accuracy, respectively. Accuracy was similar along all lead layouts for arrhythmia classification but decreased for mechanism location. Classification of AT, AFL and AF and their mechanisms' location was feasible based on BSPM features reflecting their basic electrophysiological characteristics.