We introduce a dual modeling approach that is able to detect and locate leaks reliably by analyzing telemetry data from pressure and flow sensors of district metered areas (DMAs). The superiority of this approach to conventional model-based leak detection and localization algorithms arises from a transformation of pressure signals into virtual leak flows, which magnifies and localizes leak responses of pressure sensors. The dual model is applied on data from a DMA, where artificial leaks with small flow rates are created on various positions in the system by marginally opening fire hydrants.