In this work we propose and apply a straightforward methodology for the automatic characterization of the extended earthquake source, based on the progressive measurement of the P-wave displacement amplitude at the available stations deployed around the source. Specifically, we averaged the P-wave peak displacement measurements among all the available stations and corrected the observed amplitude for distance attenuation effect to build the logarithm of amplitude vs. time function, named LPDT curve. The curves have an exponential growth shape, with 31 an initial increase and a final plateau level. By analyzing and modelling the LPDT curves, the information about earthquake rupture process and earthquake magnitude can be obtained. We applied this method to the Chinese strong motion data from 2007-2015 with MS ranging between 4 and 8. We used a refined model to reproduce the shape of the curves and different source models based on magnitude to infer the source-related parameters for the study dataset. Our study shows that the plateau level of LPDT curves has a clear scaling with magnitude, with no saturation effect for large events. By assuming a rupture velocity of 0.9Vs, we found a consistent self-similar, constant stress drop scaling law for earthquakes in China with stress drop mainly distributed between a lower level (0.23Mpa) and a higher level (3.74Mpa). The derived relation between the magnitude and rupture length can be used for probabilistic hazard analyses and real-time applications of Earthquake Early Warning systems.