In onshore wind turbines, the association of fatigue damage with the environmental conditions based on the wind field properties have already been shown to be effective. However, most of the predictive power of such approach is based on average results, being the informative power regarding a single event extremely reduced. In this paper, a new methodology based on a reduced order model where the tower top accelerations are explicitly taken into consideration is proposed, with the adequate adaptations for floating offshore and onshore applications. After a numerical validation of the model predictions using OpenFAST, the methodology is used to reconstruct the bending moments time series of both an onshore and a floating wind turbine that are currently being monitored with accelerometers and strain gauges. The existence of this experimental data allowed for a direct comparison of the accelerations based estimates with the values obtained from the strain gauges readings and the method was proven to reproduce with great accuracy the tower bending moments time series. This conclusion sustain the claim that the proposed methodology for fatigue estimation is more robust than conventional SCADA-driven approaches and more informative regarding each individual event, making it suitable to be used in real time decision making processes.