Potential evapotranspiration (PET) is a key parameter in hydrological cycle and ecological process research, and is currently mostly calculated using the Penman-Monteith (P-M) equation. However, direct measurement of PET is usually difficult, resulting in a lack of empirical verification for calculation results and unknown uncertainty. Based on the underlying surface characteristics of the Poyang Lake floodplain wetland, this study used latent heat flux observation data from an eddy covariance system, drew on Granger"s (1989) conceptual analysis of potential evapotranspiration, and combined multiple accuracy evaluation indicators to test the calculation results of the P-M equation under different temperature input conditions. Comprehensive analysis shows that the PET values calculated by different methods are highly consistent with the measured data, but the PET using surface temperature as input is closest to the observed latent heat flux and can better reflect the temporal variation characteristics of potential evapotranspiration. This research result helps to further clarify the sensitivity of potential evapotranspiration to temperature, improve the reliability of calculation results, and has positive theoretical significance and practical reference value.