Abstract Existing comparative studies on conceptual hydrological model structures have generally focused on streamflow, with less attention to evapotranspiration (ET). Yet, appropriate ET representation in conceptual models is crucial for obtaining reliable hydrological simulations, especially under climate change and in ungauged basins. To address this gap, we investigated 30 conceptual models for their ET representations and ability to reproduce two state‐of‐the‐art ET products (PML‐V2 and FLUXCOM‐X‐BASE) across 507 CAMELS‐US catchments with diverse climates and landscapes. FLUXNET ET from nearby sites was used for additional validation. The conceptual model ensemble outperformed a benchmark model that relies on PET and long‐term water balance in most catchments. Models with different ET representations showed distinct ET simulation performance, highlighting the importance of selecting appropriate ET representations. The appropriate ET representations vary across climates. Linear and nonlinear ET–soil–moisture equations with a parameter governing the long‐term ET‐to‐PET ratio are sufficient for accurately reproducing ET products in humid, summer‐rainfall regions, where model equifinality in ET simulation is high. In arid catchments, considering the contribution of lower soil storage to ET generation was necessary to reproduce ET products, especially during rainless periods. In humid winter‐rainfall‐dominated catchments, explicit representation of interception evaporation constrained by interception capacity or storage was critical. Models that performed well in reproducing product‐based ET also performed well when evaluated against FLUXNET ET, supporting the robustness of our findings. This study provides guidance on appropriately representing the conversion of PET and precipitation into ET across diverse climates, thereby helping to constrain model structural uncertainty.
Wu et al. (Sun,) studied this question.