ABSTRACT Stream temperature plays a fundamental role in the functioning of aquatic ecosystems and water quality, yet it remains sparsely monitored in many regions, including Brazil. Consequently, empirical air‐stream temperature models are often applied without adequate evaluation of their suitability to local hydroclimatic conditions. This study evaluates stream temperature dynamics and the performance of empirical air‐stream temperature models in three subtropical headwater catchments in southern Brazil. Linear and nonlinear regressions incorporating backward moving averages (BMA) were tested to represent delayed and attenuated stream temperature responses associated with thermal inertia. The results reveal strong seasonal patterns, with daily averages capturing the dominant thermal signals. Hydrological processes characteristic of subtropical headwaters (such as thermal inertia, groundwater buffering and riparian shading) were found to shape the air‐stream temperature relationship and limit the transferability of widely used models. Although locally calibrated equations outperformed those from the literature, underscoring the importance of context‐specific modelling, the analysis also provides broader insight: empirical models that account for lagged responses (e.g., BMA) tend to be more robust in systems with pronounced thermal damping. These findings help bridge a regional knowledge gap and offer guidance for adapting stream temperature models to other tropical and subtropical regions with limited data availability, supporting improved water resource management and ecological assessments.
Fernandes et al. (Thu,) studied this question.