As tourism continues its rapid expansion, its impact on regional water resource systems has become increasingly pronounced, particularly in areas characterized by uneven resource endowments and governance capacities. This study examines two representative regions in China-the arid western region and the Yangtze River Delta-to evaluate the effects of tourism development on Water Environment Carrying Capacity (WECC). A multidimensional WECC evaluation framework was constructed using the entropy weighting method, and both a fixed-effects panel regression model and a Geographically and Temporally Weighted Regression (GTWR) model were applied to capture spatial and temporal heterogeneity. The empirical results show that, at the overall sample level, domestic tourism revenue (DTR) and the number of tourists (NT) exert significant negative effects on WECC (DTR = - 0.0020, NT = - 0.0019, both p < 0.01). However, strong regional divergence exists: in the arid western region, both DTR and NT have significant positive effects on WECC (DTR = + 0.0054, NT = + 0.0095, p < 0.01), indicating that tourism income and visitor flows can enhance WECC by promoting fiscal reinvestment and infrastructure improvement in resource-scarce areas; whereas in the Yangtze River Delta, both indicators exert significant negative effects (DTR = - 0.0024, NT = - 0.0029, p < 0.01), reflecting the persistent environmental pressure caused by high-intensity tourism. The GTWR results further reveal strong spatial and temporal heterogeneity in these effects. Based on these findings, this study recommends that resource-limited regions channel tourism-generated revenues toward water infrastructure and pollution control, while developed regions should regulate tourism intensity and optimize industrial structure to alleviate environmental stress. These quantitative findings provide clear directions for future research, including exploring threshold and nonlinear mechanisms of tourism impacts, incorporating micro-level enterprise and tourist behavior data to test mediating pathways, and employing high-resolution spatiotemporal data for dynamic effect identification.