Abstract:Understanding the dynamics of p CO 2 and p CH 4 is important for evaluating carbon emissions from the aquatic environment. While temporal dynamics of p CO 2 and p CH 4 have been extensively studied, there is a noticeable gap in the literature concerning their spatial characteristics. In this study, we used boat-mounted sensors to directly measure p CO 2 and p CH 4 with high spatial resolution across seasons in Xuanwu Lake (XWL), an urban lake in Nanjing, China. Additionally, water chemistries were measured at selected sites for correlation analysis. Sensitivity analysis was performed to assess the effects of measurement location and density on carbon flux estimates.Results show that continuous on-board measurements innovatively captured the spatial distribution of p CO 2 and p CH 4 . Both gases varied significantly across seasons, exhibiting pronounced spatial heterogeneity. Peak emissions occurred in summer, with the lowest CO 2 in spring and CH 4 in fall. Both p CO 2 and p CH 4 increased from the lake center to the shoreline, with the largest fluctuations near the shore. In total, XWL acted as a net source of CO 2 and CH 4 , with mean diffusive fluxes of 11.4 ± 10.1 and 3.0 ± 0.3 mmol∙m −2 ∙d −1 , respectively. Furthermore, sensitivity analysis showed that lake-wide flux calculations depend on measurement locations and sample size. p CO 2 exhibited greater heterogeneity than p CH 4 , necessitating different sampling thresholds. The optimal threshold for capturing lake-wide CO 2 flux was 18–30 samples per km 2 , while this density was sufficient for CH 4 flux estimates. Our study highlights that both CO 2 and CH 4 exhibit significant spatial and temporal heterogeneity, necessitating high-resolution sampling for accurate flux assessments. The sampling strategy presented could guide future studies, as sampling in the intermediate zone effectively reduces the number of samples needed while maintaining accuracy.

Spatial and seasonal variations of carbon emissions in an urban lake: Flux sensitivity and sampling optimization based on high-resolution measurements