We explored the mass concentrations of water-soluble ions in PM_2.5 and their variations across different time scales and concentration levels. Using the Positive Matrix Factorization (PMF) model and backward trajectory analysis, we focused on identifying the sources of PM_2.5 and its water-soluble ion fractions, with particular emphasis on regional transport. The findings reveal that the average mass concentration of total water-soluble ions in Hainan between 1 August 2021 and 31 July 2022 was 7.0 ± 4.4 µg m⁻³, constituting 73.5% ± 24.4% of PM_2.5. Secondary ions (SO₄²⁻, NO₃⁻, NH₄⁺) were dominant, accounting for 84.0% ± 12.4% of the total water-soluble ions, followed by sea-salt particles. Seasonal variations were pronounced, with the highest concentrations observed in winter and the lowest in summer. The results of the PMF analysis showed that secondary sources, combustion sources, dust sources, and oceanic sources are the main sources of PM_2.5 at the monitoring site. The potential sources and transport pathways of water-soluble ions exhibit distinct seasonal characteristics, with the land-based outflows from the YRD–PRD–Fujian corridor controlling Hainan’s PM_2.5 maxima, while southerly marine air delivers the annual minimum; seasonal alternation between dust/secondary aerosols (winter–spring), combustion (autumn), and oceanic dilution (summer) dictates the island’s air-quality rhythm.

Vehicle emission inventories are highly sensitive to vehicle activity data, yet annual vehicle kilometers traveled (VKT) is still commonly represented using generalized default values whose representativeness at the city scale remains uncertain. In this study, large-scale vehicle inspection data from Haikou, China, were used to derive inspection-based VKT estimates and to quantify how activity assumptions affect urban vehicle emission inventories and policy evaluation. By holding vehicle population and emission factors constant across scenarios, we explicitly isolated the effect of activity representation on emission estimates. An inspection-based, age-sensitive VKT framework was further developed to capture within-fleet heterogeneity. The results showed that inspection-derived VKT accounted for only 36–75% of guideline-recommended values across major vehicle categories, with the largest discrepancies observed for diesel freight vehicles. As a result, the use of guideline-based VKT produced higher emission estimates by 34–39% for carbon monoxide (CO) and volatile organic compounds (VOCs) and by approximately 66–67% for nitrogen oxides (NOx) and particulate matter (PM). The influence of activity representation was also evident in policy assessment. In a case study of old diesel vehicle retirement, guideline-based VKT produced estimated emission reduction benefits that were more than 120% higher for most pollutants and nearly 200% higher for NOx than those derived from inspection-based VKT. These findings demonstrate that generalized activity assumptions can substantially affect both emission inventory estimates and policy-oriented assessments. Rather than merely refining a local mileage parameter, this study highlights a potential representativeness limitation of generalized activity assumptions when they are applied to city-specific emission inventories, particularly in medium-sized or geographically constrained urban systems. The inspection-based, age-sensitive approach proposed here provides a practical pathway for improving activity representation in data-rich urban environments, while its transferability should be evaluated according to local fleet structure and transport conditions.