The urban landscape pattern has critical effects on atmospheric particulate matter (PM) pollution. However, the effects may greatly differ with variations of the spatial scales and seasons, which remains poorly understood. This study established a multiple spatial scale analysis on the impact of the urban landscape pattern on PM pollution across different seasons in Nanchang, China, via regular and redundancy analysis (RDA). Six 2D and six 3D metrics were employed to characterize the landscape pattern with buffer radii of 100, 300, 600, 900, 1,200, 2,400, and 4,800 m centered at monitoring stations. Results showed that the effects of the urban landscape pattern on PM pollution have significant seasonal differences and spatial scale effects. Urban landscape pattern has a stronger influence on PM pollution in fall and winter than in spring and summer. The explanatory ability of selected metrics on PM concentrations first increased, then decreased, and finally increased as the scales increased, and the highest accumulated explanatory ability was observed at the 900 m scale. At smaller scales (buffer radii ≤ 900 m), the ability of 3D metrics was stronger than 2D metrics to explain the PM changes, and at larger scales (buffer radii ≥ 1,200 m), the 2D metrics were more explanatory. Percentage of building landscape (PLAND) and patch cohesion index (COHESION) were the key 2D metrics affecting PM pollution, and spatial congestion degree (SCD) and landscape enclosing degree (LED) were the key 3D metrics. The results provide important implications in urban planning for effective PM pollution mitigation.