Objective  In order to lay a in-depth theoretical foundation on which the impact of forest spatial layout based on landscape scale on reducing forest fire risk, this paper studies the relationship between spatial characteristics of landscape pattern and forest fire risk.

  Method  The research area of this study is Jiufeng Forest Park. We calculated the index of forest fire risk and made a fire risk rating using the analytic hierarchy process (AHP), extracted the information of landscape pattern to explore the relationship between them. The obtained GF-1 remote sensing images were preprocessed and classified. We calculated the landscape indexes of each sub-compartment in research area using Fragstats 4.2 and the index of forest fire risk by AHP and sub-forest management inventory. So the fire risk rating of Jiufeng forest was divided into 5 levels. The relationship between spatial characteristics of landscape pattern and forest fire risk was studied by grey correlation analysis and correlation analysis.

  Result  (1) Landscape shape index (LSI), patch richness (PR) and Shannon’s diversity index (SHDI) had the highest values at the first level. To some extent, it indicated that the more complex the patch shape was, the greater the landscape segmentation was, the greater the biodiversity and the fire resistance performance were. (2) In the low fire risk areas, all the seven landscape indexes, i.e. number of patches (NP), patch density (PD), contagion (CONTAG), landscape division index (DIVISION), PR, patch richness density (PRD) and SHDI, had little linear effect on forest fire risk. In the moderate fire risk areas, PD had the lowest correlation with forest fire risk index, and PRD had the highest correlation. In the area with fire risk level II, IV and V, the correlation between relative PR and forest fire risk index was the least, while the correlation between PR and forest fire risk index was the greatest, the correlation coefficients of PR and forest fire risk index were −0.452** and −0.497*, and the correlation was significant. This results indicated that the more types the landscape elements in the landscape pattern were, the higher the landscape heterogeneity was, and the lower the risk of forest fire was.

  Conclusion  To some extent, the higher the density of landscape patches and the abundance of the landscape were, with a low degree of the spread, the lower the risk of fire was, and the ability to resist interference and obstruction was the strongest in a fire. If the heterogeneity in the landscape pattern is very low and the biodiversity is relatively low, fire is very easy to occur. The larger the maximum plaque index was, the stronger the human activity was, the more prone to fire. Low heterogeneity and low biodiversity in the landscape pattern were more prone to fire. By calculating the grey correlation degree and correlation between many landscape indices and forest fire risk indices, the indices which can better represent the forest fire resistance are selected, including CONTAG, DIVISION, SHDI, PR and PRD, which can better express the fire resistance of forest landscape.