Mapping urban forest tree species using IKONOS imagery: preliminary results

Area of Science:

• Remote Sensing
• Urban Forestry
• Geospatial Analysis

Background:

• Accurate urban forest inventories are crucial for city planning and environmental management.
• High-resolution imagery offers potential for detailed species identification, but challenges remain in complex urban landscapes.

Purpose of the Study:

• To develop and assess a stepwise masking system using IKONOS imagery for identifying and mapping urban tree species in Tampa, Florida.
• To evaluate the accuracy of species identification and explore factors limiting practical application.

Main Methods:

• A stepwise masking system incorporating soil-adjusted vegetation index (SAVI), textural analysis, and near-infrared (NIR) band brightness thresholds was developed.
• Maximum likelihood classification was applied to nine spectral features derived from IKONOS imagery.
• Results were validated against independent ground survey data.

Main Results:

• The system achieved low accuracy when classifying eight individual tree species/groups, falling below practical application levels.
• Even when merging into four major groups, the average accuracy remained suboptimal (73% average, 86% overall, κ=0.76).
• Low spatial resolution, background spectral interference, and shadow effects were identified as key limitations.

Conclusions:

• IKONOS imagery's spatial resolution is insufficient for accurate urban tree species identification in this context.
• Future efforts should consider multi-temporal or hyperspectral data to enhance identification accuracy for practical urban forestry applications.