Abstract:Assessments and understanding of urban ecosystem health status are of great importance to the sustainable urban development. A health assessment of urban ecosystem was conducted in Guangzhou City in 2000 and 2005 by using a novel multi-indicator data processing method, projection pursuit model, in which the optimal projection direction was obtained based on real-rode genetic algorithm. Results predicted that the grade value of Guangzhou City in 2000 was in a sub-health status, which was in agreement with the real world data. In addition, output from the model is consistent with that obtained from the fuzzy assessment method. The grade value of Guangzhou City in 2005 was also determined as “sub-healthy” based on the actual monitoring data; whereas the model grade value for 2005 was larger than that for 2000. The increased value indicates a continuous improvement of ecosystem health in Guangzhou city during the past five years of efforts. Compared with the fuzzy assessment method, the model was able to accurately assess and project the health status of an ecosystem with higher precision.