The widespread use of multi-spectral light emitting diodes (LEDs) in landscape lighting has led to increasing ground-level light pollution and potential ecological risks, highlighting the need for improved spectral assessment tools. However, spectral measurements are expensive and time-consuming, while non-spectral measurements are limited to specific bands. This study introduces a novel colorimetric-based metric for evaluating ecological light pollution from multi-spectral lighting. By analyzing spectral data, we identified a linear relationship between the spectral responses across various species and tristimulus values of different LEDs. Our method, validated with in situ hyperspectral imaging measurements of lighting facades in downtown Shanghai, China, achieves acceptable precision, with maximum errors under 15 % for human circadian rhythms and under 10 % for plant photosynthesis, moths, and bees. To support practical applications, this study presents a general model for light pollution assessments and addresses the issue of metamerism, which affects prediction accuracy in multi-spectral lighting. The innovation of the model proposed in this study lies in its focus on predicting the absolute intensity of ecological light pollution and its applicability across different LED spectra. This colorimetric-based estimation helps to quickly predict the ecological consequences and thus mitigate the increasing light pollution under the development of lighting technologies. It is a preliminary but valuable attempt to integrate ecological light pollution research with multi-spectral lighting practice.