We present a novel and efficient approach that enables the evaluation of environmental quality in cities worldwide using high-resolution satellite imagery, based on a new green index (GI) through multivariate analysis, to compare the proportion of urban green spaces (UGSs) with built and impervious surfaces. High-resolution images were used to perform a supervised classification of 25 districts in the city of São Paulo, Brazil. Only 11 districts showed higher urban forests, green spaces, green index, and green vs. built values, and impervious surface proportions with lower impervious and built spaces. On the other hand, the remaining districts had higher population densities and unfavorable conditions for urban ecosystem development. In some cases, urban green spaces were three-times smaller than the built and impervious surfaces, and none of the districts attained a high green quality index (0.75 to 1). Artificial intelligence techniques improved the precise identification of land cover, particularly vegetation, such as trees, shrubs, and grasses. The development of a novel green index, using multivariate statistical analysis, enhanced positive interactions among soil cover classes, emphasizing priority areas for enhancing environmental quality. Most of them should be prioritized by decision makers due to the low environmental quality, as identified by the low green index and worse ecosystem services, well-being, and health outcomes. The method can be employed in many other cities to enhance urban ecosystem quality, well-being, and health. The green index and supervised classification can characterize pastures, degraded forest fragments, and guide forest restoration techniques in diverse landscapes.