Tracking a city’s livability fitness is particularly important for the continued urbanization of civilization. Urban environmental and air quality and overall fitness are decreasing due to natural and anthropogenic events, causing degradation of living quality and leading to various population health issues, including heart and lung problems and even premature death. City fitness monitoring is mostly dependent on ground sensor deployment. However, this ground sensor-based monitoring is often not continuous and extensive due to the lack of resources and a very low number of ground observations. Another approach is to estimate city fitness parameters using models built with remote sensing (RS) image and ground sensors data. However, these models are prone to lower accuracy due to the low number of ground sensor observations. In addition, those models mostly used the traditional statistical modeling approach, while deep learning shows promising results and accuracy in other fields of RS. One of the recent approaches is the Chicago Array of Things (AoT) project. AoT provides extraordinarily rich (both spatially and temporally) data. However, this initiative is very expensive and unaffordable for many cities. This study will build models for 15 parameters of city fitness, i.e., environmental and air quality parameters. This study will use ground data from the AoT to develop models using state-of-the-art deep learning, geospatial artificial intelligence (Geo-AI), and exploratory data mining processes in conjunction with remotely sensed open-source data. These models can be used to estimate and track a city’s fitness using satellite data that does not require sophisticated distributed ground sensor technology for continuous collection, such as the AoT for Chicago.