A city-wide platform of sensing devices, or Data Collection Units (DCUs), can harness vehicular networks installed in public transports to gather sensor data at a backend server. Selection of DCU deployment sites must consider connectivity patterns and achievable transfer rates to the vehicles' On-Board Units (OBUs). In this work, we characterize experimentally the connectivity patterns and the achievable data rates and volumes using WiFi between a DCU and passing OBUs of a 400 node vehicular network. Measurements were taken over approximately one month and reveal a strong correlation between connection duration and transferrable data volume. We also evaluate the impact of speed and distance on throughput. Finally, multiple linear regression is used to model attainable data volumes and identify the features that best predict the potential data volume of a tentative deployment site.