Connected Cars Could Become Mobile Weather Stations for Climate and COVID Research

Every car rolling off a Brazilian assembly line already carries a small suite of environmental sensors, and according to a new review published in AI Perspectives, those sensors could one day rival the country’s network of fixed weather stations in sheer coverage. The study, led by Philippe Cedraz Lopes and colleagues including Davidson Martins Moreira, set out to answer a deceptively simple question: how well can the sensors already embedded in ordinary vehicles measure the atmosphere, and could that data complement the readings gathered by Brazil’s automatic meteorological stations?

The answer, the authors argue, is that the two systems are strikingly complementary. Brazil’s National Meteorological Institute, known as INMET, operates 775 automatic and conventional meteorological stations across the national territory. Each station collects temperature, humidity, atmospheric pressure, precipitation, wind direction and speed, and solar radiation every minute, then compiles the information hourly for validation and storage at INMET headquarters. By contrast, the Brazilian fleet of conventional automobiles and medium and large commercial pickups is estimated at roughly 43.2 million vehicles, with about 2.8 million new cars produced in 2019 alone, according to the national vehicle manufacturers’ association ANFAVEA. The authors note that the 775 stations represent just 0.028 percent of the cars produced in 2019 alone, a quantitative gap that illustrates how dispersed a vehicle-based sensing network could become.

The sensors involved are not exotic additions. Modern vehicles identify rain on the windshield using the principle of reflection off the glass, which allows automatic activation of the wipers at variable speeds. Outside air temperature sensors, relative humidity sensors, and solar radiation sensors feed the air conditioning system, helping it adjust the temperature of the internal compartment. Light sensors measuring external luminosity, including light from the environment and other vehicles, trigger automatic headlights and adjust the direction of light projection. All of these variables are linked to the vehicle’s location through the Global Positioning System. Because they are designed for automotive applications, the authors emphasize, they carry high reliability and operating precision.

What the sensors currently lack is an outlet. Today, all of these embedded measurements serve only the local vehicle application, informing the car’s own comfort and safety systems. But automakers have begun embedding modems that connect vehicles to external servers over Wi-Fi and 3G and 4G networks, enabling the transmission of high-precision data in real time or on a schedule. That connectivity, the review argues, transforms the car from a closed comfort machine into a mobile data collection unit with high territorial coverage, one that moves constantly through the large urban centers where people actually live.

The comparison with fixed stations is not entirely one-sided in the vehicles’ favor. The automatic meteorological stations can measure at least 18 meteorological parameters distributed across eight categories of information, and their platforms are modular, allowing new sensors to be added at any station as stakeholders identify needs. Conventional automobiles cannot measure seven of those parameters: the dew point temperatures, wind speed, direction, and intensity, and the rainfall volume accumulated over a period. Vehicles also gain new sensors only when automakers launch new models, subject to technical and economic feasibility, a slower and less flexible path than upgrading a station. The authors were careful to note that the adaptation period of both platforms was not analyzed in their study. Still, the vehicles can collect 11 meteorological parameters, and their advantage lies in numbers and mobility rather than breadth.

The review is not a measurement study of its own; it is an observational analysis of vehicle technologies and weather station capabilities, combined with a survey of published research on weather and the coronavirus pandemic. That pandemic connection is what gives the argument its urgency. In 2020, as COVID-19 spread worldwide, researchers scrambled to understand whether meteorological conditions influenced transmission. Several commercial weather data platforms opened their archives to scientists, including Weather Source, which made its API available to researchers globally, and ClimaCell, whose global air quality dataset was used in research on how air quality relates to COVID-19.

The evidence the review surveys on weather and the virus is genuinely mixed, and the authors do not overstate it. Some research has found no coherent relationship between meteorological variables and COVID-19 in places like China, Spain, and the United States, while most studies focused on specific areas have reported positive correlations. A statistical study in New York City observed a relationship between average temperature, minimum temperature, air quality, and COVID-19 cases in the city. A survey in the Middle East reported that in Iran, areas with low wind speed, humidity, and solar radiation exposure showed high rates of infection, findings the authors cited as supporting the virus’s survival under those conditions. A review of studies worldwide concluded that there is a significant association between both temperature and humidity and COVID-19 incidence, consistent with earlier findings on airborne respiratory viruses including SARS, influenza, respiratory syncytial virus, and MERS. And in Brazil, a study of data from the country’s capitals found a correlation between average air humidity and COVID-19 cases among the meteorological variables examined.

The influence also runs in the opposite direction, the review points out. The pandemic itself altered climate-related measurements: since the early days of COVID-19, scientists and civilians observed a sharp improvement in air quality, especially over quarantined regions. Research from the German Aerospace Center’s Remote Sensing Technology Institute showed a healthier atmosphere over Europe, comparing nitrogen dioxide emissions between March and April 2019 with the same months in 2020. For a sensing network that would monitor the atmosphere continuously, this bidirectional relationship, weather affecting the disease and the pandemic response affecting the atmosphere, makes dense, high-resolution environmental data all the more valuable.

The authors’ central conclusion is that automakers already possess the infrastructure and fleet control needed to collect meteorological data with high territorial coverage, opening new ground for the scientific climate field. Vehicles are increasingly seen as connected mobile sensors capable of extending smart cities by mapping data across entire urban areas, gathering information on car flow, meteorological conditions, and pollutant status through communication strategies such as vehicle-to-infrastructure, vehicle-to-vehicle, vehicle-to-people, and vehicle-to-network links. A heterogeneous fleet distributed through a city would increase both the density of collected data and the granularity of the resulting information, feeding a single data center where all measurements, from cars and stations alike, would be processed for weather forecasting.

Real-world projects are already testing the concept. One recent effort led by stakeholders from industry and universities across Europe and Asia aims to obtain real-time weather information on highways from a multimodal network using vehicle locations, with data triggered by events or gathered as time series from embedded sensors, processed on a central server, projected onto a proprietary mobility platform, and trained and validated by machine learning algorithms before alerts are sent to partners, drivers, and government agencies. In its initial phase, that project analyzed the quality of data from three vehicles fitted with external sensing units, validating data quality rather than comparing all meteorological variables against stationary stations, and it plans to expand from 30 test vehicles to a wider fleet. The Brazilian review’s authors argue that the same logic, scaled to millions of vehicles, could benefit not only ordinary applications like weather forecasting, agriculture, and sailing, but also health applications such as disease prevention and diagnosis. Realizing that vision, they caution, will require investment in private digital platforms and data sharing, a careful analysis of Brazilian legal requirements for sharing private data, and further research into artificial intelligence and machine learning models that can turn the massive availability of vehicle data into better weather forecasts and, ultimately, a clearer picture of how climate variables relate to disease.

Subject of Research: The potential of vehicle-embedded environmental sensors to complement automatic meteorological stations in Brazil and support COVID-19 research

Article Title: Study of environmental data from vehicle’s sensors and its aplicability to complement climate mapping from automatic meteorological stations and assess covid-19 impact

Article References: Lopes, P. C., da Silva, J. C. S., Guarieiro, L. L. N., & Moreira, D. M. (2021). Study of environmental data from vehicle’s sensors and its aplicability to complement climate mapping from automatic meteorological stations and assess covid-19 impact. AI Perspectives, 3(1), Article 5. https://doi.org/10.1186/s42467-021-00013-y

Image Credits: AI Generated

DOI: 10.1186/s42467-021-00013-y

Keywords: connected vehicles, mobile sensors, automatic meteorological stations, Brazil, weather forecasting, COVID-19, air quality, smart cities, INMET, automotive sensors, data sharing, machine learning