FAQ

The Baltimore air quality dashboard displays maps of concentrations of fine particulate matter (PM2.5) in Baltimore, based on air quality data on Baltimore from multiple sources -- Maryland Department of the Environment (MDE), the SEARCH low-cost monitor network data, and the PurpleAir low-cost monitor network data. The MDE air quality data is the highest quality data (also called “reference data”) and is the data that the state reports to the Environmental Protection Agency (EPA). The dashboard currently offers both a continuous map covering Baltimore and some of its surroundings, and a neighborhood-level aggregate map (City View) for neighborhoods within city limits. Maps are available from October 2019 to June 2025. You can view maps over a single day, week, month, year or any custom range by selecting dates.

PM2.5 or fine particulate matter is comprised of small solid or liquid particles in the air that have a diameter of 2.5 micrometers or smaller, allowing them to enter your lungs.[1] For comparison, the diameter of a typical human hair is about 100 micrometers.

PM2.5 has been linked to many poor health outcomes including cardiovascular disease and lung disease.[2]

PM2.5 is measured as a concentration: the mass of particles in micrograms per 1 cubic meter of air. For comparison, a typical grain of table salt weighs about 60 micrograms.

[1] EPA website: https://www.epa.gov/criteria-air-pollutants

[2] Dockery, D. W., Pope, C. A., III, Xu, X., Spengler, J. D., Ware, J. H., Fay, M. E., Ferris, B. G., Jr., & Speizer, F. E. (1993). An association between air pollution and mortality in six U.S. cities. New England Journal of Medicine, 329(24), 1753–1759. https://doi.org/10.1056/NEJM199312093292401

The Maryland Department of the Environment (MDE) operates a statewide network of regulatory monitoring stations that use Federal Reference Method (FRM) or Federal Equivalent Method (FEM) instruments approved by the U.S. Environmental Protection Agency (EPA) to measure concentrations of different air pollutants. FRM monitors follow strict EPA reference protocols, while FEM monitors are alternative instruments shown through testing to produce results equivalent to FRM. Data from these high-accuracy, research-grade instruments are often called reference data, and the data is used for regulatory purposes (for example, they are used to show whether the area is meeting the National Ambient Air Quality Standards, as required by the EPA), for health assessments, and for evaluation of lower-cost sensors. More details are available in the document on MDE’s 2025 ambient air pollution monitoring plan https://mde.maryland.gov/programs/air/AirQualityMonitoring/Documents/MDNetworkPlanCY2025.pdf.

Low-cost air pollution sensors are smaller, relatively inexpensive devices that measure pollutants such as particulate matter (PM2.5), ozone (O₃), or nitrogen dioxide (NO₂). Unlike the research-grade instruments used in regulatory monitoring, these sensors are compact, easier to deploy widely, and often provide real-time data. The sensors are packaged with other electronics into a usable monitor. The monitors make it possible for communities, researchers, and individuals to monitor local air quality at a much finer scale at a reasonable price. However, their measurements can be less accurate and more sensitive to environmental conditions, so they are often compared against or calibrated with reference monitors such as those described above. More information is available on the U.S. Environmental Protection Agency’s website: EPA – Air Sensor Toolbox.
https://www.epa.gov/air-sensor-toolbox.

The SEARCH Center was an EPA funded study that was a collaboration between Johns Hopkins University, Yale University, and other university partners.As part of the center, we built and deployed low-cost multi-pollutant air pollution monitors that are placed around Baltimore. We call these monitors the SEARCH network.

More information about the sensors used in SEARCH network in available in Buehler et al. 2021. [1]

[1] Buehler, C., Xiong, F., Zamora, M. L., Skog, K. M., Kohrman-Glaser, J., Colton, S., McNamara, M., Ryan, K., Redlich, C., Bartos, M., Wong, B., Kerkez, B., Koehler, K., & Gentner, D. R. (2021). Stationary and portable multipollutant monitors for high-spatiotemporal-resolution air quality studies including online calibration. Atmospheric Measurement Techniques, 14(2), 995–1013. https://doi.org/10.5194/amt-14-995-2021

PurpleAir monitors are a widely used type of low-cost air pollution monitor designed to measure particulate matter (PM2.5 and PM₁₀) in real time. They use similar sensors as the sensors in the SEARCH network. PurpleAir monitors connect to Wi-Fi, automatically uploading data to an online map that shows air quality readings from thousands of sensors worldwide. Data from PurpleAir sensors shared by their owners are available to download from PurpleAir.com.

While low-cost sensors provide valuable local and high-resolution information, their measurements can be influenced by outdoor weather conditions, like relative humidity. Data from low-cost sensors should be used with caution and after appropriate quality control and adjustments (calibration). Often the data from low-cost sensors are calibrated against reference monitors to improve accuracy by accounting for other meteorological variables like relative humidity and temperature.

The Air Quality Index (AQI) is a simple, color-coded, unitless index that is an effective way to communicate air pollution concentrations.

The AQI provides an indication of the quality of the air and its health effects. It takes values from 0.

An AQI of 101 typically corresponds to the level that violates the national health standard.

More information about AQI can be found at https://www.airnow.gov/aqi/aqi-basics/.

There can be multiple several reasons for that:

PM2.5 is often called a regional pollutant, meaning that it doesn’t vary as much spatially as it does over time. That is partially because PM2.5 can stay in the atmosphere a pretty long time (up to a week) and because there are a lot of sources across the city (cars, industry, restaurants, etc.). However, because of weather, PM2.5 does vary a lot from day to day. The weather in Baltimore is the similar in most parts of the city, so the impact of weather is more consistent across the city on the same day.

It is also important to recognize that the color coding of the maps can make it difficult to see smaller differences within the city. The color coding used in the map is standardized, based on Air Quality Index, and all locations with concentrations that fall in the same AQI category get very similar colors on the maps, although there may be considerable difference in their values. For example, the ‘Moderate’ AQI category ranges from 9.1–35.4 µg/m³. This is a relatively wide range of concentrations but the color for this range is generally yellow (yellowish green at the lower end, and yellowish orange at the upper end) as they all correspond to the same AQI category. Hovering over the map lets you see the actual differences in concentrations.

Very localized and short-lasting peaks in concentrations are unlikely to be captured in these maps, as a) there may not be a monitor nearby to record this spike, b) even if there is a monitor nearby, a very short-term spike will get averaged out when creating the daily maps. If there is a persistent spike in concentration and a sensor nearby, it is likely to be picked by and displayed in the 98th percentile maps for that day.

You can find diurnal patterns of PM2.5 for your neighborhood by selecting the neighborhood from the dropdown list and selecting a custom date range for which you want this information. The resulting “PM2.5 Hourly Patterns” plot will then display how the average concentrations looked like in your neighborhood for each hour of the day, to help identify the hours with generally high or low concentrations. Selecting a longer time period will give you a better estimate of the typical conditions.

You can find seasonal patterns of PM2.5 for your neighborhood by selecting the neighborhood from the dropdown list and selecting a custom date range that spans at least 365 days. The resulting “Seasonal Patterns” plot will then display how the average concentrations looked like in your neighborhood for each of the four seasons.

In the morning, cold air gets trapped near the ground and the pollution is confined to a shallow layer close to the surface. This leads to higher concentrations of pollution.During the day, when the air warms the air rises leading to more mixing and dilution of the pollution resulting a lower concentration of pollution.

Similarly, in the winter, the cold air tends to be trapped near the ground, like in the morning.During winter days, it may not get warm enough to encourage as much mixing and dilution as we see during the summer. There can also be more pollution from some sources during the winter. For example, heating buildings requires more electricity during the winter, which releases more pollution than in warmer seasons.

• Conserve energy - turn off unnecessary lights and turn down air conditioning.
• Set thermostats higher in summer and lower in winter. Look for the ENERGYSTAR label.
• Carpool, use public transportation, bike, or walk whenever possible. If driving,minimize idling.
• Follow gasoline refueling instructions for efficient vapor recovery, being carefulnot to spill fuel and always tightening your gas cap securely.
• Keep car, lawnmower, boat, and other engines properly tuned. Be sure your tires areproperly inflated.
• Use environmentally safe paints and cleaning products whenever possible.
• Choose low-VOC paint or cleaning products and seal them tight.
• Mulch or compost leaves and yard waste. Avoid burning leaves, trash, and othermaterials.
• Avoid using gas-powered lawn and garden equipment.

• Avoid exercising outdoors when pollution levels are high. When the air isbad, walk indoorsor use an exercise machine.
• Avoid exercising near busy roads. Busy highways can create high pollution levels up to one-third a mile away.
• Avoid driving on the busiest roads. Choosing a parallel route only a blockaway can significantly reduce the pollution you breathe. Choose lessstrenuous outdoor activities because they make you breathe in more air.
• Avoid excessive idling of your automobile.
• Reduce the number of trips you take in your car.
• Use appropriate masks for protection. Well-fitted N95 or KN95 arerecommended to protect against small particles like PM2.5.

Currently, the dashboard only shows PM2.5 maps. We plan to include maps of other pollutants in the future. Measurements of Ozone, PM2.5, and PM10 for the MDE reference devices are available in https://gispub.epa.gov/airnow/.

Reference sensor data for the United States are available at
https://gispub.epa.gov/airnow/.

Within the Baltimore city limits, there is one MDE reference device that records hourly concentrations of PM2.5. This instrument was located in Oldtown until 2021 and moved to Lake Montebello from 2022. More details about the locations with reference devices in Maryland are available in the document on MDE’s 2025 ambient air pollution monitoring plan https://mde.maryland.gov/programs/air/AirQualityMonitoring/Documents/MDNetworkPlanCY2025.pdf.

The SEARCH monitors were installed throughout the city on a rolling basis from Dec 2018. There are more than 30 devices operating in the city. A map of locations of recently active monitors is available in Figure 1 of Heffernan et al. 2024 [1].

The locations of PurpleAir monitors operating in and around the city can be found at https://map.purpleair.com/.

[1] Heffernan, C., Koehler, K., Gentner, D. R., Peng, R. D., & Datta, A. (2024). Unified calibration and spatial mapping of fine particulate matter data from multiple low-cost air pollution sensor networks in Baltimore, Maryland (arXiv:2412.13034) [Preprint]. arXiv. https://arxiv.org/abs/2412.13034

The map shows PM2.5 concentrations (in micrograms per cubic meter). The colors applied to these values are selected following the colors scheme used to define U.S. Air Quality Index (AQI) categories. These colors and categories are the following:

green (Good, 0–9 µg/m³, AQI 0–50),

yellow (Moderate, 9.1–35.4 µg/m³, AQI 51–100),

orange (Unhealthy for Sensitive Groups, 35.5–55.4 µg/m³, AQI 101–150),

red (Unhealthy, 55.5–125.4 µg/m³, AQI 151–200),

purple (Very Unhealthy,125.5–225.4 µg/m³, AQI 201–300), and

maroon (Hazardous, 225.5+ µg/m³, AQI 301+).

There are multiple steps involved in creating the maps. Two SEARCH low-cost monitors are placed next to the MDE reference device. Because the sensors behave similarly at all locations, we can learn how to calibrate these sensors and apply that calibration to improve the accuracy of all the SEARCH low-cost monitors. For the PurpleAir monitors, there is an US-wide calibration equation published in Barkjohn et al. 2021 [1] which we used for adjustment. We only used data from PurpleAir monitors hosted outdoors within a perimeter around the city limits and from 2023 onwards when there was adequate number of PurpleAir monitors in the city (prior to 2023 there were only about 10 PurpleAir monitors in the Baltimore region). Data from all three sources went through appropriate quality control steps before use. By combining data from all three networks, we are able to create maps that best account for air quality differences across the Baltimore area.

For every hour, data from all active devices (MDE, SEARCH, PurpleAir) are combined in a Bayesian spatial model that adjusts the low-cost sensor data from SEARCH and PurpleAir using their respective calibration equations, and models spatial correlation in the data and produces modeled estimates of PM2.5 at any location in the city. Details of the methodology can be found in Heffernan et al. 2023, 2024 [2,3].

We use this methodology to produce modeled estimates of PM2.5 for each location in a 100x100 grid covering the city for every hour from Oct 2019 to Jun 2025. Daily maps are then created by taking average of the hourly estimates for that day over the 100x100 grid and plotting over a map of the city. When selecting a time range spanning multiple days, the map produced is the average of estimates of all hours within the selected time range.

[1] Barkjohn, K. K., Gantt, B., & Clements, A. L. (2021). Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor. Atmospheric Measurement Techniques, 14(7), 4617–4637. https://doi.org/10.5194/amt-14-4617-2021

[2] Heffernan, C., Peng, R., Gentner, D. R., Koehler, K., & Datta, A. (2023). A dynamic spatial filtering approach to mitigate underestimation bias in field calibrated low-cost sensor air pollution data. The Annals of Applied Statistics, 17(4), 3056–3087. https://doi.org/10.1214/23-AOAS1751

[3]Heffernan, C., Koehler, K., Gentner, D. R., Peng, R. D., & Datta, A. (2024). Unified calibration and spatial mapping of fine particulate matter data from multiple low-cost air pollution sensor networks in Baltimore, Maryland (arXiv:2412.13034) [Preprint]. arXiv. https://arxiv.org/abs/2412.13034

For every hour, data from all active devices (MDE, SEARCH, PurpleAir) are combined in a Bayesian spatial model that adjusts the low-cost sensor data from SEARCH and PurpleAir using their respective calibration equations, and models spatial correlation in the data and produces modeled estimates of PM2.5 at any location in the city. The calibration equation used for PurpleAir monitors is the US-wide equation proposed in Barkjohn et al. (2021) [1]. Calibration equation for the SEARCH monitors and and details of the methodology can be found in Heffernan et al. 2023, 2024 [2,3].

We use this methodology to produce modeled estimates of PM2.5 for each location in a 100x100 grid covering the city for every hour from Dec 2019 to Jun 2025. Daily maps are then created by taking average of the hourly estimates for that day over the 100x100 grid and plotting over a map of the city. When selecting a time range spanning multiple days, the map produced is the average of estimates of all hours within the selected time range.

[1] Barkjohn, K. K., Gantt, B., and Clements, A. L. (2021). Development and application of a united states-wide correction for pm 2.5 data collected with the purpleair sensor. Atmospheric Measurement Techniques 14, 4617–4637.

[2] A dynamic spatial filtering approach to mitigate underestimation bias in field calibrated low-cost sensor air pollution data Heffernan, C., Peng, R., Gentner, D. R., Koehler, K., & Datta, A. (2023) The annals of applied statistics, 17(4), 3056.

[3] Unified calibration and spatial mapping of fine particulate matter data from multiple low-cost air pollution sensor networks in Baltimore, Maryland Heffernan, C., Koehler, K., Gentner, D. R., Peng, R. D., & Datta, A. (2024) arXiv preprint arXiv:2412.13034.

For a selected date range, the average map shows the average of the hourly estimated concentrations at each location or neighborhood in the map. In addition to the map of average concentrations, we also map the 98th percentile of the concentrations from the hours of the same date range. For example, if you look at one week of data, the 98th percentile would show you the concentrations for the worst three hours that week.

For fine particulate matter (PM2.5), the U.S. EPA’s National Ambient Air Quality Standards (NAAQS) are defined in part using the 98th percentile of daily average concentrations. Specifically, compliance with the annual health standard is determined by the 3-year average of the 98th percentile of daily values. Mapping the 98th percentile highlights the higher-end pollution days that are most relevant to health and regulatory assessment, while reducing the influence of single extreme outliers. However, it is important to note that compliance with the NAAQS can only be evaluated using the exact form of the standard (the 3-year average of the 98th percentile of daily values). Our maps may show the 98th percentile for any user-selected date range, but those results should not be interpreted as regulatory compliance assessments. For details, see EPA’s overview of the PM NAAQS at

https://www.epa.gov/system/files/documents/2024-02/pm-naaqs-overview.pdf.

The Area Map extends slightly beyond the Baltimore city limits, roughly covering a circular region within which the SEARCH monitors are sited. This is a continuous map created by interpolating the modeled estimates of PM2.5 on the 100x100 grid to the entire area. You can hover over the map to see the concentration at the various locations.

The City Map provides neighborhood level averages of PM2.5 for each neighborhood within city limits. The neighborhood names and boundaries are provided by the Baltimore City government https://github.com/Baltimore-City-EGIS/GISdata/blob/master/Neighborhoods.geojson.

For each neighborhood, we average the estimates of PM2.5 over the subset of the 100x100 grid-cells that fall within the neighborhood. This averaged number is displayed on the map and shows up when you hover over a neighborhood.

Data from the reference MDE sensor for PM2.5 in Baltimore can be seen at https://gispub.epa.gov/airnow/. The raw PurpleAir data is available on www.PurpleAir.com. We do not display the raw PurpleAir or SEARCH data in this dashboard as the raw data from these sensors are biased and may convey incorrect information about the air quality at a location. Instead, we present the calibrated estimates, that combines all available data to get the best estimates of PM2.5 concentration at each location.

The Analytics page provides various analytics on air quality trends at a selected neighborhood in the city (using the drop-down menu) and for a date range.

• The “PM2.5 Timeseries Patterns” plot provides a time-series plot of hourly concentrations for the selected date range.
• The “PM2.5 Hourly Patterns” plots the average concentrations for each hour-of-the-day so you can see how PM2.5 changes during the day.
• The “Weekday vs Weekend Patterns” plot shows average concentrations for each hour-of-the-day grouped by weekdays or weekends. The weekday vs weekend plot is only displayed if the data range is at least 7 days.
• The “Seasonal Patterns” plot displays average concentrations for each hour-of-the-day by season (defined as winter: Dec, Jan, Feb, Spring: Mar, Apr, May, Summer: Jun, Jul, Aug, and Fall: Sep, Oct, Nov). The seasonal plot is only displayed if the date range is at last 365 days (1 year).

These plots can help you understand which parts of the day, week, or year have higher or lower concentrations. For each of these plots, the corresponding concentrations from the MDE reference device are also provided in the background to help assess if a neighborhood has better or worse quality than what is measured at the MDE site. Putting in a longer date range will help you understand how PM2.5 changes in Baltimore more generally.

• The “Air Quality Summary” plot displays the percentages of days in the selected date range, where the average concentration in the selected neighborhood falls in each of the AQI categories.

This plot provides understanding of the overall air quality of the neighborhood for the selected date range.

The reference MDE device was at Oldtown until 2021 and moved to Lake Montebello in 2022. If a selected date range includes this time of transition, then we plot both reference values in the background. Hence, when two lines are shown, one represents the measurements at Oldtown until 2021, and the other represents the measurements at Lake Montebello from 2022.

For each of the graphics in the Analytics page, we provide both the modeled neighborhood average and the values from the MDE reference data. For a selected neighborhood and a date range, you can learn about the differences between neighborhood concentrations and MDE reference concentrations in terms of the two time series, diurnal averages, week weekend averages, and seasonal averages by looking at the respective graphics in the Analytics page.

The “Air Quality Summary” plot in the Analytics page offers information on this. For a selected neighborhood and a date range, you can learn what percentage of days the average concentration in the neighborhood was in each of the AQI categories of Good, Moderate, Unhealthy for Sensitive Groups, Unhealthy, Very Unhealthy, and Hazardous.

The Calendar Daily page presents a color coded calendar from Oct 2019 to Jun 2025, with the color indicating the AQI category for the modeled average concentration for the city. This average for each day is calculated by taking the modeled average over all the hours of that day and over all the locations in the Baltimore city.