TEMPO: Reading an Hourly Pollution Map Without Mistaking It for a Surface Monitor

Place a morning pollution map beside an afternoon map of the same city and their colors may differ. Understanding that difference begins with a less visible question: what quantity does the map represent? An atmospheric column observed from space is not the same measurement as air sampled beside a pavement. TEMPO's innovation does not erase that distinction. It provides more frequent observations of changes across the same broad region. Those observations become meaningful when time, vertical extent, and processing version are read together.

TEMPO is an ultraviolet and visible spectrometer observing North America from geostationary orbit through a collaboration between NASA and the Smithsonian Astrophysical Observatory. Its first Earth observations began in August 2023. A contribution presented by the research team at the 2026 European Geosciences Union meeting reviews the first two years of nitrogen dioxide and formaldehyde products, algorithm developments, and delivery latency. This account uses a knowledge cutoff of 13 October 2026. Conference findings are not presented as universal performance rates independently established for every region. [1]

An hourly map does not represent one instant

A geostationary viewpoint permits repeated regional observations, not simultaneous continental measurement. The technical document describes a nominal scan progressing east to west through mirror steps over approximately an hour. An hourly label does not mean every pixel was acquired in the same second. [2]

The distinction matters when interpreting a moving pollution plume. Changing shapes across successive scans can provide information about transport and dispersion. But observation time also matters when comparing opposite ends of a single scan. The map can record evidence of movement without being a perfectly frozen view of the continent. Daylight operation is another fundamental feature: the instrument's input is sunlight scattered back from the Earth system. It is not a nighttime camera photographing combustion. [1][2]

Finding a gas in the spectrum

TEMPO does not draw air into a container and count its contents. It uses the wavelength-dependent structure of light. The team's retrieval begins by estimating a slant column for each spectrum, representing gas along the light path. An air mass factor then describes the relationship between that path and the desired vertical column. Viewing geometry and the gas's vertical distribution thus enter the same interpretation chain. This is more involved than treating a darker patch as a self-explanatory measure of pollution. [1]

For nitrogen dioxide, additional processing separates contributions from the stratosphere and troposphere. These are different atmospheric regions; their contents should not automatically be assigned to the same surface source. The team also applies a correction for striping in the nitrogen dioxide retrieval. That addresses structures introduced by the measurement and processing system rather than by the atmosphere. A correction is not simply the removal of an interesting-looking pattern. It is an attempt to identify more accurately what the observation represents. [1]

A column is not breathing-height concentration

A column expresses gas amount above a unit area, not breathing-height concentration. Different vertical distributions can give equal columns but different surface conditions. A pixel does not directly yield personal dose or one street's air. [2]

Recognizing this difference does not diminish the satellite's value. It establishes its domain of usefulness. Investigating how gas above a city evolves through the day is not the same task as measuring a person's exposure along a particular journey. Reading the map as a personal exposure table converts broad coverage into an unjustified claim of local certainty. When comparing a local measurement with a remote observation, their quantities and times must first be made compatible. The two sources of information need not be rivals simply because they describe the same city.

The observation remains; its processing changes

The team's 2026 presentation emphasizes the transition from version 3 to version 4. This was more than adding a new number to a filename. Updates included correcting a radiance bias to reduce overly high cloud fractions, improving surface reflectance over land, and applying nitrogen dioxide destriping. Ancillary information and models also participate in calculations of how light travels through the atmosphere. A better measurement product does not always require replacement hardware. Changing the method that extracts information from the same spectrum can change the result. [1]

An announcement from NASA's data center on 18 September 2025 described the release of standard version 4 and near-real-time version 2 products. Version 4 observations were available from 17 September onward while earlier records were being reprocessed with the updated algorithms. Processing version is therefore information separate from observation date. A day measured in the past can later be published again through a different processing chain. [3]

For a long time series, this is not merely clerical detail. Someone investigating changing pollution must examine whether the files being compared were produced consistently. Otherwise, an algorithm change may be confused with an atmospheric change. Conversely, disagreement between an older and newer file does not itself establish that the newer one is wrong. The relevant comparison needs to identify which processing approaches generated the values. A timestamp alone cannot carry that information.

Near real time is not zero latency

The 2026 presentation reports approximately three to four and a half hours of latency for standard products and two to three hours for near-real-time products. These are the product delays reported in that presentation, not a guarantee of uninterrupted delivery for every future file. Hourly observation and delivery latency are different clocks. One describes how often the atmosphere is sampled; the other describes when that sampling becomes usable by someone receiving the data. [1]

For near-real-time formaldehyde, using a radiance reference from the previous day can introduce differences from the standard product. The team's example comparison shows a strong relationship between products, but correlation does not establish identical values or the absence of bias under all conditions. Formaldehyde and nitrogen dioxide also differ in their retrieval details. Confidence in one product should not be transferred automatically to another simply because they come from the same instrument. [1]

The distinction is useful even without performing a technical retrieval. A map described as hourly and near real time may sound like a live display. The actual description contains a sampling interval, a processing delay, a measured quantity, and a particular algorithm version. Keeping those elements separate avoids expectations that the system was not designed to meet. It also makes the genuinely new temporal information easier to recognize.

What the evidence establishes

TEMPO's strength is not adding more colors to a map. It turns the daytime development of the same region into a repeated observational record. Doing so requires reading gases' spectral signatures, separating atmospheric contributions, and documenting changes in processing. The team's updates show development of that chain; they do not independently demonstrate error-free air-quality measurements at every surface location. Reading column type, observation time, and product version alongside the colors is the sounder approach. An hourly map then becomes a powerful record for the questions it can actually address, rather than a promise enlarged by questions it does not directly answer.

Sources

[1] Gonzalo Gonzalez Abad and colleagues (2026), First two years of TEMPO nitrogen dioxide and formaldehyde observations: algorithm status and highlights, EGU26-23274. Abstract and citation; author-uploaded research poster. Original explanation adapted with attribution from this work under CC BY 4.0; figures are not reproduced. Conference supplementary-material licensing terms.

[2] Caroline R. Nowlan and colleagues (2025), TEMPO Nitrogen Dioxide Retrieval Algorithm Theoretical Basis Document, v1.0. Technical document.

[3] NASA ASDC (18 September 2025), TEMPO V04 Standard and Near Real-Time V02 Level 1, 2, and 3 Product Release. Institutional data announcement.

Source access and factual review: 15 October 2026. Knowledge cutoff: 13 October 2026. The EGU abstract, supplementary poster text, and release announcement were directly read; relevant sections of the technical document were examined without claiming a complete reading. Raw data, poster graphics, and validation campaigns were not independently reanalyzed.