What you're reading: We built Luniter because we love science and want to know what's coming next, and which companies are actually building it. We track 200 of them. Every Monday we pick the one story from that world we think is worth your time and explain it in plain English. It takes about ten minutes. If you're new here, welcome.

Eyes Above the Sky

Imaging satellites give us a way to watch landscapes change. Their pictures can help people track vegetation, follow water across a region and examine damage after a disaster. Turning those observations into useful information starts with understanding what a satellite can actually see.

Different jobs make different demands. Some need a broad, frequently refreshed view. Others need enough detail to distinguish features within a single property, or a way to observe when an ordinary camera cannot get a useful picture. This week, we’ll explore the technology behind the satellite-imaging businesses we track at Luniter. Civilian applications will help us understand how their instruments work, the problems their products solve and the limits of what they can deliver.

Start with the broad view. Planet Labs flies fleets of small imaging satellites called Doves. Each photographs a narrow strip as it circles Earth. Earth turns underneath, bringing new ground into view on the next pass. Many cameras, working together, can build a frequently refreshed picture of a landscape much larger than any one image.

Earth carries earlier image strips away from the next pass. Gold follows one Dove and its fresh strip, which fades into blue. This simplified flock illustrates the motion, not actual fleet size or complete daily coverage. Original Luniter illustration; orbit geometry informed by NASA.

A flock of Doves

A Dove contains a telescope and an electronic detector. Light reflected from the ground enters the instrument and becomes a set of measurements. Solar panels provide electricity, and a radio transmits the data to receivers on Earth. The spacecraft has to collect the observation and get it home before anyone can use it.

An exterior schematic based on Planet Labs’ 2019 B14 Dove. The labels identify the optical instrument, power supply and radio link. Orange highlights the antenna for visibility; this is not an engineering blueprint.

New Doves are launched into orbits about 525 kilometers above Earth. Being relatively close helps their small telescopes see useful detail. Move the same camera twice as far away and a house appears only half as wide. A wider telescope opening can distinguish finer features from the same distance.

Lower orbits bring stronger atmospheric drag, which can shorten a satellite’s working life. With the same camera pointed straight down, they also narrow the strip of ground in view. Altitude is a compromise between detail, coverage and lifetime.

A Dove photograph spans tens of kilometers of ground. A field occupies many image pixels; a house occupies far fewer. Planet Labs supplies map-corrected PlanetScope scenes on a three-meter grid, so each pixel represents a patch of ground about three meters across.

On the ground, those observations need to be placed on a map. Overlapping pictures share features that help align them, so a road continues across the join instead of jumping sideways. Combining the pieces produces a larger view, called a mosaic. A river crossing several pictures becomes one continuous river in the assembled map.

Overlapping crops of one fictional landscape form a mosaic 4.8 kilometers across. The source drawing uses three-meter samples. These are cropped portions, not complete observation strips. California is a locator, not the map shown. This illustrates assembly, not a satellite’s measured image quality. Original Luniter illustration.

Repeat the observation and the mosaic becomes a way to follow change. A utility planning vegetation work can compare growth around its network and combine the imagery with local information about power lines and terrain. A broad view helps narrow down where to investigate, even when it cannot identify every branch that needs cutting.

The measurements extend beyond the colors our eyes see to near-infrared light reflected by the landscape. Vegetation reflects and absorbs different wavelengths in ways that give researchers information about its condition. Those patterns can help track a changing forest. They do not, by themselves, forecast where the next fire will start.

A mosaic also has a time dimension. Its pieces may come from different days, especially when processing selects clear observations to replace cloudy ones. That is useful for a broad view of the landscape, but it matters if someone needs to know what changed this morning. A new orbit does not guarantee a usable new picture. The camera may pass over the right place and photograph a cloud.

Radar brings its own light

A radar satellite can get a view beneath that cloud. It sends microwave pulses toward the ground and measures the signals that return. Those wavelengths can pass through clouds, and the instrument does not need daylight. This is a different kind of observation from the reflected sunlight collected by a Dove.

Processing the measurements produces a picture with an unfamiliar appearance. Calm water often looks dark because little energy returns toward the satellite. Rough surfaces and some arrangements of buildings send back stronger signals. Brightness depends on the surface, moisture and viewing geometry; it is not ordinary photographic color.

The fictional lower comparison shows ground hidden in the optical view but visible in radar. Radar brightness represents returned signal strength, not natural color or height. Many measurements are processed into an image. Original Luniter illustration, informed by NASA’s SAR explanation.

The result shown here is a two-dimensional map, not a 3D model of the town. A NASA radar image of Seattle through clouds demonstrates the practical benefit. Radar provides another way to observe the surface when an ordinary camera would show the cloud above it, although interpreting the picture takes knowledge of how those echoes behave.

Heat can stand out before a shape does

Finding a small fire calls for another measurement. The near-infrared light used to study vegetation is reflected sunlight. A fire-focused infrared instrument can measure thermal radiation coming from hot ground and flames, including at night. Clouds can still interfere with that observation; thermal infrared does not inherit radar’s ability to see through them.

A sufficiently hot area can contribute a strong signal even when it occupies only part of a pixel’s ground sample. The instrument may therefore detect a fire whose outline remains unresolved. NASA’s active-fire guidance distinguishes detecting a hot spot from establishing the exact size or position of the fire within its pixel.

A small hot area changes the signal from a larger ground sample. The bright cell does not trace the fire’s boundary. Colors and sizes are illustrative; detection depends on the sensor, fire and observing conditions. Original Luniter illustration.

That distinction makes thermal measurements useful early in a fire, when resolving its shape would require a much more detailed image. Other hot surfaces can also affect the signal, so detection software needs context to decide which measurements to flag as fire.

This is the job FireSat was built to address. Earth Fire Alliance and satellite builder Muon Space released the first active-fire images on September 16 from three satellites launched in July. The satellites were still being checked and calibrated in orbit, with routine delivery to early users planned for later this year.

There is a familiar Luniter name involved, too. Google Research is a founding partner in the nonprofit Earth Fire Alliance, and Google.org helped fund the early FireSat satellites.

After the fire, a closer look

After a fire, the question may move from detecting heat to examining a single property. An image can show that a block has been damaged while leaving important details within it unresolved.

If part of a roof and part of a driveway fall inside one pixel’s ground sample, their signals mix. Smaller samples can separate features that a larger sample combines. Planet Labs uses different instruments for these jobs, including its Pelican satellites for finer imagery.

In the illustration below, the coarser view already shows useful damage patterns. The finer view adds something within the properties: a remaining section of roof and a narrow strip of debris across a driveway. Both panels show the same ground at the same scale.

One fictional neighborhood sampled at 3.7 meters and 50 centimeters per pixel. Finer cells separate roof and driveway details. Only sampling changes; optics, atmosphere and noise are not modeled. This is not vendor imagery or a damage or access assessment. Original Luniter illustration.

Smaller pixels are one ingredient in a useful image. Contrast matters too, along with the instrument’s optics and the atmosphere between it and the ground. A finer grid cannot guarantee that a roof is safe or a driveway passable.

There is also a way to reach a place that lies beside the flight path. A steerable optical satellite can point toward it without passing directly overhead. That can create an earlier opportunity for a picture. The tradeoff is that an angled view spreads a detector pixel’s field of view across more ground, potentially blending details that a more direct view would separate.

A view from above shows the satellite’s path beside the target town. Steering reaches across the gap. The drawing is schematic, with no promised revisit time or image quality. Original Luniter illustration; NASA explains how viewing angle changes ground samples.

The businesses behind the pictures

Planet Labs and BlackSky sell access to observations, software that helps interpret them, and systems built for individual customers. Planet Labs has its broad mapping fleet; BlackSky emphasizes monitoring selected locations. Both must turn demand for information into enough cash to build, launch and replace expensive hardware.

Their latest reports show rapid sales growth, alongside continued losses. Planet Labs generated positive free cash flow over the six months ended July 31, even after investment in equipment and software. BlackSky’s June quarter still had a substantial loss relative to its revenue. Both also raised money by selling shares, financing expansion while spreading ownership across more shares.

Planet Labs’ growth also includes satellite hardware deliveries. Those can lift a quarter’s sales without becoming a subscription that renews each year, so the mix of business matters alongside its growth. Growing sales alone do not settle the investment case.

At the September 17 close, PL was about 67% below its highest closing price of 2026, and BKSY about 56% below its own. A lower share price makes the comparison worth revisiting. It does not tell us whether either business is cheap, or how much future growth shareholders will ultimately keep.

The customer base points toward another use for everything we have discussed. Watching a landscape change, returning to a location and examining fine detail are valuable capabilities in war. Defense and intelligence supplied about 70% of Planet Labs’ latest quarterly revenue. Almost all of BlackSky’s June-quarter revenue came from governments, though that broader category is not a defense-only total.

How armed forces use these systems, and what happens when the satellites themselves become targets, deserves its own issue.

Disclosure: I own shares of Planet Labs (PL) and BlackSky (BKSY).

Anthony

The Luniter Observatory

Research and educational analysis only — not financial advice. Every investment carries risk.