As with most high quality feeds, limitations on data refresh come from transfer speeds and processing speeds. In best case scenarios, it could take only a few minutes between data capture and data delivery. However, in most cases, for high resolution imagery, it takes a few days.
There are only a limited number of Earth observation (EO) satellites capturing data over the globe. This means not all satellites are always within range of a downlink station. It’s also important to note satellites do not capture data at all times.
Satellites capture hundreds of terabytes of data everyday. Captured data is stored on the satellite until it reaches a ‘downlink station’, where it will be decoded, sent to the data provider, further processed, and sent to its final destination. All of these steps take time.
In the past, weeks could pass between capture and final delivery. Today, the commercial space sector is booming. New entrants are getting into the ‘ground game’, the fall in computing costs is pushing space data to the cloud , and companies like SkyWatch are working hard to automate access to satellite data. This means we can expect times between scene capture and high resolution data transferred to the customer to eventually drop from days to mere minutes.
Let’s dig into the reasons why live, real-time satellite data video feeds, as portrayed by Hollywood, do not currently exist.
How many satellites are watching us right now?
Earth observation satellites are separated between geosynchronous equatorial orbits (GEO) and low Earth orbits (LEO).
GEO satellites are geo-stationary, which means they orbit at the same speed as the Earth’s rotation, thus appear to always be above the same point on Earth. To reach that orbital path, they need to be very high up [roughly 35,786 km (22,236 mi)]. Since they keep a relative position to the Earth’s surface, they can simply transfer down their data as they capture it since they don’t move out of range of the reception dish. However, it also means they are too far up to capture detailed, close-up imagery of Earth. Current spatial resolution for GEO EO satellites makes it hard to distinguish anything smaller than a city. They are most often used for weather monitoring and are considered very-low spatial resolution satellites.
Here is an example of GEO satellite data:
However, most people expect satellite data to look less like what you see on the weather channel and more like Google Maps zoomed in data. However, most people do not realise the highest ‘zoom’ Google Maps often displays aerial photography rather than actual satellite data. The current limit on commercially available optical satellite data is 30 cm, which means each pixel represent a 30 x 30 cm (12 x 12 in) area on the ground (under optimal conditions). Here is an example of what ‘very high resolution’ satellite data look like:
To take sharper images of the Earth’s surface, like the one above, a number of EO satellites are placed in LEO. Their orbital paths take them from pole to pole, ensuring they cover the entire surface of the globe, one strip at a time. Most polar-orbiting satellites take ~90 minutes to come back to the same pole. As the Earth rotates under the satellite, the next imaging strip will be taken over a different swath of ground on Earth.
To image the entire globe, satellites can either take a short number of more ‘zoomed out’, low-resolution images in a lesser number of rotations, or a higher number of ‘zoomed in’, high resolution images, over a longer number of rotations.
A wide covering satellite like MODIS, pictured above, with a swath of 2300 km (or 1400 mi), can capture the entire globe in 1-2 days, at a resolution of 250 m (820 ft). One of the Sentinel-2 satellites, which has been designed to cover 290 km (180 mi) at 10 m (33 ft) resolution, will take 10 days to take a new image of a specific area under the same viewing angle. With both satellites in the Sentinel-2 constellation (S2A and S2B), this revisit time can be reduced to 5 days.
To improve ‘refresh rates’, satellite operators can launch multiple satellites with the same sensor on board to follow the same orbit. This technique, called ‘pearling’ allows operators like Planet to provide SkyWatch, and their other partners, with the ability to obtain medium and high resolution data over the same area every day.
Here is a website that illustrates how Planet’s is able to image the entire globe daily, thanks to their very large constellation (several hundreds of satellites).
How long between satellite data capture and data delivery
Satellites can sometimes spend a few hours without getting in range of a downlinking station. This adds to the time between data captured and data transferred.
Polar orbiting satellites travel at a speed of ~7.8 km/s (4.8 miles/second). When they pass over a downlinking station, they are in range for less than 10 minutes, during which they must downlink all of the collected data since the last pass over a station. Satellite operators must ensure they do not capture more data than they can send in one pass. On top of that, downlink/processing data comes at a cost.
For this reason, many satellite operators choose to only focus on capturing ‘interesting’ scenes and are not capturing data over the Earth at all times, unless they have been tasked to do so by a client. And, as this means an operator will need to update their collection deck to accommodate a specific client, tasking data prices are often higher than archive data prices.
To ensure you receive data for the area you are interested in, it is critical you ’task’ a satellite to capture your area of interest rather than wait for archival data to be made available. If you need data to be collected within the week, we recommend you consider platforms, like EarthCache, that can automatically place tasking orders for your application.
How long does it take to process satellite data
Satellites capture data in a very different format that what can be browsed on the Google Earth engine.
Raw (or ‘level 0’) satellite data are initially processed at the station receiving the data from the satellite, decoded and transformed into usable data, with the correct geotagging, time reference, sensor information, etc. The data are then transferred from the receiving station back to the data provider. This initial step, in itself, could range from a few minutes to up to an hour.
Data are often further processed to turn it into ‘Analysis Ready Data’ (ARD). This includes image sharpening, atmospheric correction, smoothing geographical distortions due to the angle of the sensor, etc. Increasingly, this step is done automatically as manual processing could add days to the delivery date.
Finally, the ARD is stored online and made available. When developers use automated distribution platforms, like EarthCache, this step only takes a few minutes. You can even set webhooks to retrieve the data automatically.
With the recent advances to automate the capture and distribution of EO data, from in-flight pre-processing, to cloud storage and automating the ordering of satellite data, both archive and future, we can expect times between scene capture and high resolution data transferred to the customer to soon drop from days to mere minutes.
If you are looking to automate your access to low, medium, and high resolution satellite data, contact us to request access. If you prefer to only use open (free) data, we also recommend the following sources of free satellite data.