Cubesat optical payloads: 7 Earth Observation Bottlenecks

Thys Cronje

Thys Cronje

Chief Commercial Officer

People frequently compare Cubesat optical payloads with their significant counterparts. Unfortunately, due to several bottlenecks, this comparison is unfair. This blog looks at the bottlenecks experienced by Cubesat optical payloads, the challenges regarding that and possible ways to address these challenges.

Cubesat optical payloads
Cubesat optical payloads - the xScape200 and xScape100

In June this year (2021), we celebrated 6 years of Sentinel-2A in space. Less than two years after the launch of Sentinel-2A, ESA launched the second one, Sentinel-2B. These satellites form part of the Copernicus Programme to create a global continuous, autonomous, high quality Earth Observation capacity. Today, we are becoming more and more reliant on the continuous streaming of EO data to the earth to understand better and manage our activities on the planet.

In the same time frame, commercially minded start-ups are launching hundreds of smaller satellites into space. Most of these satellites fall in the nanosatellite class, or Cubesats. Some of these satellites are impressive. For example, we are seeing GSD approaching 1 meter, daily revisit times and hyperspectral capabilities.

The smartphone dilemma for Cubesat optical payloads

The challenge is that the man in the street constantly expects more while the EO mission operators want to meet these expectations with “less”. The conundrum is that smartphones have become our point of reference. They outperform nanosatellites in storage capacity, operational efficiency, communication bandwidth, processing power, image resolution and form factor by far.

One can capture a Kodak moment nearly anywhere in the world with a smartphone. Furthermore, you can filter and edit it, and share it instantly with all your friends across the globe. Unfortunately, many expect the same kind of functionality from a satellite multiple times the size of a smartphone.

Yes, over the last 20 years, smartphones have significantly impacted how we live and interact with each other. In addition, it raised our expectations of what is possible to the next level. However, one shouldn’t lose sight of the fact that much energy was spent to resolve all the bottlenecks to make the use of smartphones as seamless as possible. The smartphone is only a tiny part of the massive moble communication ecosystem. The mobile network operators build this instant and seamless way of communicating on a global network of cell towers, optical fibre networks, communication switches and agreed upon data exchange standards.

To put it into perspective, let us look at some of the bottlenecks experienced by Cubesat optical payload manufacturers.

Addressing the Earth Observation bottlenecks for Cubesat optical payloads

Bottleneck #1 - Potential to capture photons

The performance of a Cubesat optical payload, as with any other electronic imaging system, is a function of its ability to capture photons and as many of them as possible during the integration time. Again, when looking at smartphone cameras, we are becoming used to extremely smooth 12 Mpixel or more images with depth, sharpness and image quality comparable to any high-end camera. One shouldn’t forget that these smartphone cameras can automatically adjust the integration time, depending on the light conditions, to obtain the best quality image. In many cases, the object is a few feet away in full sunlight, helping a lot to make your friend jealous.

On the other hand, when looking at a Cubesat optical payload orbiting about 500km above its target at a speed of about 7km/sec, you do have a little bit less than 700us to capture one image. So there is not much time to accumulate photons. And, when using a 5um pixel pitch sensor, you are limited to about 20k electrons well depth, making your system shot-noise limited. Therefore, if you can fill up the well with electrons during the integration time, the best SNR achievable is still less than 140. However, the reality is that you can only accumulate a few thousand electrons during this short integration time.

The trick is to increase the relative exposure time of the Cubesat optical payload, and this can either be done by slowing the relative ground speed by using forward motion compensation or by using time delay integration technologies. Of course, one can also address this bottleneck by increasing the relative aperture or decreasing the f-number, but this has volume, mass, cost and manufacturing complexity constraints.

Bottleneck #2 - Cubesat optical payloads' ability to transfer spatial detail

The image detail that smartphone cameras can capture is incredible. Even when zooming in, you are still able to see a lot of contrast and features. We can attribute the ability to capture this extraordinary amount of spatial detail to several factors: the very low f-number, in many cases less than f2.0, the tiny pixel size, autofocus and image stabilization mechanisms.

For Cubesat optical payloads, there is a real bottleneck in the amount of contrast that the imaging system can transfer from the object to the sensor. The Earth Observation specialists describe these bottlenecks in terms of Modulation Transfer Function (MTF). MTF is a function of the atmospheric conditions, satellite movement and vibration, thermal gradients, aperture size, optical design, sensor selection and manufacturing quality. And, don’t forget, manufacturing imperfactions.

In reality, the biggest bottleneck of Cubesat optical payloads to transfer spatial detail is the optical systems diffraction limit. The diffraction limit is the theoretical limit of performance and is a direct function of the wavelength and inverse proportional to the diameter of the entrance pupil.

A previous blog discusses these limitations and bottlenecks in detail.

Bottleneck #3 - Capacity to capture spectral detail by Cubesat optical payloads

Capturing beautiful pictures to impress your family and friends is one thing but extracting information for decision making is an entirely different ball game. Smartphone images are only in three colours, red, green and blue. Excellent image reconstruction algorithms interpolate the captured RGB value to assign each pixel in the image three values.

Although Cubesat optical payloads sometimes employ commercial RGB filters, payload manufacturers often utilize complex multispectral and hyperspectral filter configurations. These filter configurations directly impact the complexity of the focal plane and the amount of data generated. In addition, the increase in spectral bands impacts the speed at which the data needs to be written to memory, which is usually a significant bottleneck.

Bottleneck #4 - Cubesat optical payload duty cycle

The frequency at which you can use a Cubesat optical payload in space is limited. The thermal environment of the instrument mainly determines this limit. As the only way to manage the thermal environment on a Cubesat optical payload is by conduction to the structure of the satellite, thermal management is tricky.

The instrument is consuming power, and the electronics transform this power into heat. Therefore, there are temperature limits in which one can operate the detector. As the optical payload is exposed to direct sunlight when operated, thermal gradients are induced. These gradients do have a direct impact on the optical stability and performance of the Cubesat optical payload.

Typically the duty cycle can vary from a few minutes to half an hour during an orbit. The main determining factor for the duty cycle is the thermal design of the optical payload.

Bottleneck #5 - Image downlink speed

Connectivity between a satellite circulating the earth at an orbital height of a few hundred kilometres and a ground station is limited. Of course, this has a direct impact on the ability to download large amounts of data. For instance, capturing 1 Tbit of data takes about 3 hours to download with a 100 Mbit/s X-band downlink. On top of that, with only one ground station and 20 minutes downlink time per day, it may seem like forever to download all the data.

On the other hand, since you are nearly always close to a cell tower or a wi-fi, smartphone operators do not experience this bottleneck. Moreover, we are expecting constant connectivity as a minimum requirement. In space, this is not the case. Not yet.

One way to address this downlink bottleneck of Cubesat optical payload is through intelligent onboard processing. Various methods exist to filter through the large volumes of data and only select the data of interest to download. Some call it intelligent automation of Cubesat optical payloads. However, due to limited electrical and processing power on Cubesats, this is easier said than done.

Bottleneck #6 - Image distribution swiftness

Getting your hands on the correct Earth Observation data, where and when you need it, is easier said than done. There are multiple providers, and the procurement process is quite daunting with cumbersome licensing arrangements. Furthermore, prices vary, and the process from enquiry to receiving the data can take a while. In short, although various companies are addressing these issues, Earth Observation data is not always available at the press of a button.

Additionally, larger satellites operators with deep pockets already addressed the data correction and calibration bottlenecks that exist in the process of transforming the data from raw images to a proper GeoTIFF format. Unfortunately, this is not always the case for smaller Cubesat optical payload operators.

In many cases, a data correction, calibration and distribution strategy is the last thing a new Cubesat optical payload operator addresses, and frequently no budget.

Bottleneck #7 - Imager processing time

Over the last ten to twenty years, the image processing landscape changed dramatically. This change is driven by new AI/LM computing methods, readily available computing power in cloud computing facilities and advances in big-data handling. However, the Earth Observation downstream segment is fragmented, with the more prominent EO data processing companies only relying on the established data providers as a source for data.  

Geospatial analysis and application developers are scarce resources. Furthermore, the new kids on the block are struggling to get their data in the hand of application developers and large scale adoption to the, sometimes, less valued data.

On top of that, prioritizing data processing and analytics is also a challenge. In many cases, the smaller players are required to stand in the back of the queue. However, when timely insight within specific domains, like agriculture, is required, a few hours can significantly impact the outcome of decisions.

These bottlenecks may not be a direct consequence of the Cubesat optical payload, but it directly impacts the conceived value that this industry can deliver. Nevertheless, we are seeing downstream data companies addressing these issues, but it remains the responsibility of the Cubesat optical payload operator to engage with these service providers early on. They can make or break an Earth Observation data supplier business.

The Bottom Line:

The Earth Observation bottlenecks for Cubesat optical payloads are real and there is ways to overcome it. But, then again, these solutions do have limitations, and it is the responsibility of the mission operator to understand these bottlenecks early on. Therefore, the best recommendation that we at Simera Sense can provide is to engage with all the various stakeholders along the value chain as early as possible.