Interview with Dr. Steven Tsitas - Cubesat Earth Imaging Constellations

This week I interviewed Dr Steven Tsitas of the Satellite Navigation and Positioning Lab and lead author of the paper, “6U CubeSat design for Earth observation with 6.5m GSD, five spectral bands and 14Mbps downlink.”  This paper has been peer reviewed, and appears in the November 2010 issue of The Aeronautical Journal which is published by the Royal Aeronautical Society. I analyzed the business potential of such a cubesat constellation in a previous post here.

Dr. Steven Tsitas received his BSc(Hons) in Physics from the University of Melbourne, MS in Physics (with Distinction) from California State University Fresno and MS and PhD in Planetary Science with a minor in Astronomy from the California Institute of Technology. His two part PhD thesis title is The effect of volcanic aerosols on ultraviolet radiation in Antarctica and A novel method for enhancing subsurface radar imaging using radar interferometry. After completing his PhD Steven worked as a Management Consultant at Bain & Co. in San Francisco. He recently completed a MSc in Astronautics and Space Engineering at Cranfield University, receiving the Vega Space Systems Engineering Prize for Excellent Performance in Dynamics Related Subjects 2008/2009. His most recent papers detail the system design and commercial applications for an 8 kg, 6U CubeSat that can perform Earth observation missions equivalent to those of current 50-150kg microsatellites, with a corresponding reduction in cost.

And now my conversation with Dr. Tsitas:

Q: Your paper posits the potential of a constellation of cubesat earth imaging satellites capable of performing their job on par with current industry leaders like the European company Rapideye. To fit so much capability into a 6U Cubesat is incredibility daunting. What innovations are you proposing to accomplish this?

Steven Tsitas: I employ several innovations to make such a solution possible:

• Time Delay Integration (TDI) to allow a small imager to collect as much light as a larger aperture; 
• Determining attitude during imaging by rate integration using a Fiber Optic Gyroscope to meet the requirements for pointing stability following from the use of TDI; and 
• DVB-S2 encoding and a three speed transmitter to allow fast downlink from such a small spacecraft.

One of the things that I like about space engineering is you can twist and turn around obstacles to find solutions - it is quite a creative process. However the design process isn't arbitrary, the culture of space engineering is to design to requirements, and if done well every component in the spacecraft can be traced to a top level requirement through a process of step by step logical decisions. Just as a limited palette doesn't limit an artist, this logical discipline doesn't have to limit creativity in the design of spacecraft.

Q. RapidEye produces images in five spectral bands including infrared – does your proposed 6U system do the same?

Steven Tsitas: Yes, it images in the same 5 spectral bands as RapidEye.

Q. RapidEye produces a 6.5-meter resolution image – what resolution image does your proposed 6U system produce?

Steven Tsitas: 6.5 m Ground Sample Distance (GSD), the same as RapidEye.

Q. RapidEye admits their 6.5-meter resolution is not adequate for some commodity customers like those tracking crops like grapes, strawberries, and peanuts. What is the highest resolution (better than 5-meter?) that you believe possible today in a 6U?

Steven Tsitas: Good images aren't just about resolution, but also having good contrast at medium spatial frequencies. This is quantified by the Modulation Transfer Function. I've seen a high resolution image with poor contrast at medium spatial frequencies, and it looked much worse than an image of the same scene with lower resolution but higher MTF at mid spatial frequencies. I could improve the GSD of the 6U CubeSat design at the expense of contrast, but this wouldn't necessarily give better or more useful images. Fundamentally resolution is limited by aperture size, and the 6U CubeSat design has an 89 mm aperture imager. Given the 6U CubeSat is just 100 mm thick this is obviously close to the limit. Short of some kind of foldable optics or deployable membrane mirror technology I don't think you are going to do much better than that with 6U.

Q. RapidEye’s satellites are designed to last seven years – your research indicates a 12 year satellite life per 6U. How would the orbital life of each satellite change by offering the RapidEye service of a photo anywhere on earth within 24 hours?

Steven Tsitas: To be clear, the paper indicates that the orbital lifetime could be 12 years, and in fact could exceed 25 years requiring a deorbit device, for which provision is made in the design. The orbital lifetime is not necessarily the same as the operational lifetime. Regarding the effect of imaging operations on operational lifetime, the 6U CubeSat design does not include propulsion or any consumables, so there is no direct link between a particular imaging campaign and the operational lifetime of the spacecraft.

Q. In a recent post, I speculate on the economics of a such a 6U cubesat constellation. What further are you planning in this area?

Steven Tsitas: I discuss the commercial implications of the 6U CubeSat design in an upcoming paper. Standby.  Perhaps we can continue this conversation after the release of the new paper.


Thank you Steven. Yes, let’s talk again with the release of your paper on the economics of such a cubesat system.

Business Case for a CubeSat-based Earth Imaging Constellation

The use of Commercial Earth Imaging Satellites is growing. Individuals, corporations and governments are finding varied and unique applications for images of our planet.

Futron estimates the market for commercial earth imaging topped $1B last year (2010).


Uses of Earth Imaging:

• Disaster Relief – think of all of the satellite images you saw after the Japan Earthquake (including the nuclear reactors)
• Disaster avoidance - George Clooney (among others) paying to patrol boarder of north and south Sudan using Earth imaging satellites.
• Helped with hunting down Osama bin Laden (but were any these images from commercial satellites?)
• Food Commodities tracking – allowing traders to ask and answer questions like, “how do the wheat crops in Kansas look after last night’s hail storm?”
• Remote Infrastructure observation – the oil industry uses it to keep track of their assets in remote locations
• Even the US Government is turning to Commercial providers. Last year, the U.S. National Geospatial-Intelligence Agency (NGA) awarded separate 10-year, $3.5 Billion contracts to image providers DigitalGlobe and GeoEye (these contracts are now under review).

The Commercial earth observation markets:

1. Market #1: High-Resolution images (1.5 meters per pixel). But the cost of each satellite means providers have a limited number of satellites (usually 1-2) on orbit.
2. Market #2: Med-Resolution images (5-7 meters per pixel) – lower quality images, but providers tend to have more satellites in orbit and may offer more spectral bands to choose from for each image and offer more frequent photo opportunities due to the higher number of satellites within the constellation.

In a recent Nov 2010 paper, “6U CubeSat design for Earth observation with 6.5m GSD, five spectral bands and 14Mbps downlink,” author, Dr. Steven Tsitas outlines how a constellation of 6U CubeSats could serve Market #2 (frequent med-res images) competitively. (Sorry, I think you will have to buy the paper. If a reader finds a free version of the paper online, let me know and I will change the link). I hope to post an interview with Steven Tsitas soon.

But why even consider a CubeSat at all for such a mission? Here are just a few of the advantageous of using CubeSats:

• High amount of innovation in the field – from NASA, universities, and private industry
• Low ITAR restrictions (CubeSat programs are thriving in many nations)
• Low mass of each satellite
• Reduced launch cost per satellite
• Reduced cost to replace/upgrade constellation as satellites age, breakdown, or new technology becomes available

Rapid Eye, a German company, is the current leader serving Market #2. Below I will provide some details about Rapid Eye and how a CubeSat constellation might be able to compete with Rapid Eye.  First, a little education about Rapid Eye.

Rapid Eye Details:

• Five identical sun-synchronous Earth observation satellites
• Five spectral bands
• Launched in August 2008
• Satellites built by Surrey UK
• 650KM circular orbit
• Captures 4mil km squared of earth’s surface every day
• Once an order is placed for an image, can take a photo of any location on earth (between 75 degrees N and 75 degrees S) within 24 hours.
• Offers not only images, but offers services for the analysis of images – especially good at providing comparative analysis of images taken over a period of time

Rapid Eye, the Numbers:

• Customer price for images: $1.33 per square KM (must purchase 5,000 KM at a time (at current Euro conversation rates that is equal to $6650 per very large image)
• Satellite Constellation construction: $35M 
• Expected 2009 Revenue: $29.5M (have not confirmed this number)
• Total Capital needed to break even: $224M

Assumptions about Rapid Eye’s business:

• Assumed Rapid Eye is now profitable
• Assumed the cost of the single Dnepr launch necessary to lift the five Rapid Eye sats: $15M
• Assumed a $50M infrastructure Hardware purchase (ground station and other startup infrastructure)
• Assumed a five year startup at a cost of ~$25M per year in operating (non-HW, non-infrastructure costs)

So what if we could launch a constellation of ten cubesats that could perform a very similar function as Rapid Eye’s current constellation of five small sats? Are their savings if we could? For this post, I will use Steven Tsitas’s conclusions that, yes, such a cubesat constellation would be technically possible.

I will build my business case, not from a technology discussion, but by attempting to answer the business question of - how much could an business save by using Cubesats instead of small sats?

CubeSat Venture Assumptions:

• Cost per 6U CubeSat: $400,000
• Number of CubeSats in constellation: 10
• 6U CubeSat mass: 8 lbs each
• Falcon 1 launch: $9.8M
• SpaceX willing to prorate launch cost based on mass

If we assume the CubeSat venture would operate using the same Hardware and Operating Costs as the Rapid Eye venture, then the CubeSat savings are limited to the cost of the satellites themselves and the cost to launch them into orbit:

• Rapid Eye’s satellite and launch costs: 23% of breakeven costs
• CubeSat venture’s satellite and launch costs: 3% of breakeven costs

This would mean a CubeSat venture competing with Rapid Eye could theoretically lower image prices by twenty percentage points over competitors (all other things being equal). This by itself may close the business case for some CubeSat constellation investors.

But perhaps competing toe-to-toe with Rapid Eye is the wrong business model. As a general rule, it is hard to out Wal-Mart, Wal-Mart. What-if the CubeSat earth imaging venture could, instead, become the low-price, no frills, earth imaging provider?

In the earlier example, the CubeSat advantage was limited to lower satellite costs and cheaper rides to orbit on SpaceX launch vehicles. But what-if the venture could also save money on ground costs: Hardware/ground stations and operating expenses?

CubeSats, the low-cost leader in earth imaging Assumptions:

• Continue with assumptions regarding low satellite costs
• Continue with assumptions regarding low launch costs
• Lower ground Hardware and Infrastructure costs from $50M to $25M
• Lower operating costs from $25M to $10M per year.

Here is a quick cost comparison between the options:


















Next Questions (beyond the scope of this post):

• Market price elasticity: How price sensitive is the earth imaging market? How would cutting Rapid Eye’s price by 20-60% affect demand for a CubeSat-based image product?
• What realistic cost reduction methods are possible in ground hardware and personnel?
• Admittedly, my Rapid Eye information was limited to publicly available data, a more serious effort should be conducted to understand the competitor’s cost structures and current profit forecasts
• What are the cost implications from using a CubeSat-based system? Where are system costs reduced? Where are system costs increased?
• Admittedly, images from a CubeSat are of a lower quality than the best in orbit (5-7 meters per pixel compared to 1.5 meters per pixel from the industry leaders of market #1).  How sensitive is the market to image quality?  And what can be done to increase the quality of an image taken on a 6U CubeSat?