Case Study: GEORG

GEORG is the upcoming very high resolution optical EO system of Germany. It will be in service of the Federal Intelligence Service of Germany (BND). At 20-30cm it not only the first optical VHR system built in Germany it is also one of the best in the world. Obviously whether it is a good idea to do it in one giant leap is a different question. Let’s have a look!

Disclaimer: I have no privileged access to any information about GEORG. I have not spoken to any of the people involved in the project. Nobody disclosed information, the article is based on purely on public information and my own deductions based on it. To avoid misunderstandings I have been very thorough in documenting all my sources.

TL:DR GEORG is the German VHR system. The 3 satellites will offer 20-30 GSD and cost 550MEUR. This article explores the general requirements for VHR systems, technical solutions, other nations approaches, German perfectionism and ends with a recommendation for why we need a responsive space capability which gives us good enough today.


Before looking at GEORG in more detail I will establish some general constraints for high resolution satellites: definition, requirements & challenges.

What is very high resolution

The terms high and very high resolution have shifted over the time. When civilian satellite image data with 10m GSD was introduced by Spot 1 in 1986 this was considered critical by some military observers. In fact even at this resolution Spot-1 data was used to identify previously unknown military installations in 1987 by a study conducted by an NGO [50]. Very High Resolution has thus always been that value that is at the edge or beyond the edge of what is only accessible to military systems. In the last decade this value moved down from 1m, to 50cm and has now settled at around 30cm resolution.

Why do we need the highest resolution

This is straight forward, the higher the resolution the easier it is to identify details on your target. There are however limits to what an satellite based earth observation system can observe and fortunately you most targets of interest are significantly large so that they can also be seen with less that ultimate resolution.

No alt text provided for this image

For the purpose of this you have to differentiate between the ability to detect and the ability to identify. To detect is the ability to sense the presence of an object in a certain area and to identify allows to make a statement of what it is. In above example I have chosen a ball on a green field. On the very left is the original image and then from left to right various increase in fidelity. If the contrast to the back ground is big enough the ball can be seen even if its smaller than one pixel. That said, one pixel for most purposes is the detection limit.

To detect an object it needs to by larger than 3x your ground sampling distance

To identify an object it needs to be larger than 10x your ground sampling distance

Usually latest at 3×3 pixels it becomes clear that there is an object, which you then can go further to identify at 10×10 pixels or more clearly at 20×20 pixels. From this you can deduct which type of resolution you need for your application. As a reference you can either use the NIIRS classification [51] or below table [50]

No alt text provided for this image

or more practical an example given by Maxar for the Legion constellation [37]

No alt text provided for this image

How to achieve very high resolution

In very down to earth terms there are two things that define the maximum achievable resolution: the diameter (aperture) of your optic and the distance from the target.

No alt text provided for this image

Two interesting observation can be made: while most commercial players give their resolution including some form of post processing to enhance resolution, the team that designed KompSat 3, HiROS & GEORG does not. In addition, while the native resolution (GSD) for most players is matched on the diffraction limit of 550nm the telescope builders of GEORG sized it for 630nm. This consequently leads to bigger required telescopes (but also greater sharpness). On the other hand if they had build the telescopes like everyone and used post processing the achievable (stated) resolutions would likely be factor 1.6x higher.

What are the drawbacks

You might wonder why not everyone always goes for the highest resolution. This has chiefly three reasons. First, cost, lifetime and visible area.

Cost (when increasing the telescope to reach high resolution)

From a birds eye perspective most systems follow an s-curve. That means while initially the cost and performance ratio is more or less linear, the closer you come to the edge of what is possible the steeper the curve is. In the end very few percentage performance gain will lead to drastic cost increase. Since the S-curve is individual for each technology (and the proficiency of the user) it means that something that may simple for one person (with experience) may be much more complex for somebody else.

No alt text provided for this image

Above example [52] is taken from the Korean manufacturer Satrec who is educating its customers of the same, that a system with 50cm can cost a small fraction of a system that has 30cm GSD.

All technical systems follow an s-curve: close to the technology barrier the cost grows exponentially to increase performance.

On a more technical level, doubling the resolution (at the same orbit height) will lead to a around 8-10x bigger telescope by mass and volume (using the same technology). The reason is that resolution as described above is directly coupled to the aperture (optic diameter) of the system, doubling it in one dimension will make it factor 2 bigger in the other two dimensions, too. As a result if you would build the telescope of GEORG with the same technology a low cost telescope used in the Skysat (35cm) then the telescope would weigh about 1000-2000kg!

To double the resolution, you end up with a payload 10x as complex

To make matters worse the bigger a structure gets the less stable it becomes (using the same technique). This can be compared to what happens to an ant when it is dropped from a house vs. what happens to an elephant when you try the same.

Technologies that work in smaller telescopes do not work in cutting edge telescope

As a result the bigger telescope (which is difficult enough) needs implementation of new technologies which increase cost and complexity.

Lifetime (when lowering the orbit)

In a 500km orbit a satellite will decay in 5-15 years without any effort. In 400km the satellite might be gone after 12-18 month and at 300km it may be gone in a month.

No alt text provided for this image

Note: above calculation assumes 600x600mm, 100kg, 2.2 ballistic coefficient

The reason is the increase of air drag due to increased density. The difference between 500km and 300km is two orders of magnitude.

No alt text provided for this image

As a consequence a high power propulsion system is required to keep the satellite at low altitudes. In addition the increased atmosphere will also increase the atomic oxygen content which will destroy optical surfaces which makes long time missions very difficult even if you could stay at these orbit for a prolonged time. As a consequence most commercial operators which look for long lifetimes (5-10 years) will actually not go below 450km. This has the added benefit that they stay outside the ISS orbit (400-450km).

Visible Area (when increasing resolution)

The optical design as well as the available sensors mean that only a certain number of pixels fit into one focal plane. For simple optical designs such as Cassegrain (e.g. NUSAT, Skysat, CartoSat) the available pixels that fit inside is around 8-12k (with less pixels in the smaller telescopes). More complex TMA designs like Korsch (Georg, Pleiades, LEOS-100) can support 12k-60k pixels in the focal plane, note that TDI detectors with staggered pixels can make the focal plane look bigger than they actually are. Unlike lower resolution instruments where you can place multiple payloads onto one platform typically VHR instruments are too big to have more than one aperture per satellite. As a result when assuming constant number of pixels (e.g. 24k of HiROS) then increasing the resolution by increasing the size of the telescope would mean a decreased field of view / swath.

No alt text provided for this image

Visible Area (when decreasing the orbit)

When decreasing the orbit both the field of regard of the satellite system as well as the field of view of the payload is reduced. As a consequence more satellites are required to make a 1 day revisit. For example at 700km (classical Pleiades Orbit) two satellites are required for a 1 day revisit at 45° off nadir angle. At 500km typically you need 3 satellites to do the same.

No alt text provided for this image

As shown before the field of view is inversely coupled to the increased resolution, when lowering the orbit thus increasing the resolution will equally reduce the field of view.

Summary: there is no easy way to achieve VHR. Usually it means flying a bigger telescope. The final resolution should be carefully chosen since a too high resolution would mean that your field of view is too small and you are looking for a needle in the haystack.

Practical Design Choices

As usual with optical systems the payload will be the driver for any VHR system. To break it down in the most simple terms an optical system is made out of an optic with a detector. These two key systems make or break a VHR system and so we will look on those first.


There are chiefly three choices that are being made in high resolution earth observation systems. Either two mirror catadioptic (lens & mirror combination), on-axis three mirror anastigmats (TMA) or off-Axis TMA. It can be observed that low cost earth observation systems (NUSAT, Skysat) use smaller two mirror telescopes whereas most state off the art systems use large Korsch (On Axis TMA). The only exception are state of the art systems from new players that usually chose two mirror telescopes in their first VHR generation before moving on to TMA (Satrec DubaiSat 2 & Deimos 2, ISRO Cartosat 2).

No alt text provided for this image

Note: the advantages of a TMA over a 2 mirror telescope is the much better control of image errors (aberration), no color errors, bigger field of view allows bigger sensor, better stray light protection. The advantages of two mirror telescopes are that they can be produced on simple machines found in earth bound astronomy telescopes, given the right choice of manufacturer they can be extremely low cost.

Note: the reason almost nobody chooses off-axis TMA for VHR is the size, they can easily be 2-3x as big as a two mirror or Korsch telescope of the same aperture.

Note: for smaller telescopes often two mirror telescopes with full aperture lens corrector (Maksutov or SC) are being used. Due to the full size lens corrector the mass penalty is too large and they are thus limited to systems <200mm aperture in the Cubesat domain.


There are four types of detectors used in VHR systems. Single frame detectors (e.g. NuSat), multiple frame detectors (e.g. Skysat), pushbroom or line detectors (e.g. CartoSat 2, Eros A) [55] and TDI detectors (everyone else).

No alt text provided for this image

Note: It can be observed that all low cost systems use frame or push frame detectors, and all state of the art systems use TDI push broom detectors. In the past satellites have also been using single line push broom detectors (e.g. CartoSat 2, 2A, 2B & Eros A) but advancing sensor technologies, increased GSD and SNR requirements made everyone else chose TDI.

Note: the image collecting ability of a push broom detector is usually bigger than a frame detector or push frame detector even when digital TDI and no forward motion compensation (FMC) is used. The reason is that those detectors are usually coupled with low cost optical systems (two mirror) that have significantly smaller field of view then the Korsch. Hypothetically it would be possible to combine a wide field Korsch with a very large push frame detector, these multi sensor focal plane units would however be horribly difficult to align and keep stable and consequently be very expensive. The SWIR focal plane [58] of Sentinel 2 is an example but in general it is a very unlikely choice.


Below I have given you a system that shows capability and complexity/cost classes of satellites. The inner circle is lowest performance and cost, which gradually improves going outside. Color coding is as follows, red denotes below acceptable, yellow denotes barely acceptable, green denotes acceptable, dark green denotes desired.

No alt text provided for this image

By using this benchmark I have identified four classes of systems based on their price (and capabilities): extremely low cost, low cost, medium cost and high cost. These classes have similar system choices, performance and cost across different manufacturers.

Four distinct classes of VHR systems can be identified, from ultra low cost to high(est)

On the system side the cost is mainly dominated by the size of the main mirror. Extremely low cost systems have 35cm aperture, low cost has up to 50cm aperture, mid cost has ~70cm aperture and high cost systems have >1.2m aperture.

No alt text provided for this image

It is however an interesting observation that the steps between Low Cost and Mid Cost systems can be very drastic. It should also be noted that the high cost systems seem to have only a modest increase in cost over mid cost systems. This is however only true if you consider that the NRE is split over multiple satellites. To my opinion is very likely that the virtual first unit cost of a Pleiade NEO or GEORG could easily be twice as high.

Very High Resolution Programs

I have chosen the VHR programs of three space nations with comparable budgets and ambitions as Germany. Our neighbor France which spends around 2x as much as Germany annually on space and since decades has a strong focus on optical systems and already has military and civilian systems in the resolution class of GEORG. India, who spends as much as Germany but has a much wider berth due to man power cost advantage and has traditionally a strong focus of satellites for earth applications and whose recently launched CartoSat 3 which today already has the capabilities that Germany is aiming to develop and Korea which has the smallest government program of the three but a thriving commercial program which after almost 2 decades will reach top of the line performance around 2024.

While all other nations take a step by step approach Germany’s takes “one giant leap”

All these three nations have invested over 20 years in VHR systems. They each built more than 5 systems with better than 1m GSD and it took them between 2-3 generations of actually flown systems to reach their current high performance systems at 1.2m. In contrast to that Germany is aiming to build from ground up, a cutting edge system all in just one go.

VHR Program France

Our western neighbor France has a long tradition in building optical satellite systems with very high resolution. The Helios [8] series is the military arm based on the Spot platform. Helios 2A/2B were launched at the turn of the 2000s and offer 35cm resolution which at that time was one of the best of the world. Other than the resolution, and the mass of the satellite all other performance parameters are unknown. France is also a major player in commercial satellite systems for earth observation. Starting from THEOS-1 France has built a very large number of Korsch based systems with around 600-650mm main aperture. Depending on the orbit height this is sufficient for up to 0.5m GSD. The next big step is the jump to around 1.2m main mirror for the civilian Pleiaded NEO and around 1.5m for the military CSO.

France is a leader in optical VHR since many years. After almost a decade of building civilian 0.65m aperture systems they recently went up to 1.2m aperture.

That despite the increase in mirror diameter (2x) the platforms did not drastically increase in size is testimony to the increase of performance of satellite systems made in France.

No alt text provided for this image

Note: the illustration shows key parameters such as aperture (circle diameter), GSD (number in the middle in meter, number of pixels in the detector and whether TDI as well as the orbit height. Since the actual mirror diameters of military systems are unknown they are dotted

VHR Program India

Since the inception of it’s space program by Vikram Sarabhai it has always been stated that India should be “second to none in the application of advanced technologies to the real problems of man and society” [53]. Consequently, their earth observation program is one of their strong points. Since the launch of TES, a military demonstrator for 1m GSD, India has built a fleet of CartoSat 2 satellites that have gradually improved in performace. Recently ISRO built and launched CartoSat 3 [41] which with 1.2m aperture is one of the highest resolution optical satellite systems.

Cartosat-2 had the same aperture than the French (Pleiade & NAOMI based) but they used a less advanced optical design systems. Consequently the achievable swath was smaller.

In the last 20 years India has built a global reputation for remarkable VHR systems

No alt text provided for this image

Where France used TMA since THEOS-1 and could thus fit up to 30k pixels into the focal plane, the Cassegrain & subaperture lens corrector based systems could only support 12K wide sensors. This changed with CartoSat-3 which not only is India’s first TMA based VHR system it also has one of the widest swath in its class.

Note: there seems to be some uncertainty whether later versions of CartoSat (2C-2F) have not only TDI but also increased the swath to 16K pixels. The adherence to a same swath of Cartosat 2 (in 600km orbit) to the later CartoSat 2C-2F could indicate this, it could however also just mean an error in the publication.

VHR Program Korea

Korea has two notable programs. The government KompSat satellites of KARI and the commercial satellites from Satrec. While Kari has bought their payloads from foreign manufacturers, Satrec has developed commercially a full range of VHR systems. With SpaceEye-T Satrec has a very interesting satellite under development that reaches 30cm GSD from 500km.

The Korean Satrec offers VHR systems at a very attractive price & performance.

No alt text provided for this image

History of German Satellite based Reconnaissance Systems

Unlike France, Germany was historically mainly interested in radar based reconnaissance [8]. After participation in the space shuttle radar topography mission in 2000 [11], Germany launched the TerraSar-X built by Airbus which offered 1m GSD in 2007 [12], which was followed by the sister satellite Tandem-X in 2010 [13]. These satellites were civilian but had a dual use component. The first dedicated military system of Germany is the SarLupe system launched in 2007 [14]. SarLupe offers 50cm resolution [7] and consists of 5 satellites built by OHB.

Germany traditionally had a focus on radar systems. Optical systems are new.

Compared to the system that we will be talking about today, both TerraSar-X and SarLupe came at comparatively modest cost. While TerraSar-X was built as a public private partnership where the German Aerospace Center paid 100MEUR and Airbus put in 30MEUR [56], SarLupe on the other hand was 200-250 MEUR for 5 satellites which means that each of the 5 devices cost around 40-50MEUR.

No alt text provided for this image

Why did Germany not build elektro optical systems before?

Under the Schwerin accord from 2002 [8][15] France would provide imagery from the Helios 2 optical satellites (35cm GSD) with Germany and Germany in return would share radar data from the SARLupe system (50cm GSD). In addition there was a common understanding that this load sharing optical in France and Radar in Germany would persist. France therefore would not build radar satellites and Germany not build optical.

There is a image sharing agreement between France and Germany. The implicit work sharing is France builds optical German builds Radar. GEORG breaks with this.

The origins of German optical earth observation satellites

The first German satellites with optical earth observation payload were thus not built by large players but rather at the fringes. First to implement such as system were the TUBSAT [16] small satellites built by the team of Prof. Renner at Berlin Institute of Technology (TU Berlin). Starting with TUBSAT B in 1994 offered a full ADCS (star tracker and reaction wheels) as well as <10m GSD for a earth observation. The TUBSAT satellites were therefore recognized as pioneering in the field of small satellites. Very early on the satellites were also recognized for their potential applications for security and defense [24][25] and thus were the inspiration for systems like Skysat and Blacksky.

No alt text provided for this image

The TUBSAT satellites were pioneering small satellites in high resolution optical systems

The first modern small satellite built by the German Aerospace Centre was the DLR-BIRD [17] satellite in 2001. It was built by the DLR Institute of Optical Sensor Systems (DLR OS) and is a predecessor to the TET and BIROS satellites. The infrared detectors are designed for forest fire monitoring but at 180m GSD (staggered pixels) they are however nothing that could be counted as remotely high resolution.

Optical satellite systems in Germany started with small satellites. All activities originate in or fly key elements made in Berlin.

The first German operational satellite system for commercial earth observation was RapidEye [18]. Here the payload offered medium resolution 6.5m and 77km swath was built by JenaOptronic. The focal plane unit was designed and built by the DLR Optical Sensor Systems (DLR OS). DLR OS was the logical choice as they had significant focal plane experience from designing the WAOS [20] for Mars96 & Mars Express as well as one of the first modern digital high resolution aerial cameras (ADS40) [19].

No alt text provided for this image

KompSat 3/3A a blueprint for Germany? [21][22]

When KARI, started to design a new high resolution earth observation satellite in 2004 [21] they contracted Airbus in Germany to deliver the payload. Airbus in turn went to DLR OS to design the focal plane unit.

Meet HiROS – the spy satellite that wasn’t [1]

While developing the KompSat 3 payload a team at DLR OS started to lobby a German version of the satellite. The first public appearance that I was able to find is a presentation of DLR OS comparing the resolving power of an Hubble space telescope in earth observation use. To my understanding this is a mental exercise towards the question: what is the highest resolution that is possible using today’s technology.

No alt text provided for this image

What I find more interesting than the Hubble arithmetic of this paper is that it has a full spec sheet of the proposed High Resolution Optical System or HiROS.

No alt text provided for this image

It was planned to a have a triplet of 800kg satellites carrying the KompSAT 3 payload with 80cm aperture. With this, from a 500km orbit a GSD of 50cm can be achieved. It is important to note that the given GSD is the actual one, not a value generated in post processing. Most commercial satellites usually over report theirs. As a comparison Pleiades 1st gen usually claims to have 50cm GSD when in fact the satellite only has a 70cm GSD from the chosen orbit.

HiROS was modeled after the Korean satellite KompSat 3/3A whose optical payload was built in Germany

From what I can deduct the armed forces were less interested in this satellite as they preferred their radar satellites including the image sharing agreement with France. A different recipient of this data seemed to be more responsive to the proposal: the elusive German Federal Intelligence Service (Bundesnachrichtendienst – BND). The reason seemed to have been we want to be less dependent from France. [2]

It was the German Foreign Intelligence Service not the armed forces who were interested in the HiROS satellite.

What I am a little puzzled about is that instead of being dependent of France, Germany now was embracing dependence from the USA instead as the leaked embassy cables seem to suggest. Or how else can it be understood that Germany was looking for co-funding on the HiROS satellites in the USA. For all intend and purposes I am personally more willing to be dependent on core defense capabilities of my close neighbor than our admittedly biggest but sometimes distant ally.

The argument for HiROS seemed to be less dependent from France, with that in mind I find it curious that Germany would chose to be dependent from the USA instead.

In any case after the leak of the papers in 2010 France increased political pressure on Germany to not build the HiROS satellite as it was seen as a potential threat to their own upcoming system CSO (successor of Helios 2) [8].

When HiROS became public knowledge in 2010 France was not happy and put pressure on Germany to kill the project.

I believe France was furious especially as the cables seem to have suggested that Germany not only was willing to go ahead despite the Schwerin accord but also decidedly without any involvement of France. As a result of this fallout the project was dead.

GEORG – success in the second attempt

Often in life when you hit a wall but all persons involved still believe in the subject matter there will be another opportunity. As of 2015 there was renewed discussions on a satellite for the BND. As you can see from the picture below OHB presented a mock-up of the HiROS satellite (now named GEORG) at the Paris airshow. In my eyes this is a bit of an odd choice where to go public with it considering how much opposing France was to the project. It would however not be the first time that we Germans showed bad taste in the choice where we announce our major events [28].

No alt text provided for this image

Announcing GEORG at the Paris Airshow does not make for good diplomacy .

Note: you might wonder about the acronym GEORG, this however is a pointer to the patron saint of the BND [29].

How GEORG came into being

Since France was still not happy the solution was found in a classical German fashion: check-book diplomacy. Since France expressed that their saw any German high resolution optical satellite as a threat to their own CSO, Germany calmed these fears by purchasing a stake in the system. Therefore, also in 2015 Germany agreed to pay France 200MEUR [8] or 66% of the CSO3 satellite for around 20% capacity of the three satellite entire system [7].

In classic German fashion the answer whether BND satellite or CSO is: both!

As usual with compromise everyone was (un-)happy with the situation. While the French openly expressed their approval of Germany buying a stake in CSO (while being fully aware that this would likely be the end of the Schwerin accord) German players would see it as a lost opportunity to build an own system. I believe the unhappiness on the German side was quickly overcome when in 2016 it was decided that GEORG would be built [6].


At this point you might think that GEORG is a copy of the HiROS satellite. This however does not align with articles in the press whereas the satellite(s) would have a resolution of the size of a DINA4 sized piece of paper (20-30cm) [39]. In addition the satellite looked different than the version at the Paris Airshow in 2015. I therefore started looking for clues.

No alt text provided for this image

First we can start with the recent image of GEORG [31], which also co-aligns with the model that OHB had hanging from its booth at the IAC2018 (red circle – sorry for the low quality, it was the only one I could find online) the satellite uses the bus of EnMAP.

No alt text provided for this image

That means that the payload compartment has a width of 1.2m [32]. As the picture shows that the payload completely fits the payload bay in its entire thickness that means that the mirror diameter is likely around 1.1-1.2m. This would make sense as a couple of state of the art EO systems today work with similar sized main mirrors (Pleiade NEO [33][34][35], Worldview 3 [36], Worldview 4 [57], Worldview Legion [37], CartoSat 3 [38]).

No alt text provided for this image

That however means that the KompSat 3 telescope is not the blueprint of GEORG and that GEORG has in fact a 50% increased mirror. While looking through the publicly available articles on HiROS I found something interesting. not only does one article of speak of a HiROS high resolution capability (slide 8 – which is blanked out) [1], a later article from the same source denotes a SOA (state of the art?) version of HiROS which is also pegged at 50cm but flying at 950km (and has an oddly similar configuration to today’s GEORG). It is denoted as a possibility to get faster revisit using only one satellite, however since you need 3 satellites from 500km SSO to achieve daily revisit and the satellite miniatures are shown with 3 satellites of that bigger type just at 490km and so I believe what is shown here is the performance of GEORG, just transformed to a different orbit.

It seems to me that DLR published the specs of GEORG already in 2016.

Interestingly, the above illustration was, according to an article already used by DLR in presentations in 2012 [7]. For me this looks like originally there actually might have been a plan for two step approach: built HiROS (80cm aperture) first, then built this larger satellite with 1.2m aperture which we now know as GEORG. This would make sense from an engineering point of view.

HiROS/GEORG might have originally been a two step approach that was abandoned due after the 2010 delay

My guess is that somewhere during the wait between 2010 when HiROS died and 2015 when GEORG was approved the decision was made to forego the “smaller” intermediary step in order to catch up. Considering that Germany had never before attempted such a system, one would think, what could possibly go wrong?

The price of perfection

As with my previous case study on EnMAP we will look into multiple aspects of the price of perfection: delay, actual monetary cost as well as loss of capacity & opportunity.

A delay of (at least) several years

One key element of delay is the habit to do everything in one go. Where other nations gradually increase capabilities over multiple generations of satellites Germany jumps to the top in one giant leap. This did not work with EnMAP and considering the recent news [40] that announced a delay of “several” years it won’t work with GEORG, too.

The one giant leap strategy did not work with EnMAP it won’t work with GEORG, too

In article about EnMAP I therefore advocated to do intermediary steps when you go for a perfect solution as it is not a wise idea try without it. If you don’t do it you end up with a delay equal or longer as if you would have tried these intermediary steps.

My personal bet is that it will take until 2027 until the GEORG satellites are in space

And latest by 2027 we will know whether my assessment from 2017 is true that GEORG will take 10 instead of the planned 5 years.

My bet what caused the delay is that the payload is too big a a jump for the experience of those involved.

On the question what caused these delays, well since there is no public information available your guess is as good as mine. My money is on the payload though.

Cost overruns (this is why its called bleeding edge)

The two GEORG satellites were contracted in 2017 for 400MEUR. This is a large amount of money considering that 5 SarLupe satellites at 250 MEUR cost significantly less.

400MEUR for two satellites.

Satellite three adds additional 150MEUR.

GEORG stands at 550MEUR today.

According to German newspaper the BND was already anticipating cost overruns by 25% in 2018, that is two years into the project [39].

Two years into the project, BND already anticipated 25% cost overruns.

I would thus not be surprised if the actual costs will double its initial estimation (or even higher) when the satellite will finally be delivered

No capacity

GEORG, once launched will be a technological marvel unfortunately until then we have nothing. By choosing to take one giant leap this wait time will be substantial. As said my estimation is 2027 instead of 2022. That means for many more years Germany will have to rely on allies (e.g. our 20% CSO subscription) or commercial data (likely Pleiade NEO & Worldview 4 / Legion). In addition I think it is important to understand that this topic is obviously already under discussion for one decade prior.

When you put all money in one perfect VHR system like GEORG, you will wait a very long time for your capacity.

In addition there will be no money left for things that you need, too, such as a medium resolution wide area detection capability.

I’d like to pose another question: if everyone only builds VHR systems how do you actually know where to point the comparatively tiny FoV of GEORG and on what you are going to use its limited resources? I my eyes, in order to be not dependent on say the planet scope of Planet it would make sense to have a staggered approach where a VHR like GEORG is the cherry on the cake and not the whole meal. For this to happen it is however necessary that not all money is spent in one project.

Lost Opportunities

I’d imagine if Germany would have entered the optical earth observation instead of a “perfect or nothing” approach with more measured initial targets and had gradually improved that over a series of generations then not only would we have spent much less money, had capacity earlier but likely would not have attracted the ire of France as we would have in parallel communicated that, yes we stand fully committed to the Schwerin accord. Maybe over time our own optical activities would have put us in a position to be a valuable partner for CSO or other EO projects in the spirit of European collaboration.

What would I have done different

Before making suggestions what to change it is important to visualize first how much money we are talking about: 400MEUR for the first two and 150MEUR for the third. Therefore 550MEUR total. This will give us a daily capacity of 3500 scenes [39] of 9x9km (~300,000km²) with 20-30cm² [39] pan for a duration (system life) of 10 years [5]. In the following we will see how much this would that get us in less than perfect satellites (which are often good enough).

Note: the scene size can be deduced from the pixels of HiROS SOA (30k) and the resolution of 30cm (which seems to be the limit of the readout electronics) [8].

Note: the resolution of GEORG is stated comparable to a piece of paper [39], a standard DIN-A4 sized paper is 20x30cm

Option 1 – Ultra Low Cost VHR Satellites

Using ultra low cost systems, the 550 MEUR could have given us 400 satellites (60kg) which would naturally decay within ~1.5 years from 400km and deliver us, depending on their current height, images at 40-50cm GSD (post processed like planet).

Note: as the NuSat of Satellogic [59] shows this could actually even built lighter but then you likely have higher cost and would decay faster in the very low earth orbit. I therefore, in the tradition of TUBSAT and because launch prices are really not an issue any more, I have deliberately anticipated the satellite to be 50% heavier than the example set by Satellogic.

No alt text provided for this image

Option 2 – Low Cost VHR Satellites

Or if this is, too outlandish then what about 40 low cost VHR satellites (100kg) which would have used their 50cm optics at 300km to generate 30cm GSD (post processed like Pleiades).

No alt text provided for this image

But wait there is more… since the above constellation would only eat up at 450MEUR you could throw in a constellation of 16 medium resolution (3.5m GSD) wide swath (160km) satellites (e.g. BST LEOS-50MR) to give you global coverage and intra day revisit to better task your VHR capability and then still add 4 IceEye Radar satellites with up to 25cm GSD to add day and night / all weather capability for good measure.

No alt text provided for this image

Option 3 – Mid Cost VHR Satellites

Or if that that’s still to bold, what about 10 mid cost satellites similar in capability to Satrec SpaceEye-X. Or maybe follow the original plan of using a recurrent KompSat 3 payload and use them with the equally flight proven SarLupe bus, just consider them more expendable.

No alt text provided for this image

The point of this exercise was to show what the money that was spent on 3 GEORG satellites could buy as an alternative. It is important to note that not only would the wait time be shorter for these smaller and less complex satellites, and since they are in much higher number the system is inherently more resilient against any form of accident or enemy action.


I am aware that things won’t change overnight but here is what I propose: for any super perfect system we build we set aside, from the beginning, 10-20% of the budget for a good enough capability. These low cost and fast running systems will help to de-risk the bigger ones, can act as technology pathfinders and can replace capabilities if one of the flagships should get compromised. Since DLR has already taken steps into this direction with the founding of the Responsive Space Competence Centre [54] maybe we make use of it.

How can you help:

This text is part of a series of articles in which the author sets the framework to start a discussion about the wrongs of the space industry. If you have experienced similar things, leave a comment. Other views and opinions are very welcome, too, as they may present a way forward. Please be kind to each other.


The author’s views are his own do not represent the views of Berlin Space Technologies.


[1] The Bright Future of High Resolution Satellite – DLR OS 2009

Paper of above

[2] Cost of HiROS 200MEUR, No involvement of France (2011)

[3] No support of HiROS due to problems with France (2011)

[4] Model of GEORG based on HiROS (2015)

[5] Optical Sensors – DLR OS (2016)

[6] BND gets its own satellites on the second try (2016)

[7] A new German space Policy (2016)

[8] the history of the french reconnaissance satellites (2016 & current updates)

[9] small satellite constellation (2017)

[10] Guide to buying image data – UP42

[11] Space Shuttle Radar Topography mission

[12] TerraSar-X Eoportal (2007)

[13] Tandem-X Eoportal (2010)

[14] SarLupe EOportal

[15] European Collaboration in Space

[16] TUBSAT EOportal

[17] BIRD EOportal

[18] RapidEye EOportal

[19] ADS40 first digital aerial camera

[20] WAOS on Mars96 and Mars Express

[21] Kompsat 3 EOportal

[22] KompSat 3A EOportal

[24] TUBSAT B as example for future capabilities of adversaries using small satellite; MIT Lincoln Lab (2000)

[25] Pentagon eyes satellites with Joystick control, SpaceNews – 2009

[26] KompSat 3A focal plane – DLR OS

[27] Pleiade HR EOportal

[28] declaration of the German empire in Versailles – wiki

[29] St. Georg patron saint of the German Federal Intelligence Agency – wiki

[30] Germany not happy with deal on CSO

[31] BND has problems with GEORG

[32] EnMAP EOportal

[33] Pleiades NEO EOportal

[34] Pleiade NEO pre-study

[35] Pleiade NEO

[36] Worldview 3 EOportal

[37] Worldview Legion EOportal

[38] Indian Payload Capacities

[39] Merkels fliegende Augen – Zeit (2018)

[40] BND has problems with the GEORG satellites (2021)

[41] CartoSat 1 EOportal

[42] TES EOportal

[43] CartSat 2 EOportal

[44] CartoSat 3 wiki

[45] SpaceEye-X Satrec Website

[46] SpaceEye-X Cost $60M

[47] From 400m GSD to 0.5m GSD Satrec

[48] Morroco VHR satellite

[49] Goktürk 1 EOportal


[51] NIIRS classification

[52] SpaceEye-X S curve

[53] Vikram Sarabhai

[54] DLR RS3C

[55] EROS A EOportal

[56] TerraSaR PPP

[57] Worldview 4 EOportal

[58] Sentinel 2 EOportal

[58] Satellogic Investor Information






Leave a Reply

Your email address will not be published. Required fields are marked *