Doing Battle with the Sun: Lessons From LEO and Operating a Satellite Constellation in the Elevated Atmospheric Drag Environment of Solar Cycle 25

Abstract

Capella Space, which designs, builds, and operates a constellation of Synthetic Aperture Radar (SAR) Earth-imaging small satellites, faced new challenges with the onset of Solar Cycle 25. By mid-2022, it had become clear that solar activity levels were far exceeding the 2019 prediction published by the
National Atmospheric and Oceanic Administration's (NOAA) Space Weather Prediction Center (SWPC). This resulted in the atmospheric density of low Earth orbit (LEO) increasing 2–3x higher than predicted. While this raises difficulties for all satellite operators, Capella's satellites are especially sensitive to aerodynamic drag due to the high surface area of their large deployable radar reflectors. This unpredicted increase in drag threatened premature deorbit and reentry of some of Capella's fleet of spacecraft.

This paper explores Capella's strategic response to this problem at all layers of the satellite lifecycle, examining the engineering challenges and insights gained from adapting an operational constellation to rapidly changing space weather conditions. A key development was the implementation of "low drag mode”, which increased on-orbit satellite lifetime by 24% and decreased accumulated momentum from aerodynamic torques by 20–30%. The paper shares operational tradeoffs and lessons from the development, deployment, and validation of this flight mode, offering valuable insights for satellite operators facing similar challenges in LEO's current elevated drag environment.

Full text

The 4S Symposium 2024 – W. S. Shambaugh 1
DOING BATTLE with the SUN:
LESSONS FROM LEO and OPERATING a SATELLITE CONSTELLATION in the
ELEVATED ATMOSPHERIC DRAG ENVIRONMENT of SOLAR CYCLE 25
W. Scott Shambaugh
Capella Space, San Francisco, California, USA, scott.shambaugh@capellaspace.com
ABSTRACT
Capella Space, which designs, builds, and operates a constellation of Synthetic Aperture Radar (SAR)
Earth-imaging small satellites, faced new challenges with the onset of Solar Cycle 25. By mid-2022,
it had become clear that solar activity levels were far exceeding the 2019 prediction published by the
National Atmospheric and Oceanic Administration's (NOAA) Space Weather Prediction Center
(SWPC). This resulted in the atmospheric density of low Earth orbit (LEO) increasing 2–3x higher
than predicted. While this raises difficulties for all satellite operators, Capella's satellites are
especially sensitive to aerodynamic drag due to the high surface area of their large deployable radar
reflectors. This unpredicted increase in drag threatened premature deorbit and reentry of some of
Capella's fleet of spacecraft.
This paper explores Capella's strategic response to this problem at all layers of the satellite lifecycle,
examining the engineering challenges and insights gained from adapting an operational constellation
to rapidly changing space weather conditions. A key development was the implementation of "low
drag mode”, which increased on-orbit satellite lifetime by 24% and decreased accumulated
momentum from aerodynamic torques by 20–30%. The paper shares operational tradeoffs and lessons
from the development, deployment, and validation of this flight mode, offering valuable insights for
satellite operators facing similar challenges in LEO's current elevated drag environment.
1 BACKGROUND
LEO is not a vacuum. Satellites orbiting between ~80 and ~700 km fly through the thermosphere, an
upper layer of Earth’s atmosphere where gas floats in free molecular flow conditions. This thin air
exerts a drag force that causes a spacecraft’s altitude to gradually decay over time, until it hits the
denser mesosphere and burns up in atmospheric reentry.
A satellite’s acceleration due to drag , is a function of atmospheric density , velocity relative
to the air stream , coefficient of drag , planar area in the velocity direction , and mass of the
satellite  (Equation 1). The last terms are specific to the design of a satellite and can be grouped
into a single term called the ballistic coefficient  (Equation 2).

The 4S Symposium 2024 – W. S. Shambaugh 2
 =1
2||
(1)
=
(2)
While most of these terms are relatively stable, thermosphere density varies widely in both space and
time (see Section 4.3). In the 400–600 km range, every 10 km increase in altitude corresponds to a
~14% decrease in density. Density also varies strongly with solar flux and geomagnetic conditions.
Solar flux over the 11-year rising and falling solar cycle pumps energy into the thermosphere, causing
it to swell and densify at higher altitudes. This solar flux is generally measured by the number of
visible sunspots (SSN) or the intensity in the 10.7 cm radio band (F10.7), which are strongly
correlated metrics that can be converted between [1]. These nonlinearities in density must be carefully
handled when predicting satellite drag.
In most cases, the phenomenon of satellite lifetime is looked at through the lens of minimizing space
debris. The Federal Communications Commission (FCC) which issues communication licenses for
commercial U.S. satellite operators, requires that spacecraft operating in LEO and licensed under its
Part 25 “standard” rules reenter within 25 years after an up-to 15-year operational lifetime. This
reentry time drops to five years in October 2024 [2]. Spacecraft licensed under its “streamlined” small
satellite rules must have a total in-orbit life of no more than six years [3]. The lifetime analysis must
follow NASA guidance in [4] and generally be performed with NASA’s Debris Assessment Software
(DAS) [5]. From this lens, higher drag environments and rapid post-mission deorbit are preferred,
and predictions of a lower density atmosphere make for conservatively long lifetime estimates.
A commercial satellite operator looks at the drag environment from the opposite perspective. For
them, the goal is to ensure they meet their minimum design lifetime, and generally to maximize
mission duration past that as far as regulatory limits permit. To do this they must protect for higher
drag environments that would cause a premature end to on-orbit operations. As Solar Cycle 25’s
growth towards maximum over the past few years has greatly exceeded predictions, the delta between
lower and higher density estimates has taken satellite operators by surprise, often with dramatic
consequences.
In February 2022, SpaceX lost 38 of 49 Starlink satellites during a geomagnetic storm immediately
after launch to a 340 x 210 km staging orbit, due to atmospheric drag increasing up to 50% higher
than expected [6]. This event brought attention to the need for better “nowcast” space weather
predictions so that satellite operators can anticipate and react to short term space weather events [7],
however such improvements to real time data would not help satellite designers with their need to
predict the trajectory of the solar cycle many years into the future.
In August 2023, a TechCrunch news article shared that Capella Space’s satellites were having their
own struggles with the stronger than predicted solar cycle [8]. The company’s Whitney-generation
SAR satellites were deorbiting faster than expected and reentering sooner than their intended three-
year lifetime. This paper shares a behind-the-scenes look at Capella’s thinking and response to this
issue.

The 4S Symposium 2024 – W. S. Shambaugh 3
2 INTRODUCTION to CAPELLA SPACE
2.1 Capella Space and its Mission
Capella Space is an American space company which designs, manufactures, and operates a
constellation of Synthetic Aperture Radar (SAR) Earth-imaging small satellites. Founded in 2016, it
was the first U.S. company to fly a constellation of SAR satellites and continues to deliver the highest
quality SAR imagery commercially available anywhere in the world. Using X-band radar allows for
any-time imaging of anywhere on Earth, regardless of cloud cover or lighting conditions. This
represents a significant advantage in imaging availability over optical sensors, which are prevented
by clouds and darkness of night from resolving ~75% of Earth’s surface at any given time (and are
fully denied from extreme latitudes during polar night). This technology is key to the company’s
mission to make timely Earth observation an essential tool for commerce, conservation, and well-
being. Figure 1 shows an example of Capella’s imagery.
Figure 1: A spotlight SAR image of Palma de Mallorca, Spain taken in June 2022 by the first
Whitney-generation satellite, Capella-3. Inset shows details of a ship at port. Colors indicate the
squint angle over the duration of the collect, and brightness indicates intensity. See [9] for details.
Figure 2 shows the timeline of Capella’s satellite launches. This paper will focus on the six Whitney-
generation satellites which were in operation in 2022.
Figure 2: Capella’s launch timeline, with satellite name, generation, target altitude, and inclination.

The 4S Symposium 2024 – W. S. Shambaugh 4
2.2 The Capella “Whitney” Satellite
The Whitney satellite design is distinguished by several large deployable structures which unfold
from the satellite bus after the satellite is delivered to orbit. These deployable structures consist of
solar panels, a 3-meter folding boom which carries the radar instrument, and a 3.5-meter diameter
radar reflector made from fine wire mesh. To capture images, the radar instrument emits microwave
pulses that bounce off the reflector towards an Earth target. The return echo is concentrated by the
same reflector back to the radar instrument, and this signal data is downlinked over the satellite’s
high gain antenna and then processed to generate the SAR images. Whitney also carried an antenna
to communicate with Inmarsat’s constellation of geostationary communication satellites via their
Inter-satellite Data Relay Service (IDRS), which allows the satellite to share telemetry and be tasked
to take imagery in near real time without needing to wait to fly over a ground station. See Figure 3
for a rendering of the satellite with these components labeled.
Figure 3: A rendering of a Whitney generation satellite in flight (left), and its 2022 flat plate
aerodynamic model (right).
2.3 Capella’s Satellite Concept of Operations (ConOps)
Since the Whitney satellite does not use any mechanically or electronically steerable components, the
satellite must frequently change its orientation to point hardware components at different targets.
Scheduled pointing activities include imaging collects, ground station contacts, and propulsive
maneuvers. While time allocated to these activities for a given orbit varies significantly with different
satellite schedules, over the long term these use approximately 30% of the satellite’s time. The rest
of the orbit is filled with background activities which depend on whether the satellite is in sunlight.
When in sunlight, the satellite pivots to point its solar panels at the sun for power generation (“Sun
pointing”). When in eclipse, the satellite pivots to point its IDRS antenna to track the nearest Inmarsat
satellite to maintain a communication link (“IDRS pointing”). Over a time period of several days,
analysis of telemetry has shown that the satellite’s orientation to the velocity stream is well
approximated by a uniform random sampling of the unit sphere.

The 4S Symposium 2024 – W. S. Shambaugh 5
2.4 Whitney Drag Estimation
The satellite’s use of a large, high gain parabolic radar reflector is a key enabler of its best-in-class
imagery data. However, even with the use of wire mesh reducing the area fill fraction of the reflector’s
geometric envelope (Figure 4), the reflector contributes a full three-quarters of the satellite’s entire
planar surface area at only one quarter of its mass.
Figure 4: An on-orbit photograph of the fully deployed wire mesh radar reflector on the first
Whitney-generation satellite Capella-3, taken from a camera at the end of its boom. The low fill
fraction of the mesh gives the mesh optical transparency.
To estimate satellite ballistic coefficient prior to the launch of the first Whitney, Capella deemed that
numerical simulation via Direct Simulation Monte Carlo (DSMC) would be unreliable, given the
meshing difficulties of an enormous numbers of thin wires following a complex geometry. Instead,
fill-fraction-weighted surface areas and centers of pressure were estimated and combined into a flat
plate model for each of the primary spacecraft faces (Figure 3). Since the ConOps result in a random
orientation of the spacecraft to the airstream, the overall effective area of the spacecraft was taken to
be the average of these faces. During the early Whitney design effort in 2020 this was estimated at
3.0 m2, and later revised in 2022 with better estimates to 3.6 m2. The coefficient of drag was initially
estimated at 2.2, using the default from NASA’s Debris Analysis Software (DAS) version 3.1, which
resulted in a 2020 ballistic coefficient estimate of = 0.059 m2/kg.
During the Whitney design effort, Capella used this information along with the solar cycle prediction
from the NOAA SWPC consensus model (Section 3.1) to predict atmospheric drag, which fed into
an estimate of orbital lifetime. Analysis at that time found the propulsion system adequate to keep the
spacecraft on-orbit for the 3-year design goal, by combining a propellant reserve for collision
avoidance (COLA) maneuvers with drag compensation needs at a 525 km insertion altitude.

The 4S Symposium 2024 – W. S. Shambaugh 6
After launch, the ballistic coefficient was measured by fitting modeled satellite altitude decay using
the NRLMSISE-00 density model [10] against measured orbit data, resulting in ballistic coefficient
= 0.085 m2/kg. From the satellite mass and updated area estimate, this implies a coefficient of
drag of = 2.7, which compares reasonably to estimates of shapes in near-vacuum molecular flow
[11], but is also 44% higher than originally estimated.
A comparison to ballistic coefficients of other satellites is difficult to generate, as it depends on
knowledge of satellite attitude and propulsive ConOps that is usually not publicly available. However,
a literature review found a few collections of estimates for various satellites, shown below in Figure
5. From this, we can put in context Whitney’s unique susceptibility to atmospheric drag as 2-10x
higher than typical satellite missions.
Figure 5: Estimates of ballistic coefficient  for satellites from [12], [13], and [14], vs Capella’s
Whitney satellite. Note that the METEOR 1-1 and METEOR 1-7  estimates in [13] were hidden
as they were not in family with the other METEOR satellites and believed to be erroneously high.
3 SOLAR CYCLE 25 and THE IMPACT on CAPELLA’S SATELLITES
3.1 Early Solar Cycle 25 Predictions
Ever since the end of Solar Cycle 21, the SWPC has convened a panel of experts during solar
minimum to compare predictions for the upcoming solar cycle. The December 2019 prediction panel
chaired by NOAA, NASA, and ISES published in December 2019 its “consensus” forecast of Solar
Cycle 25, predicting a weak cycle in-line with Solar Cycle 24 [15] [16]. NASA Marshall Space Flight
Center also releases a long-term prediction, updated monthly [17]. From 2020–2022, this model’s
mean prediction lined up with the SWPC panel, albeit with much larger variability. The other major
long-term prediction of the cycle was put out in 2020 by the National Center for Atmospheric
Research (NCAR). This used the newly hypothesized McIntosh et al. “Terminator” model of moving
magnetic bands on the surface of the Sun, and contradicted the consensus to predict a stronger-than
average solar cycle [18].
Common tools for analyzing satellite lifetime use different solar cycle predictions. Ansys’ Systems
Tool Kit’s (STK’s) satellite lifetime tool uses by default a prediction from its SolFlx_CSSI.dat file,
last updated in 2017. The source of this prediction is unclear, but it aligns closely with the SWPC
panel’s model. STK’s lifetime tool can also use its space_weather.txt data [19], which pulls from
Celestrak Space Weather [20]. Celestrak prior to 26 May 2023 used the SWPC panel consensus
prediction, after which it switched to using the NASA Marshall Solar Cycle Forecast. NASA’s DAS
tool uses the SWPC panel prediction for Solar Cycle 25 [21], however it appears to be biased towards
the 1 value, presumably to bake in some conservatism to the tool for ensuring timely satellite
reentry timelines. Table 1 shows the peak solar flux estimates from the major prediction sources, and
Figure 6 plots the full predictions against the raw F10.7 solar flux observations from GFZ Helmholz
Center Potsdam [22].

The 4S Symposium 2024 – W. S. Shambaugh 7
Table 1: Solar Cycle 25 Predictions. Predictions marked with * have been converted from sunspot
number SSN to F10.7 solar flux via the 4th order model covering 1981-2015 from Table 9 in [1].
Prediction Source
Prediction Date
Solar Cycle 25 Peak Flux
Uncertainty
NOAA SWPC prediction
panel consensus [15]
9 December 2019
138 SFU*
in July 2025
± 8 SFU*, ± 8
months
Ansys STK lifetime tool:
SolFlx_CSSI.dat [19]
January 2017
145 SFU
in December 2025
± 9 SFU 95%
confidence interval
NASA Marshall Solar Cycle
Forecast [17]
August 2020
145 SFU
in December 2023
102–229 SFU 90%
confidence interval
April 2022
138 SFU
in December 2023
85–199 SFU 90%
confidence interval
NASA DAS [5] [21]
v3.2.0, v3.2.3
29 December 2021,
29 June 2022
135 SFU
in July 2025
-
McIntosh et al. [18]
24 November 2020
221 SFU*
in ~February 2024
± 10 SFU* 68%
confidence interval,
± ~6 months
McIntosh/Leamon
Unpublished 2022 Update [26]
2 March 2022
194 SFU*
in June 2025
± 13 SFU* 68%
confidence interval
McIntosh/Leamon/Egeland
Final Prediction [27]
30 January 2023
190 SFU*
± 12 SFU* 68%
confidence interval
Figure 6: Historical observed F10.7 solar flux from [22], with predictions from Table 1 sources
circa April 2022. Where applicable, confidence intervals were converted to 1 values using a
normality assumption. Numbers indicate the solar cycle.
While the stronger than expected solar cycle did not go unnoticed by the research community (see
[23] from June 2022), the SWPC was slow to update its official estimates. It was not until October
2023 that it launched an experimental solar cycle prediction that updates monthly and better reflects
the current state of the cycle [24].
3.2 Capella Experiences Increased Drag on Orbit
Capella first noticed an issue with atmospheric drag in April 2022, with noticeable acceleration of
altitude decay across its constellation. For a representative 500 km circular orbit at =45°

The 4S Symposium 2024 – W. S. Shambaugh 8
inclination, the SWPC panel model predicted a density of 1.7e-13 kg/m3, but the actual density over
that month was around 4.2e-13 kg/m3 (~2.5x higher), with peaks up to 9.9e-13 kg/m3 (5.5x higher).
Searching for a better flux model, Capella found the NCAR team had recently announced that the
Sun’s “Terminator” event had occurred in December 2021, and that their internal forecasts had been
updated based on this timing [25]. With the help of the Austrian Space Weather Office (ASWO) and
its open-source “heliocats” tool [26], Capella was able to obtain a copy of this unpublished forecast
and found it to match observations well. This 2022 McIntosh/Leamon model formed the basis of
Capella’s solar flux modeling moving forward, with a slight update after their finalized prediction in
January 2023 [27]. In retrospect, the Marshall 1-high prediction would also have worked.
To compare solar flux predictions, Capella chose a representative 500 km circular orbit at inclination
=45°, and calculated the orbit-average density over all RAANs using the NRLMSISE-00 model at
a density-weighted historical average geomagnetic Ap index of 10. This calculation was sped up by
several orders of magnitude with the development of an internal tool that precalculated this data over
a range of altitudes, inclinations, solar fluxes, and Ap indices, and then used 4-dimensional spline
interpolation to find intermediate values. This density estimate later fed into another internally
developed tool for probabilistic lifetime predictions. Figure 7 shows the differences in density
predictions between the SWPC panel model and McIntosh/Leamon 2022 model for this
representative orbit. Density over the whole solar cycle had increased by 2–3x, which decreases
satellite lifetime a corresponding amount.
Figure 7: Density estimates for Solar Cycle 25 for a 500 km circular orbit at =45° inclination,
with predictions from Table 1 sources circa April 2022. Where applicable, confidence intervals
were converted to 1 values using a normality assumption.

The 4S Symposium 2024 – W. S. Shambaugh 9
This analysis predicted that if current trends continued, six of the seven spacecraft on-orbit would re-
enter before the end of the year. Additionally, the upcoming Capella-9 spacecraft was slated for a
launch to ~525 km, but at this insertion altitude would only survive for nine months even with full
use of its onboard propellant. Mitigating the impending loss of these satellites instantly became a top
priority for the company.
4 WAYS of INCREASING SATELLITE LIFETIME
Capella immediately began investigating strategies for increasing satellite lifetime. Different
approaches broadly apply to different points in the satellite lifecycle:
1. Modifications to the baseline satellite design or mission architecture to make it less sensitive
to drag. This was largely infeasible for Capella’s situation, due to the extensive effort and
time required to design and qualify a new satellite bus.
2. Subsystem modifications or upgrades, especially the propulsion system. While free space in
Capella’s internal bus volume is limited, its system architecture is modular and not tightly
coupled. Subsystems can be “hot swapped”, given adequate structural, power, and volume
margins, and enough time for the associated design effort and procurement.
3. Choosing higher or more inclined orbits. For a small satellite operator flying on rideshare
launches, this is largely constrained by launch provider orbit availability. However, using
orbital transfer vehicles on a rideshare or procuring dedicated launches from “small launch”
providers allows for more flexibility in orbit selection.
4. On-orbit ConOps modifications. This is the only way to mitigate drag for existing on-orbit
assets.
4.1 Increase Altitude
Onboard propellant can be used to maintain altitude in the presence of atmospheric drag, or it can be
used to perform orbit change maneuvers. Because atmospheric density continues to decrease as
altitude increases, it is more propellant efficient from a lifetime perspective to raise the satellite as
high as possible and coast back down than it is to hold altitude in a denser environment. Prior to the
observation of increased on-orbit drag in 2022, Capella had performed a minimal number of
propulsive maneuvers across the constellation in anticipation of future demonstrations of precision
orbit control schemes. Lifetime concerns postponed these plans, in favor of immediate use of all
propellant available for altitude raising across the entire constellation. Due to the low thrust of
Whitney’s electric thrusters and need to fit propulsive maneuvers between other activities in the busy
satellite imaging schedule, these “burn campaigns” lasted several months. Figure 8 shows the effects
of the successful burn campaigns on orbital altitude.

The 4S Symposium 2024 – W. S. Shambaugh 10
Figure 8: Orbital altitudes for Capella’s satellite constellation over time, with burn campaigns
labeled. The solid lines are semi-major axis altitude and the shading shows perigee to apogee.
Instead of burning propellant to raise altitude, it’s far easier to simply be dropped off at a higher orbit
by the launch vehicle. Being able to go higher than available rideshare missions was a major factor
in Capella’s decision to pull Capella-9 from its planned launch and procure a dedicated ride from
Rocket Lab for Capella-9 and Capella-10, which were launched in March 2023 on an Electron rocket
to a targeted 600 km [28]. This was also a major motivation behind the follow-on contract for four
dedicated launches on Electron rockets for Acadia-generation satellites [29], the first of which
launched to a targeted 640 km in August 2023.
There was a compromise to reach when choosing launch altitudes, between ensuring a minimum
three-year operating life under conservatively denser solar flux profiles, and a maximum six year in-
orbit life under conservatively less dense profiles for the streamlined small satellite FCC licensure.
Meeting both requirements required Capella to reject the default solar flux profile in NASA’s DAS
tool, and substitute a custom profile based on the lower bound of the McIntosh model. See the Capella
9&10 Orbital Debris Assessment Report (ODAR) for details [30]. As a condition of accepting this
rationale, Capella’s FCC operating license stipulates that the company must continue to monitor space
weather conditions, reserving propellant as needed to perform powered deorbits and not exceed an
in-orbit lifetime of six years.

The 4S Symposium 2024 – W. S. Shambaugh 11
4.2 Increase Size of Propulsion System
Increased atmospheric drag has two implications for the sizing of a satellite propulsion system. First,
the total impulse of the system must increase to counteract the cumulative drag which the satellite
will experience over its lifetime. This can be accomplished with increased thruster efficiency (specific
impulse) or more propellant mass. Second, given a power budget, thermal limitations on maximum
burn duration, and ConOps scheduling constraints, thrust levels must be high enough to deliver
adequate impulse per orbit to prevent altitude decay.
By mid-2022, Capella had already procured the propulsion systems for the last two Whitney satellites,
as well as a planned propulsion upgrade for the first two Acadia-generation spacecraft Capella-11
and Capella-12, which increased total impulse by 5x. With the higher altitude afforded by dedicated
Rocket Lab launches (Section 4.1), these systems were deemed adequate to meet a nominal three-
year design lifetime even in elevated drag conditions.
However, thrust levels on this planned upgrade were deemed too low to deliver enough impulse per
orbit for the elevated drag environment for future rideshare missions to more typical orbit insertion
altitudes (500–550 km). So in late 2022, Capella kicked off design and procurement of a new electric
propulsion system that had an additional 4x increase in thrust and 2.5x increase in total impulse. This
new propulsion system flew and operated successfully for the first time on Capella-14 in April 2024.
4.3 Increase Orbital Inclination
At a given orbital radius, Earth’s thermosphere is denser at the equator than at the poles due to more
direct solar heating and the planet’s equatorial bulge. Roughly, a satellite launched in an inclination
=0° equatorial orbit will encounter 30% higher orbit-averaged atmospheric density than one flying
in an =90° polar orbit (Figure 9). While raising altitude is a much more efficient use of onboard
propellant than changing inclination after launch, inclination does play into orbit selection.
Figure 9: Orbit-averaged atmospheric density for a satellite flying in a circular orbit, as calculated
using the NRLMSISE-00 density model and assuming a geomagnetic Ap index of 10 (see Section
3.2). Left shows density as a function of inclination and altitude during solar flux F10.7 = 160 SFU
conditions, and right shows density as a function of inclination and solar flux at 500 km altitude.

The 4S Symposium 2024 – W. S. Shambaugh 12
4.4 Lower the Operational Altitude Floor
Capella’s original plan for satellite operations was to capture images during orbital decay down to
450 km, but its ultimate floor for imaging operations is driven by the company’s FCC license
conditions and NOAA license issued by the Commercial Remote Sensing Regulatory Affairs office.
Extending imaging operations to lower altitudes was an easy way to gain additional lifetime.
The ability of the satellite to operate at lower altitudes is also driven by its capacity to absorb
aerodynamic torque transients and bleed off accumulated momentum with its attitude determination
and control system (ADCS). Whitney’s asymmetry and large aerodynamic lever arms (see Figure 3)
stress its ADCS far more than a typical axisymmetric satellite design. To this end, Capella sized up
the torque rods on all satellites past Capella-12 by 63%. For existing satellites on-orbit, Section 5.3
talks about ConOps adjustments that reduced accumulated momentum.
4.5 Increase Satellite Mass
Satellites are generally designed to be as mass efficient as possible, and the addition of raw mass is
rarely desirable. But it does decrease the ballistic coefficient. Shortly before launch of Capella-11,
Capella decided to add 10 kg of steel ballast mass to increase its mass to the upper end of the 160–
170 kg range provided in its ODAR filing. This was estimated to add 3–4 months of life in a
contingency scenario if its new propulsion system failed on first flight, and was left off later satellites.
4.6 Decrease Satellite Coefficient of Drag
Aerodynamically shaping the satellite to reduce its coefficient of drag was deemed of dubious value.
Drawbacks were the complexity of performing iterative DSMC over a range of orientations, the lack
of design heuristics for shapes in free molecular flow that reduce drag, the importance of the existing
geometry to the radar imaging mission, and the likelihood that any benefits in coefficient of drag
from added cowling would be more than negated from increases in overall surface area.
There have been a few proposals to decrease satellite coefficient of drag by smoothing surfaces, which
increases the ratio of specular to diffuse molecular reflections and thus reduces skin friction.
Extremely high levels of surface polishing [31] or atomic layer deposition of polymers [32] have been
suggested as methods to accomplish this, however these surface treatments were deemed immature
and in conflict with mesh reflector performance and bus thermal design.
4.7 Decrease Satellite Area
Removing or miniaturizing external components to reduce the satellite’s area is only of benefit to
decreasing drag if they are low enough mass to have a ballistic coefficient  greater than the satellite
average. No external parts on Whitney were identified as suitable to delete or substantially modify.
If the physical surface area of a satellite cannot be modified, then for a satellite with heterogenous
geometry the effective surface area can be made smaller by modifying ConOps to fly while presenting
smaller planar areas to the velocity stream. This is the principle of Capella’s “low drag mode” flight.

The 4S Symposium 2024 – W. S. Shambaugh 13
5 LOW DRAG MODE
5.1 Low Drag Concept
By flying with its radar reflector and solar panels “edge-on” (see Figure 10), Whitney can avoid
presenting its Z face to the velocity stream, which has ~2.2x the planar area of its X and Y faces
(Figure 3). Of those two approximately equal-area options, the Y face is preferred for its lower
aerodynamic lever arm, which is 14% that of the X face. Some intermediate orientation that perfectly
balances aerodynamic torques could have been chosen, but with the uncertainties associated with the
vehicle aerodynamics in free molecular flow, moving away from an orthogonal orientation of the
spacecraft axes with respect to the airstream was deemed low impact and potentially confusing.
“Low drag mode” was thus defined as the Whitney satellite flying with its Y axis aligned with the
velocity vector, with a rotation about that vector to best point the relevant hardware toward its target
(e.g., solar panels towards the Sun). This pointing approach is also known as the “align-constrain”
algorithm [33]. With Whitney’s average planar area of 3.60 m2 reduced to the Y area of 2.55 m2 for
the ~70% of flight spent conducting background activities, Capella expected a (schedule-dependent)
~21% reduction in drag, and approximately equal increase in satellite lifetime.
Changing a satellite’s drag profile by adjusting its flight orientation is not a new concept. Several
LEO satellite constellations have used differential drag as the primary method of orbit phasing and
formation flying (see [34] section II for an overview), in lieu of an onboard propulsion system.
Additionally, SpaceX has shared that their Starlink satellites were commanded into an edge-on low-
drag orientation during the February 2022 geomagnetic storm [7], much like their regular “ducking”
maneuvers to reduce frontal areas during conjunction events. However, the author is not familiar with
this technique being adopted for active satellites that were not already designed for such a flight mode.
5.2 Trade-offs
The downside of flying in low drag mode is loss of precision pointing for power generation and IDRS
communications link. IDRS connectivity was deemed easier to test than to model, but the impact to
the power budget was more difficult to test since stressing conditions are not necessarily present at
any given time in an orbit. So, power impacts were analyzed beforehand.
Figure 10: Low drag mode geometry when solar angle =0°.

The 4S Symposium 2024 – W. S. Shambaugh 14
The fraction of a circular orbit which is available for power generation depends on how much of that
orbit is spent in eclipse. This is a complex calculation in the general case (see [35] and [36]). However,
the only scenario of interest in this case was the minimum power stressing scenario. This happens
when the solar  angle (angle between the orbital plane and sun vector) is minimized at 0°, and the
geometry of this scenario is shown in Figure 10. For a circular orbit in LEO where eclipse can be
approximated as a cylinder rather than umbral and penumbral cones, the half angle of solar
illumination  is a function of the orbital altitude  and Earth’s radius  (Equation 3).
=
2+cos 
+(3)
For baseline direct-pointing ConOps, the minimum-sunlight orbit that results in power generation
fraction  occurs at solar angle =0° and is given by Equation 4.
 =1
2 

 =
(4)
Solar angle =0° is also the stressing case for the low drag mode align-constrain pointing algorithm.
Since Whitney’s solar panels point orthogonal to the Y axis, the optimally aligned solar pointing
vector will be aligned with orbit radial or antiradial. The light hitting the solar panels will be
attenuated by cosine losses, and so assuming no minimum incidence angle, the minimum power
generation fraction in low drag mode is given by Equation 5.
 =1
2|cos()|

 =2sin()
(5)
For a representative 500 km circular orbit, the ratio of Equations 4 and 5 show that low drag mode
results in a 55% drop in power production capacity during the stressing low sunlight scenario. This
was acceptable for the Whitney satellites that had completed their power-hungry burn campaigns.
5.3 On-orbit Performance
Capella-7 and Capella-8 were launched together on the SpaceX Transporter-3 mission in January
2022, and after their burn campaign their orbital altitudes gradually decayed until they re-entered the
atmosphere in August 2023. Since they shared an orbit, they provided an excellent natural experiment
to compare the effects of low drag mode. In May of 2023, low drag mode was enabled on Capella-7,
with Capella-8 remaining in its baseline ConOps as a reference. After a month, the flight modes were
swapped to confirm the generality of the results, and this configuration was kept through reentry.

The 4S Symposium 2024 – W. S. Shambaugh 15
Figure 11: The effects of low drag mode on Capella-7 and Capella-8. Top shows altitude over time,
where the solid lines are semi-major axis altitude and the shading shows perigee to apogee. Middle
is daily average torque rod duty cycle, which tracks momentum accumulation due to aerodynamic
torques. Bottom shows micro tumble events.
Figure 11 shows the effects of low drag mode on satellite altitude decay, with the change in behavior
being visually apparent in plots of altitude over time. By fitting predicted decay against actuals,
Capella found that low drag mode resulted in a 24% decrease in aerodynamic drag for both spacecraft,
comparing well to the ~21% predicted.
Figure 11 also shows the effect of low drag mode on torque rod duty cycles, which is taken as a proxy
for accumulated spacecraft momentum from aerodynamic torques. The satellite with low drag mode

The 4S Symposium 2024 – W. S. Shambaugh 16
enabled consistently showed 20–30% lower torque rod duty cycles. Additionally, towards the end of
a spacecraft’s life as it falls to denser layers of the atmosphere, it begins experiencing “micro tumble”
events where the reaction wheels saturate and are unable to control attitude for several minutes. Low
drag mode lowered the altitude where these started occurring from 404 km on Capella-7, to 365 km
on Capella-8.
The downsides of reduced IDRS link and power accumulation were assessed during this test. While
high-fidelity power model validation was not performed due to limitations in telemetered data, the
satellites in low drag mode maintained adequate battery charge levels. IDRS link was characterized
by the duration of outage between subsequent connections. Figure 12 shows a 35% increase in mean
outage duration and longer tail cases, but Capella found this to have minimal impact on operations.
Figure 12: Boxplot of the effects of low drag mode on IDRS outage duration.
One more unexpected benefit of low drag mode was observed. Reducing the susceptibility to drag on
the spacecraft by ~24% also reduced the variability of medium-term orbit propagation predictions.
This improved the timing accuracy when scheduling imaging collects and ground downlink contacts.
6 LESSONS LEARNED
There are several lessons to be learned from Capella’s experience:
• Solar physics is still a nascent field, and solar cycle predictions have significant epistemic
uncertainty outside of the aleatory uncertainty within each prediction. This extends not only
to the magnitude and timing of the solar cycle, but to whether there will be a solar cycle at all
— we do not yet have an explanation for the “Maunder Minimum” between 1645–1715 where
solar activity remained at prolonged low levels. For Solar Cycle 26, Capella recommends
sizing spacecraft propulsion systems based on historical extremes regardless of the forecast.
• The default solar cycle prediction or coefficient of drag in common satellite lifetime tools
should not be blindly trusted for satellite lifetime design purposes.
• Propulsion is a system worth oversizing. Deorbit from LEO is a hard lifetime limiter, and the
added value from marginal mission extension far exceeds the cost of additional propellant.
Solid propellant electric thrusters currently in development will make propulsion a much more
compact system to scale up by the time Solar Cycle 26 arrives.
• Differential drag or low drag flight modes are feasible ConOps even for vehicles not
specifically designed for them, though they come with power and other operational trade-offs.

The 4S Symposium 2024 – W. S. Shambaugh 17
• Satellite lifetime estimation needs to be a continuously monitored metric after launch, rather
than being just a design requirement. These estimates should be driven by real-time telemetry
and the latest solar cycle prediction updates. They should also be a probabilistic metric for the
range of possible futures rather than a point estimate.
• Vertical integration within a satellite company of design, manufacture, and operations is
extremely powerful. Capella was able to quickly tackle its drag problem at all levels of the
satellite lifecycle, and trade off operational solutions against hardware solutions with clear
visibility into both sides.
• This is a case study that proves out the value of dedicated small launch vehicles to provide
the flexibility and agility that satellite operators may need in unique situations.
• This is also a case study to show the value of a proliferated LEO constellation architecture
with regular launches and iterative improvement. Satellites lost to drag were able to be
backfilled by satellites already in production, and the designs of the satellites in production
were able to be modified in real time to better handle the changing space environment.
7 SUMMARY
In April of 2022, Capella Space discovered that their Whitney satellites were deorbiting faster than
NOAA SWPC’s consensus solar cycle model predicted, due to increased solar flux densifying the
atmosphere in LEO to 2–3x over predictions. Based on on-orbit data that increased Capella’s estimate
of Whitney’s ballistic coefficient from 0.059 to 0.085 m2/kg and the 2022 McIntosh/Leamon solar
flux model, Capella revised their satellite lifetime estimates and found that they expected to lose six
out of their seven operational satellites by the end of the year.
Faced with the impending early loss of its satellites, Capella acted quickly and decisively to address
the increased drag environment and its effects. The company immediately started burn campaigns to
raise orbits of its on-orbit satellites, procured a new more performant electric propulsion system,
switched launch providers for upcoming missions to “small launch” dedicated rockets that could
reach higher orbits, added ballast mass to already-built satellites to extend life, sized up torque rods
on future satellites to lower the operational altitude floor, and deployed and validated a new “low
drag” flight mode that extended in-orbit lifetime by 24%. While all the satellites that were flying in
April 2022 have since reentered, these efforts extended the lives of those vehicles by a cumulative
2.5 years, and that of Capella’s expanding constellation by many times that.
8 ACKNOWLEDGEMENTS
The author thanks the entire Capella Space team for its rapid response and hard work solving the
satellite drag situation. Thanks also to Scott McIntosh of NCAR and Chris Möstl of ASWO for their
assistance in obtaining and interpreting pre-released solar cycle predictions.

The 4S Symposium 2024 – W. S. Shambaugh 18
9 REFERENCES
[1] Clette, Frédéric. Is the F10. 7cm–Sunspot Number relation linear and stable?. Journal of Space
Weather and Space Climate 11: 2, 2021.
[2] Federal Communications Commission. FCC 22-74 “Mitigation of Orbital Debris in the New
Space Age”. 30 September 2022.
[3] Federal Communications Commission. FCC 19-81 “In the Matter of Streamlining Licensing
Procedures for Small Satellites”. 2 August 2019.
[4] National Aeronautics and Space Administration. NASA-STD 8719.14 Revision A with Change 1
“Process for Limiting Orbital Debris”. 25 May 2012.
[5] National Aeronautics and Space Administration. NASA/TP-2020-5002380 “Debris Assessment
Software User’s Guide”. May 2020.
[6] SpaceX. Geomagnetic Storm and Recently Deployed Starlink Satellites. SpaceX Updates. 8
February 2022.
[7] Berger, T. E., et al. The thermosphere is a drag: The 2022 Starlink incident and the threat of
geomagnetic storms to low earth orbit space operations. Space Weather 21.3. 2023.
[8] Alamalhodaei, Aria. Capella’s Earth-Imaging Satellites Are Deorbiting Faster than Expected.
TechCrunch. 5 August 2023.
[9] Castelletti, Davide, et al. Capella Space VHR SAR constellation: Advanced tasking patterns and
future capabilities. IGARSS 2022-2022 IEEE International Geoscience and Remote Sensing
Symposium. IEEE. 2022.
[10] Picone, J. M., A. E. Hedin, D. P. Drob, and A. C. Aikin, NRLMSISE-00 empirical model of the
atmosphere: Statistical comparisons and scientific issues, Journal of Geophysical Research: Space
Physics, 107(A12). 2002.
[11] Moe, Kenneth, and Mildred M. Moe. Gas–surface interactions and satellite drag coefficients.
Planetary and Space Science 53.8: 793-801. 2005.
[12] Zhejun, Lu and Weidong, Hu. Estimation of ballistic coefficients of space debris using the ratios
between different objects. Chinese Journal of Aeronautics. 2017.
[13] Saunders, Arrun, Hugh Lewis, and Graham Swinerd. Further evidence of long‐term
thermospheric density change using a new method of satellite ballistic coefficient estimation. Journal
of Geophysical Research: Space Physics 116.A2. 2011.

The 4S Symposium 2024 – W. S. Shambaugh 19
[14] Saunders, Arrun, Graham G. Swinerd, and Hugh G. Lewis. Deriving accurate satellite ballistic
coefficients from two-line element data. Journal of Spacecraft and Rockets 49.1: 175-184. 2012.
[15] Space Weather Prediction Center. Solar Cycle 25 Forecast Update. 9 December 2019.
[16] Petrovay, Kristóf. Solar cycle prediction. Living Reviews in Solar Physics 17.1:2. 23 March
2020.
[17] Marshall Space Flight Center. Archived Solar Cycle Forecasts. 12 March 2024.
[18] McIntosh, Scott W., Leamon, Robert, et al. Overlapping magnetic activity cycles and the sunspot
number: forecasting sunspot cycle 25 amplitude. Solar Physics 295.12: 1-14. 2020.
[19] Ansys. Solar Flux File Format. STK Help. March 2024.
[20] Celestrak. Space Weather Data Documentation. 26 May 2023.
[21] Manis, A., Matney, M. Updated Solar Cycle Predictions for Orbital Debris Modeling. Orbital
Debris Quarterly News 24: 4. November 2020.
[22] Matzka, J., et al. The geomagnetic Kp index and derived indices of geomagnetic activity. Space
Weather 19.5. 2021.
[23] Carrasco, V. M. S., and J. M. Vaquero. An early assessment of the forecast by the Solar Cycle
25 prediction panel. Research Notes of the AAS 6.6: 121. 2022.
[24] Miesch, Mark. Solar Cycle Progression Updated Prediction (Experimental): Validation
Document. NOAA Space Weather Prediction Center. 18 January 2024.
[25] Phillips, Tony. The Termination Event has Arrived. Space Weather Archive. 25 February 2022.
[26] Möstl, Christian. heliocats, github commit 2e62a7f. Austrian Space Weather Office. 2 May 2022.
[27] McIntosh, Scott W., Robert J. Leamon, and Ricky Egeland. Deciphering solar magnetic activity:
The (solar) hale cycle terminator of 2021. Frontiers in Astronomy and Space Sciences 10: 16. 30
January 2023.
[28] Clark, Stephen. Rocket Lab deploys two Capella radar satellites after launch from Virginia.
Spaceflight Now. 16 March 2023.
[29] Capella Space. Rocket Lab Signs Multi-Launch Deal to Deploy Satellite Constellation for
Capella Space. Capella Space Press Release. 28 February 2023.
[30] Riggi, Lucas. Capella 9&10 Orbital Debris Assessment Report (ODAR) v3. Capella Space FCC
Regulatory Filings. 29 November 2022.

The 4S Symposium 2024 – W. S. Shambaugh 20
[31] Jiang, Yifan, et al. Aerodynamic drag analysis and reduction strategy for satellites in Very Low
Earth Orbit. Aerospace Science and Technology 132. 2023.
[32] Minton, Timothy K., Thomas E. Schwartzentruber, and Chenbiao Xu. On the utility of coated
POSS-polyimides for vehicles in very low earth orbit. ACS Applied Materials & Interfaces 13.43.
2021.
[33] Magner, Robert. Extending target tracking capabilities through trajectory and momentum
setpoint optimization. Small Satellite Conference. 2018.
[34] Falcone, Giusy, Jacob B. Willis, and Zachary Manchester. Propulsion-Free Cross-Track Control
of a LEO Small-Satellite Constellation with Differential Drag. 2023 62nd IEEE Conference on
Decision and Control (CDC). 2023.
[35] Cunningham, Fred G. Calculation of the eclipse factor for elliptical satellite orbits. National
Aeronautics and Space Administration. 1962.
[36] RM, Sumanth. Computation of eclipse time for low-earth orbiting small satellites. International
Journal of Aviation, Aeronautics, and Aerospace 6.5: 15. 2019.