Greenhouses in extreme environments: The Arthur Clarke
Mars Greenhouse design and operation overview
, Alain Berinstain
, Stephen Braham
, Thomas Graham
, Keegan Boyd
, Matthew Silver
, Alexis Lussier-Desbiens
, Pascal Lee
, Keith Cowing
, Michael Dixon
Canadian Space Agency, Space Science, 6767 route de l’aeroport, St-Hubert, Que., Canada J3Y 8Y9
PolyLAB, 515 West Hastings Street, Simon Fraser University, Vancouver, BC, Canada V6B 5K3
University of Guelph, Dept. of Environmental Biology, 50 Stone Road East, Guelph, Ont., Canada N1G 2W1
Mars Institute, SETI Institute, and NASA Ames Research Center, MS 245-3, Moﬀett Field, CA 94035, USA
SpaceRef Interactive Inc., P.O. Box 3569, Reston, VA 20195-1569, USA
Received 15 November 2005; received in revised form 14 June 2006; accepted 2 July 2006
Since its deployment on Devon Island, Canadian High Arctic, in 2002, the Haughton Mars Project’s Arthur Clarke Mars Greenhouse
(ACMG) has supported extreme environment related scientiﬁc and operation research that is relevant to Mars analogue studies – each at
a speciﬁc level of ﬁdelity and complexity. The Greenhouse serves as an initial experimental test-bed supporting ﬁeld research, from which
lessons may be learned to support the design and implementation of future ﬁeld facilities, and enabling higher ﬁdelity demonstrations.
This paper provides an overall description of the ACMG, describes the diﬀerent subsystems, explains its operational modes, details some
results over the three years of operation and discusses future development plans.
Crown copyright 2006 Published by Elsevier Ltd on behalf of COSPAR. All rights reserved.
Keywords: Greenhouse; Autonomous operation; Extreme environment; Space analog studies
Long-duration space exploration missions will include
plant production as a key component of a biologically
regenerative life-support system designed to meet crew
nutritional requirements while cleaning water and main-
taining a suitable atmospheric composition. In this context,
advanced life support systems, based on bio-regenerative
principles, have emerged as an important area of research
in the space sciences. One of the areas of particular interest
is the study of crop behavior under novel environmental
conditions (pressure, gas composition, etc.) that typically
fall outside of the normal conditions under which the spe-
cies developed. In addition to the plant physiology research
interests, there are also many operational aspects of plant
growth and climate control in extreme environments that
require research attention. Luckily, many of these research
objectives can be studied here on Earth.
In 2002, the Haughton Mars Project (HMP) established
the Arthur Clarke Mars Greenhouse (ACMG) as an exper-
imental ﬁeld research facility at the HMP Research Station
site at 75260N, 8952 0W, on the northwestern rim of
Haughton Crater, Devon Island, Nunavut, Canadian High
Arctic. Construction and inaugural operation of the
ACMG involved the participation of researchers at the
Canadian Space Agency (CSA) along with other HMP
partners from NASA, industry and academia in Canada
and the United States (see Fig. 1). Each summer, this moon
and Mars analogue site brings together researchers from
around the world with diverse research interests such as
geology, astrobiology and space exploration technology,
0273-1177/$30 Crown copyright 2006 Published by Elsevier Ltd on behalf of COSPAR. All rights reserved.
E-mail address: Alain.Berinstain@space.gc.ca (A. Berinstain).
Advances in Space Research 38 (2006) 1248–1259
to name only a few. Where Devon Island is also useful
from an analogue perspective is its utility as an operational
model for a remote ﬁeld research station.
Advanced life support systems have been, and continue
to be studied by the major space agencies. Among these
are, ESA, which initiated the MELiSSA project in 1989
(Lasseur et al., 2005) and NASA which has been develop-
ing bioregenerative capabilities since the early 1980s
(Wheeler et al., 2001). Also, the Russian space program
had its series of Bios experiments (Bartsev et al., 1996;
Salisbury et al., 1997) performed by the Biophysics Insti-
tute. Recent proposals for long-term exploration missions
by NASA and ESA have been a great catalyst for advanc-
ing research in this area from a diverse spectrum of contrib-
utors. The results of this advancement has been the
development of many feasible concepts for space green-
houses (Bucklin et al., 2001; Perino et al., 2002; Scaras-
cia-Mugnozza and Schettini, 2002; Clawson et al., 2005;
Janssen et al., 2005).
Another fruitful area of research concerns the study of
the eﬀect of the space environment on plant morphology
and physiology, especially the reduced pressure conditions
that will be predominant in early Mars greenhouse settings
(Andre and Massimino, 1992; Goto et al., 1995; Rygalov
et al., 2002; Dixon et al., 2005). It is important that we
understand the behaviour of plants growing under these
unique conditions to ensure a reliable and responsive life
Greenhouses have also been investigated as a part of
advanced life support systems for planetary outposts
(Wheeler and Martin-Brennan, 2000). A popular mode
for investigating the aforementioned research areas is
through the exploitation of analogue systems. The most
well-known greenhouse based bioregenerative life-support
analogue is the Biosphere II experiment (Allen, 1992),
but smaller life-support test facilities have also been based
on greenhouse environments (Drysdale and Collins, 2005).
These experiments focus on large-scale closed-ecology sys-
tems and help understand the complexity of such an
The ACMG is not a full-featured, high ﬁdelity simula-
tion of a greenhouse to be established on Mars. Rather,
as is the case of most analogue studies being conducted
on Devon Island, it supports scientiﬁc and operations
research in an operational setting that is relevant to Mars
in unique ways – each at a speciﬁc level of ﬁdelity and
The current research goal of the ACMG (speciﬁc goals
updated annually) is to develop an autonomous and
remotely monitored greenhouse operating continuously in
the harsh environment of the Canadian High Arctic. This
project is unique in the world: no one has ever built and
remotely operated a greenhouse in an extreme environment
like the polar desert of the Canadian High Arctic. It
addresses many Mars-like challenges, including: short win-
dow of operation to conduct maintenance and overhaul,
extreme environment for plant growth, limited electrical
power production and heat generation, and low-bandwidth
communication for remote monitoring and control.
This paper ﬁrst describes the greenhouse structure and
systems. The operational modes of the greenhouse are then
addressed and an example of year-round operation is
detailed. Selected results from the ﬁrst few years of opera-
tion are presented and discussed in the following section.
Finally, future developments and new research endeavors
conclude the paper.
2. System description
The greenhouse is composed of many systems, namely
the physical structure, the plant growth system, the envi-
ronment control system, the power system, the communi-
cation system, and ﬁnally, the data acquisition and
control system. The latter acts as the nexus for all the inter-
actions among the systems. Fig. 2 shows the high level
architecture of the systems and their interactions.
The arrows indicate information ﬂow. Each system, as it
currently exists, is described in the following subsections.
Fig. 1. Map showing location of Devon Island, Nunavut, Canada.
Fig. 2. High level architecture of greenhouse systems.
R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259 1249
The evolution of the greenhouse systems is discussed in the
discussion section of the paper.
The structure of the greenhouse is based on the commer-
cial ‘‘Garden Grower Hobby Kit’’ [http://www. igcusa.
com/hobby-greenhouse-garden-grower.html]. The dimen-
sions are 3.6 m ·7.2 m ·3 m, with the long end aligned
east to west to maximize sun exposure. The cover is made
of 6 mm Twinwall Lexan polycarbonate sheets, with a
R-value of 1.55, supported by 1.2 m spaced steel struts
complying with a 163 kg/m
live load speciﬁcation and
resisting up to 144 km/h wind speeds. A porch (1.3 m wide)
on the east end has been installed to facilitate access to the
greenhouse while reducing energy losses through the main
greenhouse door by acting as an air buﬀer.
2.2. Network, communication, and telemetry
The greenhouse has a local Ethernet network that links
diﬀerent components through an EISX9-100T CTRLink
Ethernet hub. An embedded computer based on a
400 MHz ARM processor, 64 MB RAM, 64 MB Flash
and running embedded linux manages the network and
exchanges the information between the outside world and
the data acquisition system. The computer has 2 Ethernet
ports and 4 serial ports. Communication with other parts
of GH is via Ethernet and IP protocols. An SDX-1100
modem (serial communication) uses the MSAT satellite
to transmit data south, while commands can also be sent
Fig. 3 shows the ACMG components of the communi-
cation infrastructure. Transmission is ﬁle-based in both
directions. The embedded computer collects the data gath-
ered by the data acquisition and control (DAQ& CTRL)
system. A UNIX (Solaris) workstation at the operations
center at Simon Fraser University (SFU) controls the com-
munication, providing autonomous fetching of data from
the embedded computer, and can command the packet
satellite terminal in the greenhouse to reboot the DAQ &
CTRL system when it detects failure conditions. Upon
reception of data, and as represented in Fig. 4, the informa-
tion is processed and stored, and then made available to the
researchers through the Internet.
Also, the operations center can send commands to the
data acquisition and control (DAQ & CTRL) system to
modify the mode of operation. These diﬀerent modes of
operation will be described in the operation section of this
2.3. Data acquisition and control
The data acquisition is performed by a National Instru-
ments Field Point controller (FP-2015) and distributed I/O
(two AI-100 analog input modules, four TC-120 thermo-
couple input modules, one DI-330 digital input module
and two RLY-420 relay modules). Using 512 MB non-vol-
atile storage and 32 MB of RAM, the controller is able to
manage up to nine I/O modules, has one RS232 serial port
and is accessible through an Ethernet port.
The DAQ & CTRL modules are dedicated to acquiring
sensors data and to triggering the actuators that will be
mentioned in the following subsections. Fig. 5 depicts the
interactions between the acquisition modules and the con-
troller. The analog inputs include DC voltage inputs and
thermocouple element inputs. Discrete inputs are used to
sense the state of operation and relays are designed to trig-
ger actions of diﬀerent systems. The RS232 port of the con-
troller acquires the weather station data.
The main control loop operates at a frequency of 5 s
per cycle. The numerical average of each data point is
logged every 10 min and is appended to a continuous
global archive ﬁle on the FP-2015 controller ﬂash memo-
ry. In addition to writing this data line to the continuous-
ly growing main log ﬁle, the data is written to three other
ﬁles: the ﬁrst is a single log ﬁle that only contains the
information from the last 10-min log, the second is a 2-
h ﬁle that only contains the most recent 2 h of log infor-
mation, and the third contains the most recent 24 h of
logging information. The data acquisition archiving
scheme is depicted in Fig. 6. Under normal operations,
approximately every two hours, the autonomous systems
at the operations center send a request via the embedded
Fig. 3. Greenhouse network and communication system.
Fig. 4. Operations center (SFU) communication system and data
1250 R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259
computer over the satellite link for a 2-h data ﬁle. The
embedded computer requests the 2-h ﬁle from the FP-
2015, and sends over the satellite link. The embedded
computer also automatically fetches a backup 24-h data
ﬁle from the FP-2015. In case of communication outages,
a real-time 24-h ﬁle is autonomously requested by the
operations center in order to ﬁll in any gaps of data that
may have been created. It is also possible to request the
backup 24-h ﬁles, which are stored permanently in the
embedded computer’s storage, after periods of extended
communication loss. Ten minute real-time data may also
be fetched on-demand by the operations center, as can
embedded-computer housekeeping and status data ﬁles.
2.4. Plant growth
The core of the plant growth system is a drip-line system
(American AgriTech 8 tray Jetstream) similar to the one
shown in Fig. 7. This system allows for reduced reservoir
volumes and a potentially lower power consumption
depending on irrigation settings compared to an ebb and
ﬂow system. The plants are grown in rockwool slabs and
are watered every hour for a speciﬁc duration. As for all
drip-line systems, the excess water is drained back into
A Flojet Quad II diaphragm pump (model 4306 series) is
used to feed the drip lines and is activated by the controller.
A Plant Products 20-20-20 (with micronutrients) general
purpose fertilizer is used to prepare the nutrient solution
(approximately 100 mg/L total N). Presently, the nutrient
solution composition is monitored via HI2910B/5 pH sen-
sors and HI3001 electrical conductivity (EC) measure-
ments, supplied by Hanna Instruments. These parameters
are monitored; however, solution maintenance systems
(i.e., Nutrient injection pumps, acid and base injectors) have
not yet been deployed or incorporated into the power bud-
get. The solution temperature is also monitored for opera-
tional reasons that will be explained later in the paper.
The greenhouse has two distinct growth trays to accom-
modate dual operational requirements (to be discussed in a
later section): one for the fall crop season and another for
the spring crop season. The crop growth process is moni-
tored by two StarDot Technologies NetCamcameras
that provide daily or on-demand pictures to qualitatively
evaluate growth progress. Fig. 8 summarizes the plant sys-
2.5. Environment control system
The environment control system of the greenhouse is of
major importance due to the extreme climate of the green-
house location. At this stage of the project, only the tem-
perature is controlled, and this is done through the use of
heaters and exhaust fans. However, other environmental
parameters are logged for comparative analysis.
Fig. 7. American AgriTech Jetstream growth tray.
Fig. 6. Data acquisition archiving.
Fig. 5. Data acquisition and control.
R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259 1251
Table 1 shows the diﬀerent sensors used in the green-
house for the environmental monitoring.
Only the temperature sensors (type Tthermocouples) in
the vicinity of the crops provide feedback to the climate
control algorithm. The other thermocouples are used to
monitor the general climate of the greenhouse in addition
to the outside temperature. Two custom sensors (Tand
RH sensors), developed at the University of Guelph, mea-
sure the temperature and relative humidity inside the
greenhouse. They both use RS232 protocol to transmit
their data. Because of the limitation of RS232 ports on
the NI controller at hand, the embedded computer is used
to gather the custom sensor data. The data are then read by
the FP-2015 via the local Ethernet network.
The quantum sensors (model QSO-SUN), the pyranom-
eter sensors (model PYR) and the ultra-violet sensors
(model UVS) from Apogee Instruments indicate the level
of solar radiation in diﬀerent wavelength bands. Also, a
customized PORTLOG-M weather station from RainWise
monitors the outdoor climate parameters that inﬂuence the
greenhouse interior climate. Finally, an external webcam
(StarDot Technologies NetCam) provides a general out-
look of the weather and the structural integrity of the
greenhouse. Fig. 9 summarizes the sensor architecture of
There are three main actuators in the temperature con-
trol system: the heaters, the circulation fans and the
exhaust fans, as illustrated at Fig. 10.
The three heaters are equally spaced along the northern
wall. Each of them is a commercial non-vented open ﬂame
propane greenhouse heater, delivering 21.1 MJ (20 000
BTU) each (model C1-200SR from Southern Burner
Co.). Twelve 86 kg (190 lbs) standard LP tanks supply pro-
pane to the heaters. The tanks are located outside along the
north end of the greenhouse. Each tank is ﬁlled with 45 kg
(100 lbs) of propane.
In addition to heat-induced movement fans on top of
each heater, two 220 L/s (466 CFM) Schaefer circulation
fans are installed allowing equal distribution of the heat
in the greenhouse. To cool down the greenhouse, two
30.5 cm (12 in.) exhaust fans are situated on the upper part
of the west wall. At an exhaust capacity of approximately
377.5 L/s (800 CFM) each, they are used when the inside
temperature reaches a predetermined threshold.
2.5.3. Climate control
The climate control system currently comprises of only
basic greenhouse temperature control within the limits of
Humidity Relative humidity sensor
300 to 1100 nm Pyranometer
400 to 700 nm Quantum sensor
250 to 400 nm UV sensor
Relative humidity RH sensor
Wind speed Anemometer
Wind direction Wind vane
Barometric pressure Barometer
Solar radiation Pyranometer
Outside picture Webcam
Fig. 8. Plant system components and interactions.
Fig. 9. Environment control system sensing elements.
Fig. 10. Environment control system actuators.
1252 R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259
the previously described actuators. Fig. 11 depicts the con-
trol logic, where the X-axis represents the temperature
above the canopy and the state sequence of the temperature
control is determined by a virtual dot following the
arrowed hysteresis curve given a temperature variation.
Given a desired minimum temperature, the heaters are trig-
gered on at a temperature above that minimum to account
for heat transfer inertia. During heating events, only two
heaters are turned on at a time and the three pairs of heat-
ers are cycled in a predetermined sequence. Following a
hysteresis control strategy, the heaters are shut oﬀ at a tem-
perature slightly higher than the one at which they were ini-
tially turned on.
As for the cooling strategy, the exhaust fans are turned
on before the temperature reaches the maximum setpoint.
Both fans are always turned on at the same time. Although
they are very power consuming, they are only needed dur-
ing high solar radiance, which in turn implies high power
output from the solar panels. The fan control also follows
a hysteresis control scheme.
The circulation fans can be actuated by two modes: con-
tinuous or periodic operations. The continuous operation
allows the fans to operate unless they are manually turned
oﬀ. This mode is activated by a direct command from the
operations center and is used only occasionally. The peri-
odic operational is the nominal operation mode and
requires a frequency and duration input. The nominal set-
tings for this mode are periods of 15 min for a 5-min acti-
vation timeframe (duty cycle of 33%).
2.6. Power system
Since the greenhouse is located where there is no infra-
structure, electrical power is provided by a hybrid solar-
wind system. In addition to wind-power generators and
solar panels, the system is composed of charge controllers,
a battery bank, a load dump, a line switch logic and distri-
bution lines. Fig. 12 illustrates each component of the sys-
tem and their interconnections. The power system is
intentionally isolated (not controlled) from the DAQ &
CTRL system to allow robustness and continuous opera-
tion even in the case of a computer failure. However, power
production and consumption are monitored.
2.6.1. Power production, storage, and charge control
The system has two separate lines of power production,
each of which contains three solar panels, one wind gener-
ator and one charge/load controller. The two power pro-
duction lines feed the same battery bank. The relevant
technical information of the production system is summa-
rized in Table 2.
The Xantrex C60 charge/load controllers manage the
power to and from the battery bank. Fig. 13 represents
the control logic of the charging sequence, where the X-axis
represents the main bus voltage and the state sequence of
the charge logic control is determined by a virtual dot fol-
lowing the arrowed hysteresis curve given a voltage varia-
tion. Batteries are always in charging mode unless the
battery bank voltage approaches its maximum voltage set-
point. In this case, the excess power produced by the wind
generators or the solar panels is diverted to a load resistor
immersed in a barrel of an antifreeze/water mixture. Each
production line has one 1600 W water-heating element that
converts the excess electrical energy into heat. The heat
inertia of the barrel of liquid then provides extra heating
capacity to help maintain/buﬀer the temperature inside
2.6.2. Load control and power distribution
Two diﬀerent sets of lines supply the greenhouse subsys-
tems: the critical lines and the utility lines. The critical lines
feed the computers and the communication system; the
utility lines supply the pumps and fans. Each distribution
set contains one 24 V line, two 12 V lines and one 5 V line.
When the charge of the battery bank goes below a thresh-
old voltage, the utility line is ﬁrst disconnected by the char-
ge controller, making available more power to charge the
batteries. If the voltage still drops below the second thresh-
old, then the critical line is also disconnected and all the
power generation is directed towards the charging of the
batteries. When the battery bank voltage rises, the critical
line is ﬁrst reconnected, followed by the utility line. This
logic is represented by the hysteresis curves in Fig. 14,
where the X-axis represents the main bus voltage and the
state sequence of the load logic control is determined by
a virtual dot following the arrowed hysteresis curve given
a voltage variation.
Fig. 11. Temperature control strategy.
Fig. 12. Power generation, accumulation, and distribution.
R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259 1253
Finally, as shown in Fig. 12, a line switch (designed in-
house) is used to redirect the utility line power to the crit-
ical line in case of a failure of the latter. This feature adds
robustness to the power system.
2.6.3. Power monitoring
Although the power system is not controlled by the
DAQ & CTRL system, it is still monitored to evaluate its
performance. Shunt resistors are placed in series with the
line and the voltage drop is measured. The measurement
is designed so that a voltage drop of 1 mV at the shunt
resistor corresponds to 1A of current passing through it.
The power is then simply calculated.
The wind generators, solar panels, load dump, and load
consumption power are currently monitored, as depicted in
3. Greenhouse operation
The greenhouse is meant to operate year-round, but
with diﬀerent modes of operation, depending on the condi-
tions: normal fall, normal spring, safe, and dormant
modes. In the fall mode, only the growth tray with the fall
crop is activated. That allows the spring crop system to
remain purged and dry to be able to pass through the cold
winter period without damage, as will be explained later in
this section. Obviously, the spring mode activates the
spring crop system. Both normal modes are activated dur-
ing the crop-growing period. During that time, all systems
are operational. However, the safe mode stops the environ-
ment control system (heating and cooling) to conserve elec-
trical power in preparation for the harsh winter season.
Data acquisition is still performed to monitor the condition
of the greenhouse. In the case of a dangerously low voltage
level in the batteries (risk of freezing), it is possible to
switch to a dormant mode where all systems are shut-down
except the communication infrastructure to conserve ener-
gy. In this mode, even the communication system is in low
The following sequence of events details the current
planned yearly operations of the ACMG.
The greenhouse is situated in a very harsh and remote
environment. Hence, the research team can only do work
to improve and repair the systems during 3–4 weeks a year
(ﬁeld season). The month of July is usually the most
human-friendly and logistically compatible time of the year
for such work. Upon arrival, the team investigates the
greenhouse status and begins maintenance. Also, the fall
Power production and storage characteristics
Wind generators Total of 2 AIR industrial wind generators providing a maximum total power of 800 W @ 24 V for wind speed of 45 km/h
Solar panels Total of 6 Shell SM110-24 solar panels for a total peak power production of 660 W @ 35 V (irradiance level 1000 W/m
DC/DC converters Four converters SPS2412-5 (24 to 12 V) from Soltek Powersource
Two converters VTC60-24-5 (24 to 5 V) from analytic systems
Battery bank 36 Extreme EX-1000 12V batteries in a serial/parallel conﬁguration providing 1800 A h @ 24 V nominal
Fig. 13. Charge control logic.
Fig. 14. Load control logic.
Fig. 15. Power system monitoring.
1254 R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259
growth tray is re-outﬁtted and seeded. Manual watering is
performed until the data acquisition and control system is
brought on line after any improvements/maintenance have
been completed, allowing germination of the crop. New
propane tanks replace empty ones and heaters are ignited.
The spring growth tray (for the following year) is seeded,
although its watering system is purged with air (no germi-
nation of the spring crop yet). No water other than in the
nutrient reservoir is present in the spring crop watering sys-
tem. At the end of the ﬁeld season and upon departure of
the site, the crop in the fall tray has already started to grow.
The greenhouse operates in the normal-fall mode. Tem-
perature is controlled by means of heating and venting. The
data acquisition system returns daily data about the system
performance and operation. Crop yield is qualitatively
evaluated with the web cam pictures, as depicted at
Fig. 16. On this ﬁgure, the spring crop tray is also visible,
and it can be seen that no crop is growing.
After nearly two months of autonomous operation, the
fall crop has grown signiﬁcantly. However, the propane
level is getting low and the heating system has to be shut
oﬀ. In order to have propane fuel for the next ﬁeld season,
a shut-oﬀ limit of 20% propane level (80% consumption) is
set. Also, the winter period has less sun and usually milder
wind activity. The batteries can survive down to 60 Cif
fully charged. Hence, power consumption has to be
In the following example illustrated by Figs. 17 and 18,
the safe mode command was sent via the operations center
on 10 October 2004, resulting in a distinct drop in the
power demand and the inside greenhouse temperature.
This conﬁrmed that the environmental control system is
shut down in addition to the watering system of the fall
plant growth system. Fig. 17 shows the controlled green-
house inside temperature and the outside temperature for
one month (September–October 2004). Starting October
10th, it is clear that the inside temperature follows with a
certain margin of error the outside temperature (this mar-
gin is induced by the natural radiation heating of a green-
house). In Fig. 18, it can be seen that the critical line is still
in operation keeping the DAQ & CTRL system and the
communication system active, although the utility line
power consumption is near zero since all utility loads have
been shut down. It is interesting to note that although all
loads have been turned oﬀ, there is still a marginal power
reading in the utility line from the power-sensing device
(shunt resistor). Although still under study, the inaccurate
reading seems to originate from the sensing loop and not
from a malfunctioning load.
The major impact of the safe mode is the lack of heating
while outside temperature goes below zero celsius, implying
Fig. 16. Web cam picture taken remotely on 9 August 2004 (fall crop on
top right, spring crop on bottom left).
Fig. 17. Greenhouse inside temperature over time.
Fig. 18. Power consumption of critical and utility lines.
R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259 1255
the freezing of any water in the greenhouse. However, the
safe mode also shuts down the watering of the fall plant
growth tray. As expected, the remaining water in the pump
circuit might damage the fall watering system and is pres-
ently treated as an unavoidable collateral damage, in addi-
tion to the freezing of the fall crops. Since the spring crop
watering system has been purged, no damage is expected in
The general greenhouse behavior during winter is mon-
itored: open-loop temperature and humidity inside and
outside the greenhouse, and especially the power system
performance. If there is enough light, pictures are also tak-
en to assess the environment, as shown in Fig. 19.
In order to get data for the entire winter, strict power
management is required. Fig. 20 shows an example of
power data gathered during the beginning of safe mode
operation. If the safe mode still draws too much power, a
dormant mode can be activated from the operations center
powering oﬀ all systems except the communication system.
During this period, the sun will start to warm up the
greenhouse to a level where the water in the reservoirs will
have melted. Being still in safe mode, the water tempera-
ture of the spring growth tray reservoir is monitored. After
several days of the water temperature being above 5 C, it
can be inferred that the water is in liquid state. Then, the
watering process of the spring crop can begin as soon as
the water temperature reach a predetermined temperature
that ensure decent germination rate. The operation mode
of the greenhouse is then turned to normal-spring mode.
Temperature is controlled but only by means of venting.
The heaters cannot presently be turned on remotely
because of propane valve mechanical fail-safe operation.
If the temperature inside the greenhouse is suﬃciently con-
stant and favorable for plant growth, crops will germinate
and start to grow. Upon the crew arrival for the ﬁeld sea-
son in July, remotely started crops would have been
The current operation sequence has some advantages
and drawbacks that will be discussed in the next section
of the paper.
4. The greenhouse systems and operation evolution
This research project has continuously evolved since its
inception. This section will describe the evolution of the
greenhouse systems and highlight some lessons learned.
Fig. 21 gives a broad summary of the improvements
throughout the project.
The greenhouse structure was erected during the sum-
mer of 2002. A preliminary suite of sensors was also
deployed to assess the requirements for the environment
The summer ﬁeld season of 2003 was dedicated to the
installation of the environment control system (heaters
and ventilation), the data acquisition and control system,
the power system, the communication system and the ﬁrst
plant growth system. At that time, the power system con-
tained one wind generator and 3 solar panels managed
by one dedicated power controller. Also, a battery bank
of 6 batteries was installed. A preliminary heat recovery
system was implemented with one dump load. The current
DAQ & CTRL system was installed at that time and was
Fig. 19. Webcam picture of the ACMG after a snowstorm at the end of
Fig. 20. Power monitoring.
1256 R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259
basically the same as described in the preceding section,
although some sensors have been added as the project
has evolved. The communication system was the same
MSAT modem telemetry system. Since the initial green-
house deployment was focused on the establishment of
the greenhouse structure, environmental sensors, and con-
trol actuators, it left little resources for actual plant growth
system development. The plant growth system was then a
basic ebb-and-ﬂow nutrient delivery system. Low cost/
low power supply pumps were used to feed a basic
1.2 m ·2.4 m ﬂood bench from Grotek.
Upon completion of the ﬁeld season and departure from
the site in August 2003, the ﬁrst autonomous and remotely
monitored greenhouse in the world was successfully oper-
ating. The ﬁrst crop was growing within an average tem-
perature range of 15–30 C. However, the operations
center lost communication with the greenhouse on Septem-
ber 11th 2003. After several unsuccessful attempts to rees-
tablish the communication link, a helicopter mission went
to the site in May 2004 to investigate the failure. It was
concluded that the capacity of the power system was not
suﬃcient to deal with extreme cold (due to lack of charge)
and the production of power should be enhanced. The
demonstration of a remotely initiated spring crop growth
was not attempted due to failure of subsystems that could
not be replaced during the May 2004 visit.
To address the issues of the previous year, the power sys-
tem was upgraded during the 2004 ﬁeld season. Doubling the
battery capacity (6 to 12 batteries), doubling the power pro-
duction and adding a second power controller were among
the modiﬁcations. Also, the heat recovery system was mod-
iﬁed to incorporate the second dump load from the new
power controller. In addition, the current plant growth sys-
tem (drip-line) was installed during that ﬁeld season.
The beginning of the autonomous operation went well
until another communication outage occurred on August
11th 2004; only two weeks after the research team left the
site. The symptom of this communications outage was very
similar to the one of the previous year, despite the improve-
ment of the power system. However, after much hypothe-
sizing and veriﬁcation, it appeared that the satellite
Fig. 21. High-level summary of greenhouse improvements.
Fig. 22. System improvements planned for summer 2006.
R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259 1257
communications terminal system was not operating in a
way that was optimized for such high latitudes, due to both
software and conﬁguration issues in the MSAT network
and the satellite terminal. This was adjusted in collabora-
tion between SFU, Infosat, and MSV (the operators of
the MSAT spacecraft). The communication was reestab-
lished on September 14th 2004; however it was too late
to save the fall crop. The system is designed to generate a
general shut down command after three days of no com-
munication between the ACMG and the operations center.
Experimentation on environmental control continued,
and the system was put in safe mode on October 10th.
Once again, communication was lost in December 2004,
this time, due to a failure in the power system. It was later
determined that a blown fuse on the 12-V power line pro-
viding power to the communications system had caused the
failure. The cause of the blown fuse was not determined
since all devices were working properly when examined
the following ﬁeld season.
The 2005 ﬁeld season was dedicated to consolidation of
assets and no major changes were performed with the
exception of a power storage increase. The 12 batteries
from the 2004 system were replaced with 36 batteries, total-
ing 1800 A h at 24 V. The controller and embedded com-
puter cabinets were insulated to protect them from
extreme cold. The line switching logic on the distribution
lines was integrated, to safeguard against the type of failure
that was experienced during the fall of 2004.
The research team left the site at the end of July 2005
with the fall growth process on its way. Although the plants
seemed to demonstrate signs of illness after some time, the
fall growth season was successfully conducted. Although
the exact cause of the plant growth issues has not yet been
determined, it is believed that the lack of an active nutrient
control mechanism is the limiting factor at this time. On
September 7, 2005, the safe mode command was issued,
stopping circulation fans to conserve power and stopping
heating to conserve propane. All systems are performing
nominally during the writing of this paper (October 2005).
During the upcoming winter, the power system will be
closely monitored to evaluate its eﬃciency to get through
the harsh conditions and keep a suﬃcient level of charge
to allow the start of the system in spring 2006. The current
performance results are very encouraging and suggest that
remote spring startup should be successful.
5. Discussion on future research and development activities
The ﬁrst two ﬁeld seasons of operation of the ACMG
(2003 and 2004) generated many valuable lessons. It has
become clear that a second ‘‘development’’ greenhouse,
in a more accessible location, is required in order to expe-
dite the implementation of new technologies and algo-
rithms in the ACMG. Access to the Arctic greenhouse is
only possible for a few weeks every summer. Before deploy-
ing systems in the Arctic, testing is conducted in the labo-
ratory; however, the ﬁrst time they are deployed in an
actual greenhouse is on Devon Island. Between the inevita-
ble surprises, a very short ﬁeld season, and a lack of infra-
structure on Devon Island, the implementation of new and
complex systems in the Arctic greenhouse is limited. To
address this important issue, the establishment of a readily
accessible development greenhouse similar to the one in the
Arctic was proposed. Currently under construction, this
development greenhouse is on the grounds of the Canadian
Space Agency headquarters in Quebec, Canada.
For the next ﬁeld season (summer 2006), several system
improvements are planned based on the past experiences,
current investigations and future needs, as listed in
Fig. 22. The issues with crop health during the fall season
in 2005 have served to expedite the implementation of a
more robust nutrient feedback control system. A new con-
troller will be added to control this new addition and pave
the way to a more distributed DAQ & CTRL system. Also,
the environmental control system is not optimal in its cur-
rent operation and is currently the subject of improvement
eﬀorts. The CSA-based development greenhouse will be a
great resource to test new hardware and software. Areas
under investigation are new fuel/heating systems, heat
recovery systems to take better advantage of the excess
power production and more advanced control algorithms
for climate control. Remote ignition of the heaters will also
be a priority for system improvement. Finally, based on the
current ﬁeld season data of the power system, ﬁne-tuning
might involve a more powerful generation system and/or
a third production device independent from the two cur-
rent types of energy production.
The primary greenhouse systems have been performing
successfully throughout this year, enabling more fundamen-
tal research to be performed in the coming years. Four main
areas have been identiﬁed and will form the basis of engineer-
ing and space science research: biological systems, environ-
mental control systems, power and communication
systems, and operations strategies for exploration purposes.
As the planning for the human exploration of the moon
and Mars continues, the use of terrestrial analogues of
space environments will prove to be a valuable resource
for preparing scientiﬁc investigations and science opera-
tions, developing technologies, understanding/developing
exploration protocols (i.e., traverse standard operating
procedures), and for the training of future ﬂight crews.
This project is demonstrating the value of such analogue
sites, since the development team must deal with con-
straints that cannot be reproduced nor anticipated in a lab-
oratory environment. It will eventually see life sciences
being carried out within the greenhouse. The concept of
how a greenhouse can be incorporated into a research base
will also be explored.
1258 R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259
This greenhouse serves as an initial experimental, ﬁeld-
deployed test bed that supports ﬁeld research of inherent
and immediate value, and from which lessons may be
learned to support the design and implementation of sec-
ond-generation ﬁeld facilities enabling higher ﬁdelity dem-
onstrations. It took several ﬁeld seasons of development to
attain the level of readiness we are now capable of. The sys-
tem failures over the years have pointed us underestimated
challenges in the design. These insights have helped us
improve the greenhouse operation to a point where we
are now able to operate the critical systems year around
in an extreme environment. Now that the engineering por-
tion of the project is mature enough, plant science will start
to take more ﬁeld season time and eﬀort. Ultimately,
through a sequential and iterative program of experimenta-
tion, it is hoped that a better understanding of the opera-
tional challenges faced by future astronauts on the
surface of Mars (or other planetary bodies) will be gained.
To some extent, the technology developed (and to be
developed) for the Arthur Clarke Mars Greenhouse could
have direct applications for Canada’s estimated $2 billion
commercial greenhouse industry. In the future, new tech-
nologies for optimizing greenhouse energy use eﬃciency,
and autonomous and remote operation, could provide
direct beneﬁts to this ground-based industry.
The Haughton-Mars Project (HMP)’s Arthur Clarke
Mars Greenhouse (ACMG) was donated by SpaceRef
Interactive, Inc. and established at the HMP Base Camp
(now HMP Research Station) with initial sponsorship sup-
port from NASA. The ACMG facility is currently man-
aged and operated by the Mars Institute, in partnership
with the SETI Institute and Simon Fraser University.
The authors would like to thank the Ontario Centers of
Excellence (OCE) and MacDonald Dettweiler & Associates
Ltd. (MDA) for their partial ﬁnancial support for the re-
search described in this paper.
Allen, J. Biosphere 2: Description, Purpose, and Conceptual Design.
International Conference on Life Support & Biospherics, University of
Alabama in Huntsville, Alabama, 1992.
Andre, M., Massimino, D. Growth of plants at reduced pressures:
experiments in wheat – technological advantages and constraints. Adv.
Space Res. 12 (5), 97–106, 1992.
Bartsev, S.I., Mezhevikin, V.V., Okhonin, V.A. Bios-4 as an embodiment
of CELSS development conception. Adv. Space Res. 18 (1/2), 201–204,
Bucklin, R.A., Leary, J.D., Rygalov, V., et al. Design parameters for Mars
deployable greenhouse. Proceedings of the 31th International Confer-
ence on Environmental Systems. Orlando, FL, SAE aerospace:
Technical Paper 2001-01-2428, 2001.
Clawson, J.M., Hoeln, A., Wheeler, R.M. Inﬂatable transparent structure
for Mars greenhouse applications. Proceedings of the 35th Interna-
tional Conference on Environmental Systems. Rome, Italy, SAE
aerospace: Technical paper 2005-01-2846, 2005.
Dixon, M., Wehkamp, C.A., Stasiak, M. Physiological response of lettuce
(Lactuca sativa) to reduced atmospheric pressure, in: Proceedings of
the 35th International Conference on Environmental Systems. Rome,
Italy, SAE aerospace: Technical paper 2005-01-3074, 2005.
Drysdale, A., Collins, R. Mars base zero – a terrestrial analog, in:
Proceedings of the 35th International Conference on Environmental
Systems. Rome, Italy, SAE aerospace: Technical paper 2005-01-2756,
Goto, E., Iwabuchi, K., Takakura, T. Eﬀect of reduced total air pressure
on spinach growth. J. Agric. Meteorol. 51, 139–143, 1995.
Janssen, E.G.O.N., Tse, T.C., Mas, J.L., et al. Design approach and
implementation of a Mars surface food production unit, in: Proceed-
ings of the 35th International Conference on Environmental Systems.
Rome, Italy, SAE aerospace: Technical paper 2005-01-2824, 2005.
Lasseur, C., Paille
´, C., Lamaze, B., et al. MELISSA: overview of the
project and perspective, in: Proceedings of the 35th International
Conference on Environmental Systems. Rome, Italy, SAE aerospace:
Technical paper 2005-01-3066, 2005.
Perino, M.A., Lobascio, C., Pastor, S. Greenhouse: a strategic element to
support humans in space, in: Proceedings of the 7th ISU Annual
International Symposium. Strasbourg, International Space University,
Rygalov, V.Y., Buklin, R.A., Drysdale, A.E., et al. Low pressure
greenhouse concept for Mars: atmospheric composition, in: Proceed-
ings of the 32th International Conference on Environmental Systems.
San Antonio, TX, SAE aerospace: Technical Paper 2002-01-2392,
Salisbury, F.B., Gitelson, J.I., Lisovsky, G.M. Bios-3: Siberian experi-
ments in bioregenerative life support. Bioscience 47, 575–585, 1997.
Scarascia-Mugnozza, G., Schettini, E. Preliminary indication on the
technological systems for an Italian space greenhouse, in: Proceed-
ings of the 32th International Conference on Environmental Systems.
San Antonio, TX, SAE aerospace: Technical paper 2002-01-2486,
Wheeler, R.M., Martin-Brennan, C. Mars Greenhouses: concepts and
challenges, in: Proceedings from a 1999 Workshop, NASA Tech.
Mem. 208577, 2000.
Wheeler, R.M., Stutte, G.W., Subbarao, G.V. Plant growth and human
life support for space travel, in: Handbook of Plant and Crop
Physiology. M. Pessarakli, New York, pp. 925–941, 2001.
R. Giroux et al. / Advances in Space Research 38 (2006) 1248–1259 1259