Observing Nightlights from Space wi= th TEMPO
James = Carr1,*, Xiong Liu2, Brian Baker3, Kelly Chance2
[1]<= /a> 1Carr Astronautics Corp., Greenbelt, MD, USA
2Harvard-Smithsonian Center for Astrophysics, Cambridge, MA, US= A
3Ball Aerospace and Technologies Corp., Boulder, CO, USA=
Received
05 December 2016, Accepted 08 February 2017
Abstract
The Tropospheric Emissions: Monitoring=
of
Pollution (TEMPO) instrument is a NASA Earth Venture Instrument set to fly =
in
the 2018-2021 timeframe and cover North America from a geostationary
orbit. TEMPO is a powerful
spectrometer sampling UV and visible wavelengths in two channels (290-490 nm
and 540-740 nm) in 0.2 nm steps.
This spectroscopic capability will enable TEMPO to characterize many
types of artificial lighting at night in addition to its regular daytime air
quality and atmospheric chemistry mission. This paper describes the TEMPO
instrument and our planned approach to make nighttime observations. A spect=
ral
retrieval algorithm will estimate the contributions from sources in a spect=
ral
library to map out the usage of various types of lighting (e.g., Hg vapor, =
high
and low pressure Na vapor, LED, fluorescent, and incandescent) over North
America. We calculate the sensitivities of such nightlight retrievals using=
the
measured properties of our flight detectors and optics, and conclude that T=
EMPO
will complement well the VIIRS Day-Night Band for the study of artificial
lighting at night from space by offering a new and powerful capability to
discriminate between lighting types.
Keywords: Nightlights, Spectroscopy, VIIRS day-n=
ight
band
1. Introduction
The Tropospheric Emissions: Monitoring of Pollut=
ion
(TEMPO) instrument [1] is supported by the NASA Earth Venture Instrument (E=
V-I)
program. It will fly as a hosted payload on a geostationary commercial
communications satellite sometime between 2018 and 2021 depending on availa=
ble
flight opportunities. TEMPO w=
ill
cover all the United States except for Alaska and Hawaii, Mexico as far sou=
th
as Mexico City, Canada as far north as the Alberta oil sands, and much of t=
he
Caribbean. This entire area will be covered once per hour with the possibil=
ity
of more rapidly revisiting smaller subareas. TEMPO collects and spectrally
disperses reflected sunlight to reveal the presence of various trace gas
species through their distinctive absorption spectra. Its primary mission is to monitor =
air
quality and study the chemistry of trace gases in the atmosphere and support
the study of aerosols, clouds, and chlorophyll fluorescence. TEMPO will be
capable of producing vertical column densities of various gas species, incl=
uding
O3, NO2, SO2, H2CO, C2H2O2, BrO, IO, and H2O as well as ozone profiles
including tropospheric ozone on an hourly basis. The primary mission of TEM=
PO
is conducted only during daylight hours, which leaves TEMPO available at ni=
ght
for the spectroscopic observation of nightlights, which is the subject of t=
his
paper. Artificial lighting is a marker for human economic activity with
implications for energy efficiency, darkness of the nighttime sky, and
ecological impacts on both humans and other lifeforms. These implications are explored as=
part
of the Artificial Lighting at Night (ALAN) conference series and this paper=
is
based on our presentation at ALAN 2016. For a review of nightlights imagery
sources and several research application areas with a good bibliography, see
Kyba et. al. [2].
Interest in the observation of nightlights from
space has grown since the launch of the Visible Infrared Imaging Radiometer
Suite (VIIRS) with its Day-Night Band (DNB) on the Suomi National
Polar-orbiting Partnership spacecraft in 2011. The VIIRS-DNB represents a
significant improvement [3] in capability relative to the legacy Operational
Linescan System (OLS) flown on Defense Meteorological Satellite Program (DM=
SP)
spacecraft since 1972 (for a good review of research applications using OLS
nighttime imagery conducted over the years, see Huang et. al. [4]). The
VIIRS-DNB capabilities for worldwide mapping of light pollution were recent=
ly
featured by the publication of an atlas of night sky quality [5] that updat=
es a
similar atlas produced from OLS data in 2001 [6]. However, the nightlight
sensing capability of VIIRS is monochromatic leaving it unable to distingui=
sh
between different types of artificial lighting. It is known that the spectr=
al
characteristics of artificial lights are important. For example, blue light=
is
scattered more strongly in the atmosphere (Rayleigh scattering) than red li=
ght
and therefore blue light contributes more to the brightening of the night s=
ky
than red light, which is the same reason that scattered sun light makes the=
sky
appear blue. A second example is the impact on human health through the
mechanism of melatonin suppression in response to exposure to primarily blue
light. Concern over the disruption of circadian rhythms from artificial
lighting at night and its impact on human health including elevated cancer
risks prompted the American Medical Association to issue a policy statement=
on
light pollution in 2012 [7]. As Light Emitting Diode (LED) luminaires become
more prevalent in outdoor lighting, the spectral characteristics of nightli=
ghts
will be changing too, with typically more light being emitted in the bluer
wavelengths, which is a concern relative to both the examples just cited. T=
he
VIIRS-DNB, with a lower wavelength cutoff at 500 nm, is also less well suit=
ed
to observing LED lighting that emits a significant fraction of its light in
bluer wavelength regimes.
Multispectral and hyperspectral nighttime imagery is therefore desir=
able
to aid in the analysis of nighttime lighting patterns and particularly the
adoption of LED technology. A collection of nighttime multispectral images =
of
cities is presently available to researchers from an astronaut photography
program conducted from the International Space Station (ISS). An atlas exis=
ts
to facilitate accessing these photographs from an even larger collection in=
the
NASA gateway for astronaut photography [8]. These photographs have been use=
d to
provide insight into changes in lighting of some European cities where patt=
erns
of illumination have been observed over time of night and year to year and
lighting types in use have been identified by their color [2]. We anticipate
that TEMPO and possibly some other spectrometers that we discuss in Section=
5
should provide additional capabilities to more richly describe the artifici=
al
lighting of our planet at night.
2. TEMPO Instrument and Data<= o:p>
Fig. 1 shows the components of the TEMPO
instrument. Light enters thro=
ugh
the calibration mechanism, which can be open for collecting Earth data, clo=
sed
for safety, or have either one of two diffuser plates inserted into the opt=
ical
path for calibration using the sun as a source. A scan mechanism moves a flat scan
mirror that allows the TEMPO field of view to be scanned across the Earth f=
rom
east to west during collection of a scan.&=
nbsp;
A full scan nominally consists of 1282 scan steps in about one hour.=
At
each scan step, light reflected off the scan mirror is projected by a
three-mirror telescope onto the entrance slit of an Offner-type spectrometer
where it is spectrally dispersed onto two Charged Coupled Device (CCD) focal
plane arrays nominally operated at -20¡ÆC to reduce dark current. Each CCD
records the spectra of ~2038 spatial pixels arranged north to south in ~1001
spectral bins, with exact numbers to be known after final assembly of the
spectrometer. One CCD covers the spectral range from 290 nm to 490 nm (UV to
blue) in 0.2 nm steps and the other CCD covers the spectral range from 540 =
nm
to 740 nm (green to red) also in 0.2 nm steps. The size of the spectral gap=
between
490 nm and 540 nm is dictated by the number of spectral pixels on the CCDs,=
the
physical restrictions on the devices as to how closely they can be spaced, =
as
well as the needs of the scientific community to adequately cover ozone
absorption in the visible for lower tropospheric ozone retrievals and extend
the spectral range to 740 nm for vegetation products using chlorophyll
fluorescence. The intrinsic spectral resolution of the TEMPO spectrometer is
0.6 nm (full width at half maximum).
The focal plane has a read-while-integrating
architecture, with the flexibility to specify the desired integration time =
per
frame and number of co-added frames. We co-add frames for daylight observat=
ions
to avoid having the CCD wells overfill while achieving a dwell time per spa=
tial
pixel equal to the integration time times the number of co-adds. Overfillin=
g is
not a concern for nightlights observations, so we will take advantage of the
operational flexibility of TEMPO to program it to integrate longer than
nominally planned for daylight observations and co-add fewer frames. Reduci=
ng
co-adds saves on read noise from the readout circuitry. The dominant noise
sources for nightlights observations are Poisson noise, mostly from the dark
current background, and read noise. We analyze the sensitivity of TEMPO for
nightlights observations using the measured characteristics of its flight
parts, including CCD dark currents and quantum efficiencies and the efficie=
ncy
of the spectrometer grating. Dark current will gradually increase in flight
with the total ionizing dose of radiation, resulting in best performance at=
the
beginning of life and graceful degradation thereafter.
Fig.
1. Components of the TEMPO Instrument are shown in an exploded view. The
optical system comprises a calibration mechanism, a scan mechanism, a
three-mirror telescope, and a grating spectrometer. Instrument control
electronics and thermal control surfaces will be integrated alongside TEMPO=
on
the nadir facing deck of a commercial communications satellite. The cant an=
gles
for TEMPO to orient it towards the desired field of regard will be known on=
ce
its exact longitude station has been determined after host spacecraft
selection.
A nominal full-scan level 1b product is a data
¡°cube¡± of 1282 (spatial) x 2038 (spatial) x 2002 (spectral) voxels. Each voxel holds a calibrated spec=
tral
radiance value and each spatial plane within the data cube has a nominal
wavelength assignment; however, there is a small spectral cross-coupling
between the two axes of the CCDs, so calibrated wavelengths are included as
metadata in the product to more precisely mark the spectral content of voxe=
ls.
Each spatial pixel has an instantaneous field of view equal to approximately
120 rad (East-West) x 41.8 rad (North-South), which corresp=
onds to a
footprint on the ground of 4.4 km x 2.1 km at the center of the field of re=
gard
(34.45¡ÆN, 94.72¡ÆW) when TEMPO is stationed at 100¡ÆW longitude above the
Equator. No remapping or resampling is performed on the spatial pixels to
compensate for spacecraft motion, scanner errors, or other geometric
distortions, but metadata is provided in the product to accurately geolocate
each spatial pixel. The footp=
rint
of a spatial pixel on the ground will grow in area as the limb is approached
introducing effectively more spatial averaging of sources within it. Scanni=
ng
across the field of regard is nominally in equal angle steps and there is no
compensation for the growth of the area of spatial pixels moving further off
nadir as is done with VIIRS. =
In
addition, TEMPO will collect dark images when the calibration mechanism is
closed before and after nominal Earth viewing, and TEMPO will collect solar
calibration images using one of two transmissive diffusers once a day. One diffuser is for daily use and =
the other
is a reference diffuser to be used once every few months to track any chang=
e in
the diffuser that is used daily. The solar calibration images will help to
trend instrument response over time and will be utilized to track CCD
performance in the way a typical flat-field image is used.
TEMPO data, including radiometrically calibrated=
and
geographically located level 1b radiances and higher level products, will be
first released to the public 4-6 months after the beginning of routine
operations. Thereafter, level 1b data will be available within ~3 hours of
observation through the Smithsonian Astrophysical Observatory (SAO) website,
kept online for 30 days, and then permanently archived at NASA¡¯s Atmospher=
ic
Science Data Center for public distribution.
3. Plan for Nighttime Observi=
ng
Nightlights observing will be conducted when TEM=
PO
is not performing its primary air quality and atmospheric chemistry mission.
However, there is an important safety constraint because the TEMPO spectrom=
eter
entrance slit can be damaged if intense sunlight impinges on it. To avoid t=
his,
we do not open the TEMPO aperture door whenever the sun is within 60¡Æ of t=
he
optical axis. This still leaves plenty of opportunities to collect nightlig=
hts
data, particularly in the northern hemisphere winter, as shown in Fig. 2 wh=
ere
solar illumination is depicted for three major metropolitan areas from the =
east
to the west across the field of regard as a function of U.S. Central Standa=
rd
Time (CST) and season (declination of the sun above the Equator). Nightlights observations over the =
East
Coast and Midwestern U.S. can be conducted prior to midnight CST and over t=
he
West Coast after midnight CST. The aperture door will need to be shut for u=
p to
eight hours roughly centered around midnight CST. Although in principle we could reo=
pen
the aperture door during the umbra of the eclipse of the sun by the Earth, =
we
are unlikely to do so because of the risk of not being able to close the
aperture in time.
Fig. 2. Nighttime observing with TEMPO depends on the season (solar declination), time of day, and location. Daylight observing times when the solar zenith a= ngle is less than 80¡Æ are indicated by the yellow fill and nighttime observing = times when the solar zenith angle is greater than 108¡Æ (astronomical night) are indicated by the dark blue fill. The aperture door will be shut for safety = when the sun is within 60¡Æ of the optical axis (light blue). The most desirable longitude for TEMPO is 100¡ÆW, which puts it in the U.S. Central Time zone = with Chicago and observing over eastern cities such as New York naturally would occur earlier and observing over western cities such as Los Angeles later.<= o:p>
We intend to dwell for about 10 seconds per spat=
ial
pixel, about four times longer than dwells for daylight collections, which
means that TEMPO will cover the field of regard about four times more slowl=
y as
well. It will take roughly two hours to cover the eastern half and two hour=
s to
cover the western half of the field of regard and we may collect the full f=
ield
of regard piecemeal over several nights. A dark signal calibration is plann=
ed
before beginning nightlights observing and again to verify the dark calibra=
tion
at the completion of observing. The
dark calibration will consist of 16 dwells of about 10 seconds each with the
aperture door shut that will then be averaged together to estimate the dark
signal. Because of hardware limitations on the maximum integration time for=
a
frame, we will co-add two frames each integrated for five seconds to form o=
ur
dwells. A solar calibration is performed daily, regardless of whether we are
conducting nightlights observations, with one of the diffuser plates in pla=
ce
to attenuate and diffuse incoming sunlight. This activity is performed when=
the
sun is 30¡Æ from the optical axis.
Since 30¡Æ is within our exclusion zone, there is no conflict betwee=
n the
two calibration activities or nightlights observations. 4. Nightlights Retrievals 4.1. Spectr=
al
Library The fine spectral resolution offered by TEMPO
enables light from various types of known sources to be identified by
origin. The NOAA National
Geophysical Data Center (NGDC) has made publicly available many examples of
spectra for various categories of sources [9-10]. These spectra form the
initial basis set for our retrieval algorithm that assigns the light from e=
ach
spatial pixel to a linear combination of sources in accordance with the
strength of each source. Fig. 3 shows the spectra of ten different lighti=
ng
types to be used as a basis set for spectral fitting in our retrieval analy=
sis.
Each spectrum, Fig.
3. A spectral library from NOAA National Geophysical Data Center (NGDC) wil=
l be
used to retrieve the strengths of contributing sources. Each basis function=
in
the library is normalized so that it would produce a signal in the VIIRS-DN=
B of
1 nW sr-1 cm -2. The spectral passband of the VIIRS-DNB is indicated by the
dashed box. The passbands for each TEMPO CCD are similarly indicated, but
unlike for VIIRS, each is subdivided into ~1000 spectral planes. The data
plotted from NGDC has not been convolved with the spectral line response
function of the TEMPO spectrometer and has been sampled at in 1 nm steps, w=
hich
is coarser than the spectral resolution of TEMPO. The
gap in the spectral coverage of TEMPO is not a significant handicap relativ=
e to
the sources cataloged by NGDC as most spectral features are covered by one =
of
the two TEMPO bands. 4.2. Retrie=
val
Sensitivity The
nightlights signal level for a voxel is the number of photoelectrons
accumulated during the dwell minus the calibrated dark signal photoelectron=
s.
We model the nightlights signal to be a linear combination of sources in our
spectral library. Let
The
matrix
The
dominant noise contribution comes from Poisson noise in each co-added frame=
(<=
![if !msEquation]>
Using
our measurements from the flight parts, we can calculate the retrieval
sensitivity for each type of lighting.&nbs=
p;
The square-roots of the diagonal components of the covariance matrix=
in
Equation (1) are these sensitivities (i.e., the one-sigma error on zero
signal). Note that since the spectral library basis functions are normalize=
d to
a unit response in the VIIRS-DNB, the uncertainties are calculated in DNB
units. These sensitivities are
plotted in Fig. 4 for retrievals using TEMPO alone. Our spectrometer is very
good at detecting lighting types with very sharp spectral features, but less
sensitive to thermal sources, such as incandescent lamps and not as adept at
discriminating between subtypes of such sources. For Fig. 4, we have aggreg=
ated
all thermal sources into a single type for retrieving the strengths of the
various lighting types. VIIRS-DNB, on the other hand, has a broad spectral
bandwidth and is quite good at detecting lighting of all types, but incapab=
le
of discrimination between types. By fusing data from TEMPO and VIIRS-DNB, we
can improve our retrieval sensitivity, especially for thermal sources. We do
this by remapping a VIIRS-DNB image or composite image into the TEMPO geome=
try,
binning it into the TEMPO spatial resolution, and then considering the rema=
pped
and binned VIIRS-DNB pixels to be samples of an additional spectral plane in
our retrieval. The resulting covariance for the retrieved strengths is
Here,
<=
![if !msEquation]>
Fig.
4. Retrieval sensitivities are calculated for seven classes of sources in u=
nits
of equivalent response in the VIIRS-DNB (1 DNB is equivalent to a radiance =
of 1
nW sr-1 cm -2 in the DNB passband). Each sensitivity is the minimum detecta=
ble
signal of that type as defined by a signal-to-noise ratio of one.
4.3. Backgr=
ound
Sources of Light
The prior analysis looked at the problem of
retrieval sensitivity only in the presence of instrumental noise. Natural sources of illumination wi=
ll
exist that may impact the retrievals of light from anthropogenic sources and
are interesting on their own, including backgrounds due to moonlight,
starlight, and zodiacal light, lightning, auroras, and airglow. Normally, w=
e do
not consider these backgrounds in our daytime retrievals. They will induce a retrieval signa=
l to
the extent that their spectra are correlated with the spectra in our source
library. We know from VIIRS that such backgrounds can be observed [12]. Moo=
nlit
scenes should present the brightest backgrounds and can have responses in t=
he
VIIRS-DNB >10 DNB units.
Moonlight exhibits a nearly continuum spectrum [13]. Aurorae and air=
glow
will be the next most important backgrounds and will include sharp spectral
features within the two TEMPO passbands, including atomic lines O (555.7, 6=
30,
636 nm). Na (D lines: 589.0, 589.6 nm) and the excited radical OH* (>500
nm). Aurorae are confined to =
the
polar regions and should be visible to TEMPO at the northernmost reaches of=
its
field of regard. Airglow is m=
ore
ubiquitous; VIIRS detects atmospheric waves in the upper atmosphere as airg=
low
emission features and can observe reflected airglow from the tops of clouds
[14]. Stray light from the sun even if it is >60¡Æ from the optical axis=
when
we do our nighttime observing could also produce a background to interfere =
with
retrievals.
We have not yet fully analyzed the impact of the=
se
dim background signals on retrievals. The Na D lines in nighttime airglow w=
ill
correlate well with the spectra of Na-vapor lamps and may cause a spatially
smooth retrieved Na lamp signal to be retrieved. We may need to include int=
erfering
sources in our spectral library or otherwise model or calibrate their
contributions. When spectral fitting residuals are found to exhibit structu=
re,
additional modeling will be required.
4.4
Distribution of Nightlights
We can envision the value of the TEMPO retrieval=
s by
looking at a VIIRS-DNB clear-sky composite remapped and binned to the TEMPO
geometry and resolution. This is shown in Fig. 5, where radiance levels are
color coded in levels relevant to the sensitivities shown in Fig. 4. There =
are
many areas where the radiance levels exceed 10 DNB units and typical values=
in
large cities are 100 DNB units or more. The signature of the aurorae is evi=
dent
in the northernmost part of the field of regard.
Fig.
5. VIIRS-DNB monthly clear-sky radiances have been remapped and binned to t=
he
TEMPO resolution and scan geometry (with TEMPO at 100¡ÆW longitude) to repr=
esent
the TEMPO field of regard. High signal-to-noise retrievals will certainly be
possible over major metropolitan areas.&nb=
sp;
Note the evidence of light from the aurorae in this monthly composit=
e.
5. Opportunities for
International Collaboration
TEMPO will be part of an international geostatio=
nary
constellation of spectroscopic instruments primarily for the study of air
quality [15]. The Korean Geostationary Environment Monitoring Spectrometer
(GEMS) [16] set to launch ~2019 and the European Sentinel-4 [17-18] set to
launch ~2022 will cover Asia and Europe respectively from Geostationary Orb=
it
(GEO). In addition, similar air quality missions Sentinel-5p [19] and -5 [2=
0]
will cover the Earth from Low-Earth Orbit (LEO) in the 2017 and ~2021
timeframes respectively, providing global coverage but with only one
measurement daily per site. GEMS and Sentinel-4 in principle could be opera=
ted
in a fashion similar to TEMPO at night to broaden the geographic scope of
nightlights spectroscopy. However, their spectral coverage is less favorable
than that of TEMPO for this purpose. GEMS only covers the wavelengths 300-5=
00
nm and Sentinel-4 covers 305-500 nm and 750-775 nm; both GEMS and Sentinel-4
miss the important spectral structures seen in Fig. 3 between 500-700 nm. A
spectrometer in LEO cannot generally dwell as long per pixel as an instrume=
nt
in GEO, which is a disadvantage for collecting data for nightlights retriev=
als.
Sentinel-5p and -5, moreover, also lack spectral coverage in the 500-675 nm
region but do cover 270-495 nm and 270-500 nm respectively. On the other ha=
nd,
Sentinel-4, -5p, and -5 will all have colder focal planes than TEMPO and GE=
MS
and much lower detector dark currents. Recalling that Poisson noise due to =
dark
current is the dominant noise source for TEMPO nightlights observations, the
reduced dark current could be an important mitigating factor to improve the
sensitivity for at least those types of lighting radiating in the region fr=
om
~300-500nm. Other LEO hyperspectral imagers like the Environmental Mapping =
and
Analysis Program (EnMAP, to be launched in 2017, 420-2450 nm, Guanter et al=
.,
2015) and proposed Hyperspectral InfraRed Imager (HyspIRI, 380 nm-2500 nm, =
Lee
et al., 2015) cover most of the spectral range (~400-900 nm) needed to reso=
lve
lighting types shown in Fig. 3 at high spatial resolution (tens to hundreds=
of
meters) but with coarser spectral resolution of ~10 nm, and thus can also p=
rovide
some capabilities to measure different lighting types. In summary, the
capabilities of each spectroscopic instrument must be examined individually
considering their dwell time per spatial pixel, spectral coverage, and dete=
ctor
dark current. An instrument i=
n GEO
will generally permit longer dwell times per spatial pixel but of course not
generally possess the finer spatial resolution of an equivalent LEO instrum=
ent.
6. Conclusion
We have shown that TEMPO will provide a powerful=
new
tool to complement VIIRS for the study of nightlights. TEMPO spectral retrievals will map=
the
utilization of artificial lighting by type over North America with good
sensitivity. Including VIIRS-DNB data in the retrievals further improves
retrieval accuracy for sources lacking sharp spectral features. Other air
quality instruments, such as GEMS, Sentinel-4, -5p, and -5 may offer some
possibilities for the study of nightlights over other geographic areas but
their spectral coverage is apparently less well suited than that of TEMPO to
this particular purpose. Nevertheless, we invite our international colleagu=
es
to examine these possibilities.
References
[1]
Zoogman, P., Liu, X., Suleiman, R. M., Pennington, W. F., Flittner, D. E.,
Al-Saadi, J. A., Hilton, B. B., Nicks, D. K., Newchurch, M. J., Carr, J. L.,
Janz, S. J., Andraschko, M. R., Arola, A., Baker, B. D., Canova, B. P., Chan
Miller, C., Cohen, R. C., Davis, J. E., Dussault, M. E., Edwards, D. P.,
Fishman, J., González Abad, G., Grutter, M., Herman, J. R., Houck, J=
., Jacob, D.
J., Joiner, J., Kerridge, B. J., Kim, J., Krotkov, N. A., Lamsal, L., Li, C=
.,
Lindfors, A., Martin, R. V., McElroy, C. T., McLinden, C., Natraj, V., Neil=
, D.
O., Nowlan, C. R., O¡¯Sullivan, E. J., Palmer, P. I., Pierce, R. B., Pippin=
, M.
R., Saiz-Lopez, Spurr, R. J. D., Szykman, J. J., Torres, O., Veefkind, J. P=
.,
Veihelmanna, B., Wang, H., Wang, J., Wulamu, A., Chance, K. (2017).
Tropospheric emissions: monitoring of pollution (TEMPO). Journal of
Quantitative Spectroscopy and Radiative Transfer, 186, 17-39.
[2]
Kyba, C., Garz, S., Kuechly, H., de Miguel, A. S., Zamorano, J., Fischer, J=
.,
& Hölker, F. (2014). High-resolution imagery of Earth at night: new
sources, opportunities and challenges. Remote sensing, 7(1), 1-23.
[3]
Elvidge, C. D., Baugh, K. E., Zhizhin, M., & Hsu, F. C. (2013). Why VII=
RS
data are superior to DMSP for mapping nighttime lights. Proceedings of the
Asia-Pacific Advanced Network, 35, 62-69.
[4]
Huang, Q., Yang, X., Gao, B., Yang, Y., & Zhao, Y. (2014). Application =
of
DMSP/OLS nighttime light images: A meta-analysis and a systematic literature
review. Remote Sensing, 6(8), 6844-6866.
[5]
Falchi, F., Cinzano, P., Duriscoe, D., Kyba, C. C., Elvidge, C. D., Baugh, =
K.,
... & Furgoni, R. (2016). The new world atlas of artificial night sky
brightness. Science advances, 2(6), e1600377.
[6]
Cinzano, P., Falchi, F., & Elvidge, C. D. (2001). The first world atlas=
of
the artificial night sky brightness. Monthly Notices of the Royal Astronomi=
cal
Society, 328(3), 689-707.
[7]
American Medical Association House of Delegates Annual Meeting 2012. Counci=
l on
Science and Public Health Report 4, Light pollution:
adverse health effects of nighttime lighting. Chicago, Illinois.
[8] Sánchez =
de Miguel,
A., Castaño, J. G., Zamorano, J., Pascual, S., Ángeles, M., C=
ayuela, L.,
Martinez, G. M., Challupner, P., Kyba, C. C. M. (2014). Atlas of astronaut
photos of earth at night. Astronomy and Geophysics, 55(4) ,4.36.
[9] Laboratory spec=
tra.
NOAA: National Center for Environmental Information. Retrievedhttp://[10] Elvidge, C. D.,
Keith, D. M., Tuttle, B. T., & Baugh, K. E. (2010). Spectral identifica=
tion
of lighting type and character. Sensors, 10(4), 3961-3988.
[11] Liao, L. B.,
Weiss, S., Mills, S., & Hauss, B. (2013). Suomi NPP VIIRS day
[12] Miller, S. D.,
Mills, S. P., Elvidge, C. D., Lindsey, D. T., Lee, T. F., & Hawkins, J.=
D.
(2012). Suomi satellite brings to light a unique frontier of nighttime
environmental sensing capabilities. Proceedings of the National Academy of
Sciences, 109(39), 15706-15711.
[13] Kieffer, H. H.,
& Stone, T. C. (2005). The spectral irradiance of the Moon. The
Astronomical Journal, 129(6), 2887.
[14] Miller, S. D.,
Straka, W. C., Yue, J., Smith, S. M., Alexander, M. J., Hoffmann, L., ... &=
amp;
Partain, P. T. (2015). Upper atmospheric gravity wave details revealed in
nightglow satellite imagery. Proceedings of the National Academy of Science=
s,
112(49), E6728-E6735.
[15] CEOS Atmospher= ic Composition Constellation (ACC). (2011). A Geostationary Satellite Constellation for Observing Global Air Quality: An International Path Forwa= rd. Retrievedhttp://ceos.org/document_manageme= nt/Virtual_Constellations/ACC/Documents/ACC_White-Paper-A-Geostationary-Sat= ellite-Cx-for-Observing-Global-AQ-v4_Apr2011.pdf[16] Kim, J. (2012, April). GEMS (Geostationary Environment Monitoring Spectrometer) onboard the GeoKOMPSAT to monitor air quality in high temporal and spatial resolution <= o:p>
[17] Ahlers, B.,
Courrèges-Lacoste, G. B., Guldimann, B., Veihelmann, B., Stark, H., =
&
Dobber, M. GMES Sentinel-4/UVN instrument concept and calibration approach.
(2011). Proceeding of 20th CALCON Technical Conference, Utah, USA.
[18] Courrèg=
es-Lacoste,
G. B., Ahlers, B., Guldimann, B., Short, A., Veihelmann, B., & Stark, H.
The Sentinel-4/UVN instrument on-board MTG-S. (2011). Proceedings of 2011
EUMETSAT Meteorological Satellite Conference, Oslo, Norway.
[19] Veefkind, J. P=
.,
Aben, I., McMullan, K., Förster, H., De Vries, J., Otter, G., ... &=
; Van
Weele, M. (2012). TROPOMI on the ESA Sentinel-5 Precursor: A GMES mission f=
or
global observations of the atmospheric composition for climate, air quality=
and
ozone layer applications. Remote Sensing of Environment, 120, 70-83.
[20] Sierk, B., Bez=
y,
J. L., Caron, J., Meynard, R., Veihelmann, B., & Ingmann, P. (2012,
October). The GMES Sentinel-5 mission for operational atmospheric monitorin=
g:
Status and Developments. In Proceedings of the ICSO (International Conferen=
ce
on Space Optics), Ajaccio, Corse, France, Oct (pp. 9-12).
[21] Guanter, L., Kaufmann, H., Segl, K., Foerster, =
S.,
Rogass, C., Chabrillat, S., Kuester, T., Hollstein, A., Rossner, G., Chlebe=
k,
C., Straif, C., Fischer, S., Schrader, S., Storch, T., Heiden, U., Mueller,=
A.,
Bachmann, M., Mühle, H., Müller, R., Habermeyer, M., Ohndorf, A.,=
Hill, J.,
Buddenbaum, H., Hostert, P., van der Linden, S., Leitão, P. J., Rabe=
, A.,
Doerffer, R., Krasemann, H., Xi, H., Mauser, W., Hank, T., Locherer, M., Ra=
st,
M., Staenz, K., Sang, B. (2015). The EnMAP spaceborne imaging
spectroscopy mission for earth observation, Remote Sensing. 7 (7), 8830-885=
7.
[22] Lee, C. M., Ca=
ble,
M. L., Hook, S. J., Green, R. O., Ustin, S. L., Mandl, D. J., & Middlet=
on,
E. M. (2015). An introduction to the NASA Hyperspectral InfraRed Imager
(HyspIRI) mission and preparatory activities. Remote Sensing of Environment=
, 167,
6-19.
J. Carr et al. /International Journal of Sustainable
Lighting 20 (2017) 26-35