International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
ESTIMATION OF NIGHT LIGHT FROM THE DMSP/OLS
Hiroshi Yagia,b*, Hesiletub, Masanao Haraa, Fumihiko Nishiob
a
VisionTech Inc., 2-1-16 Umezono , Tsukuba-city, Ibaraki 305-0045, Japan, yagi@vti.co.jp
Centre for Environmental Remote Sensing (CEReS), Chiba University,1-33 Yayoi-cho, Inage-ku, Chiba 263-8522, Japan
b
Commission VIII, WG VIII/8
KEY WORDS: DMSP, OLS, Night right, Noise Reduction Filter (NRF), Stable light
ABSTRACT:
Night light on land includes the temporal light which occurs by forests and fields fire, city fire and a volcanic explosion. Ho wever,
the most part of the night light based on human activity, and this light can say a symbol of energy consumption. To observe night
light can be said the barometer of this human activity quantitatively, and it can also be one way to measure spatial distribution of
energy consumption handily. For these possibilities there is a satellite, DMSP (Defense Meteorological Satellite Program) observing
night light. The OLS (Operational Linescan System) sensor onboard DMSP mission observes the night light periodically and
globally. However, , there are few issues to be resolved in the observation data, such as the moon surface reflection of sunlight,
secondary reflection from surface of the sea in influence of sunlight and an age of the moon by a change in solar altitude by night
observations of OLS. The sensor gain of OLS is adjusted to minimize the influence. Therefore, analyzing quantitatively using the
observed intensity of light, a method to revise the gain adjustment has to derive. But the information concerned with this gain
adjustment isn't exhibited. So the object of the research is to take gain tuning parameter by these sunlight and surface of the moon
light and to develop a correction way to the data observed by the fixed gain.
1. INTRODUCTION
The light coming from the bio-mass fire and volcanic
eruptions generate temporarily but one part of the lights at night.
However, the bigger part of the night light is coming from the
artificial origin or sources. Therefore, this part of night light can
also be said to be a symbol of energy consumption. In other
words, it’s a barometer of human activities which can be
observed quantitatively, providing one method for measuring
spatial distribution of energy consumption on the earth surface.
Now, the U.S. polar orbit circumference type meteorological
satellite DMSP (Defense Meteorological Satellite Program) is
operative and observing light at night. The sensor called OLS
(Operational Linescan System) is carried in DMSP, and
observation is performed periodically and globally. However,
these nights light observations by OLS is influenced by the
changes in solar altitudes and the impact of the secondary
reflection from sea surface. In order to minimize this influence,
adjustment of the sensor gain of OLS is performed. Therefore,
when conducting quantities analysis using the observed
intensity of light, you have to rectify a part for gain adjustment.
However, the information in connection with this gain
adjustment is not well published. In this research, the amount of
gain adjustments by this sunlight and moon's surface catoptrics
light was removed, and focused on developing a correction
method to the data observed by the fixed gain.
Fig1. Analysed Area
㪛㪤㪪㪧㩿㪛㪼㪽㪼㫅㫊㪼㩷㪤㪼㫋㪼㫆㫉㫆㫃㫆㪾㫀㪺㪸㫃㩷㪪㪸㫋㪼㫃㫃㫀㫋㪼㩷㪧㫉㫆㪾㫉㪸㫄㪀㩷㫊㫇㪼㪺㫀㪽㫀㪺㪸㫋㫀㫆㫅
㪦㫉㪹㫀㫋
㪘㫃㫋㫀㫋㫌㪻㪼
㪪㫌㫅㩷㫊㫐㫅㪺㪿㫉㫆㫅㫆㫌㫊㩷㫅㪼㪸㫉㩷㫇㫆㫃㪸㫉㩷㫆㫉㪹㫀㫋㩷
㪏㪊㪇㫂㫄
㪩㪼㫍㫀㫊㫀㫋
㪈㪇㪈㫄㫀㫅
㪪㪼㫅㫊㫆㫉
㪦㪣㪪䋨㪦㫇㪼㫉㪸㫋㫀㫆㫅㪸㫃㩷㪣㫀㫅㪼㫊㪺㪸㫅㩷㪪㫐㫊㫋㪼㫄䋩㩷
㪪㪪㪤㪆㪠䋨㪤㫀㪺㫉㫆㫎㪸㫍㪼㩷㪠㫄㪸㪾㪼㫉䋩㩷
㪪㪪㪤㪆㪫䋨㪘㫋㫄㫆㫊㫇㪿㪼㫉㫀㪺㩷㪫㪼㫄㫇㪼㫉㪸㫋㫌㫉㪼㩷㪧㫉㫆㪽㫀㫃㪼㫉䋩㩷
㪪㪪㪤㪫㪆㪉䋨㪘㫋㫄㫆㫊㫇㪿㪼㫉㫀㪺㩷㪮㪸㫋㪼㫉㩷㪭㪸㫇㫆㫉㩷㪧㫉㫆㪽㫀㫃㪼㫉䋩㩷
㪪㪪㪡㪆㪋䋨㪧㫉㪼㪺㫀㫇㫀㫋㪸㫋㫀㫅㪾㩷㪜㫃㪼㪺㫋㫉㫆㫅㩷㪸㫅㪻㩷㪠㫆㫅㩷㪪㫇㪼㪺㫋㫉㫆㫄㪼㫋㪼㫉䋩
㪪㪪㪠㪜㪪䋨㪠㫆㫅㩷㪪㪺㫀㫅㫋㫀㫃㫃㪸㫋㫀㫆㫅㩷㪤㫆㫅㫀㫋㫆㫉䋩㩷
2. DATA SOURCES
The 1999 data observed in the visible band (night mode) of
DMSP/OLS was used in the study. The geometric correction of
the data has done by U.S. NGDC, and spatial resolution has set
to 1km. Moreover, the region made applicable to analysis falls
from 124 degrees to 146 degrees east longitudes and 24 north
degrees to 46 north degrees latitude as presents by the Fig. 1.
DMSP and specifications of OLS which is the loading sensor
are shown in Fig. 2.
㪦㪣㪪㩷㪪㪼㫅㫊㫆㫉㩷㫊㫇㪼㪺㫀㪽㫀㪺㪸㫋㫀㫆㫅㩷
㪙㪸㫅㪻
㪪㫇㪼㪺㫋㫉㪸㫃㩷㫉㪸㫅㪾㪼
㪭㫀㫊㫀㪹㫃㪼䋨㪛㪸㫐䋩
㪭㫀㫊㫀㪹㫃㪼䋨㪥㫀㪾㪿㫋䋩
㪫㪿㪼㫉㫄㪸㫃㪄㪠㪩
㪇㪅㪋㪇䋭㪈㪅㪈㪇㱘㫄
㪇㪅㪋㪎䋭㪇㪅㪐㪌㱘㫄
㪈㪇㪅㪇䋭㪈㪊㪅㪋㱘㫄
㪪㫇㪸㫋㫀㪸㫃㩷㫉㪼㫊㫆㫃㫌㫋㫀㫆㫅㩷
㪽㫀㫅㪼㩷
㫊㫄㫆㫆㫋㪿㩷
㪇㪅㪌㪌㫂㫄㩷
㪉㪅㪎㫂㫄㩷
㪇㪅㪌㪌㫂㫄㩷
㪉㪅㪎㫂㫄㩷
㪇㪅㪌㪌㫂㫄㩷
㪉㪅㪎㫂㫄㩷
㪪㫎㪸㫋㪿
㫎㫀㪻㫋㪿
㪊㪇㪇㪇㫂㫄
㪊㪇㪇㪇㫂㫄
㪊㪇㪇㪇㫂㫄
㪩㪸㪻㫀㫆㫄㪼㫋㫉㫀㪺
㫉㪼㫊㫆㫃㫌㫋㫀㫆㫅
㪍㪹㫀㫋
㪍㪹㫀㫋
㪏㪹㫀㫋
㪥㫆㫋㪼㩷㪑㩷㪈㩷㫂㫄㩷㫊㫇㪸㫋㫀㪸㫃㩷㫉㪼㫊㫆㫃㫌㫋㫀㫆㫅㩷㪻㪸㫋㪸㪃㩷㫎㪿㫀㪺㪿㩷㫀㫊㩷㫉㪼㫊㪸㫄㫇㫃㪼㪻㩷㪽㫉㫆㫄㩷㫋㪿㪼㩷㪽㫀㫅㪼㩷㫄㫆㪻㪼㩷㫉㪼㫊㫆㫃㫌㫋㫀㫆㫅㪃
㫀㫊㩷㪻㫀㫊㫋㫉㫀㪹㫌㫋㪼㪻㩷㪹㫐㩷㪥㪞㪛㪚㩷㫀㫅㩷㪬㪅㪪㪅㪘㪅㩷
Fig.2 DMSP specification
831
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
3. METHOD
altitude of the influence is performed. Here, it was possible that
gain adjustment gap is obtained as primary frequency by
reprocessing NRF data set of obtained fcomp, and the
performed processing was.
DMSP goes the direction of a pole of the earth around in
101 minutes, and whenever it completes one earth round, a shift
of 25.25 degrees in the direction of latitude, is observed due to
the global rotation. Therefore, DMSP passes to the same point
in about 23.5 days (returning cycle). It means that about 23
degrees of earths had rotated on the circumference (revolution)
orbit of the earth [thing / on the 23.5th / to shift] cantering on
the sun, and a solar altitude changes. By neither the gain of a
sensor, nor change of the solar altitude according that the offset
is constant to global rotation and revolution in this way, the
exaggerated scale (saturation) of the dynamic range of a sensor
cannot be carried out, or it becomes a under scale and sufficient
observational data can be obtained. Then, the sun with
periodicity; assuming that adjustment of offset was considered
as the gain which can see by carrying out cancellation out of the
advanced influence, and considering it as a picture, the intensity
of light of arbitrary specific subjects performed the
compensation which is not based on change of time (season)
but serves as the fixed intensity of light (luminosity).
The cloud-less data set of 36 scenes was created for visible
data with the time maximum synthetic method (10-dayMVC) in
the DMSP/OLS night for one year in 1999, and NRF(Noise
Reduction Filter : one formula) *1 was processed to the data set.
The frequency component which the arbitrary time series pixels
in a picture have, its amplitude, and a phase are obtained by
performing NRF processing (two formulas). It reconstructs as a
10-dayNRF data set by carrying out inverse transform of this
NRF processing result and carrying out imaging. Supposing the
source of luminescence of a terrestrial subject is always
constant here, it should not come together at Time t, but
C0(Fig.3) should become fixed. However, it is *2,3 which does
not become constant [C0] by change of a solar altitude with
periodicity. Furthermore, the gain adjustment depending on a
solar altitude does not perform gain adjustment which the
intensity of light of the arbitrary subjects which have emitted
the fixed intensity of light can always observe uniformly, but is
considered to be adjustment in the range which can do it. It is
possible that the solar altitude contained in the amplitude
intensity of the sin ingredient of C0 and two formula from this
and the amount of gain adjustments depending on it have
influenced change of C0. Then, the influence and its amount of
gain adjustments of a solar altitude will be removed about
reconstructed 10-dayNRF by having subtracted a sin ingredient
and its amplitude intensity from ft of every time t. However, ft
obtained in this way always is not obtained as a fixed DN value
in the intensity of light of arbitrary specific subjects.
It is possible that the calibration based on a specific
reference is not performed as the cause. Fig. 4 -- the East China
Sea -- the charity under operation -- it being what was plotted
on the picture observed by OLS, and the operation position
which carried out tracking of the fishing fire (the total intensityof-light 250kw) of a benevolence ships (medium size cuttlefish
fishing boat) during January [ about ], and was notified by
vessel fax DN value of the position, i.e., charity, -- a ship
releases -- shining (Fig. 5) -- it is Fig. 6 which read and was
expressed with graph *4. It turns out that the source of
luminescence of the total intensity of light of 250kw is not
observed as a fixed DN value which clearly displays in graph.
Therefore, the gap of the amount of gain adjustments and a
reference value and the phase difference of the change cycle of
a solar altitude may remain in fcomp which subtracted the
amplitude intensity of the sin ingredient in Time t from ft. Gain
adjustment is considered to be has lower influence of the phase
difference if gain adjustment depending on the large solar
ft
N
­
§ 2Skt ·
§ 2Skt ·½
c0 c1t ¦ ®c2l 1 sin¨
t ¸ c2l cos¨
t ¸¾
© M ¹
© M ¹¿
l 1 ¯
࡮࡮(1)
Where M = Cycle (1 Year =36 for 10-day MVC)
t = Time period ( 1Year = 36, 2Years = 72,
3Year=108, 4Years=144)
kt
= Frequency (1,2,3,4,6,12 month(s))
N = Number of pixels
c0 , c1 , c2l , c2l 1 = Coefficient of
ft
each function
N
­
§ 2Skt
·½
c0 c1t ¦ ®al sin ¨
t ǰl ¸¾ © M
¹¿
l 1 ¯
Where al
ǰl
c 2 2l c 2 2l 1 tan 1
c2 l
c2l 1
࡮࡮(2)
࡮࡮amplitude at
࡮࡮amplitude at
kl
kl
Fig.3 NRF C0 data
䂾㩷 Aug.20䋨first point䋩
䃂㩷 Sep.07䋨last point䋩
Shinseimau tracking line
Kawakamaru tracking line
Japan-China fisheries agreement line
Japan-Korea fisheries agreement line
Fig.4 Tracking of benevolence ships
832
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
No.28 Kawakamaru
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪡㪸㫅
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪝㪼㪹
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪤㪸㫉㩷 㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪘㫇㫉
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪤㪸㫐
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪡㫌㫅
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪡㫌㫃㩷
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪪㪼㫇
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪦㪺㫋
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪥㫆㫍㩷 㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪛㪼㪺
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪘㫌㪾
No.27 Shinseimau
Fig.7 NRF data set for one year in 1999
Fig.5 Lights of benevolence ships (250kw)
㪍㪇
㪛㫀㪾㫀㫋㪸㫃㩷㪥㫌㫄㪹㪼㫉
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪡㪸㫅
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪝㪼㪹
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪤㪸㫉㩷 㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪘㫇㫉
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪤㪸㫐
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪡㫌㫅
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪡㫌㫃㩷
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪪㪼㫇
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪦㪺㫋
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪥㫆㫍㩷 㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪛㪼㪺
㪢㪸㫎㪸㫂㪸㫄㪸㫉㫌
㪌㪇
㪪㪿㫀㫅㫊㪼㫀㪤㪸㫉㫌
㪋㪇
㪊㪇
㪉㪇
㪈㩷㫊㪼㪸㫊㫆㫅 㫀㫅㩷㪘㫌㪾
㪈㪇
㪇
㪏㪆㪉㪇 㪏㪆㪉㪉 㪏㪆㪉㪋 㪏㪆㪉㪍 㪏㪆㪉㪏 㪏㪆㪊㪇 㪐㪆㪈
㪦㪹㫊㪼㫉㫍㪸㫋㫀㫆㫅㩷㪛㪸㫋㪼
㪐㪆㪊
㪐㪆㪌
㪐㪆㪎
Fig.6 Changing DN value of benevolence ships
Fig.8 NRF frequency data set for one year in 1999
4. RESULTS AND DISCUSSSIONS
Fig.7 shows the NRF data set for one year in 1999. Fig. 8
shows the amplitude (sin ingredient) picture of primary
frequency ingredients by season units from the results of
performed NRF processing, using about a 10-dayMVC data set.
Moreover, Fig. 9 shows the condition when reduced the
amplitude of primary frequency ingredients corresponding to ft
for every season unit about the result of the NRF processing
expressed with one formula. Here, supposing the influence and
its amount of gain adjustments of a solar altitude are removed
by reducing this amplitude intensity, the intensity of light (DN
value) of arbitrary specific points will show a fixed value. Then
DN value of the season unit in arbitrary selected two or more
points change of year the time a series have plotted (Fig. 10,
Fig.11 and Fig.12). Dashed line graph shows change of DN
value before compensation, and a solid line shows change of
DN value after compensation. When DN value compensation
before compared with the after compensation conditions, it
turns out that the influence of a solar altitude or its gain
adjustment is decreasing clearly. However, in order to fulfil the
conditions of "always having fixed DN value", results are
agreed very much.
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪡㪸㫅㩷
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪝㪼㪹
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪤㪸㫉㩷 㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪘㫇㫉㩷
㪈 㫊㪼㪼㪸㫊㫊㫆㫅㫅 㫀㫅 㪤㪸㪸㫐
㪈 㫊㪼㪼㪸㫊㫊㫆㫅㫅 㫀㫅 㪡㫌
㫌㫅
㪈 㫊㪼㪼㪸㫊㫊㫆㫅㫅 㫀㫅 㪡㫌
㫌㫃
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪪㪼㫇
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪦㪺㫋
㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪥㫆㫍㩷 㪈㩷㫊㪼㪸㫊㫆㫅㩷㫀㫅㩷㪛㪼㪺
㪈 㫊㪼㪼㪸㫊㫊㫆㫅㫅 㫀㫅 㪘㫌㫌㪾
Fig.9 NRF fixed data set for one year in 1999
833
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
its impact on compensation of this sensor characteristic can be
stated as future research interests.
㪈㪍㪇
㪈㪋㪇
REFERENCE
㪈㪉㪇
㪛㪥
㪈㪇㪇
1.
㪥㪩㪝㩷㫆㫉㫀㪾㫀㫅㪸㫃
㪥㪩㪝㩷㪽㫀㫏㪼㪻
㪏㪇
㪍㪇
㪋㪇
㪉㪇
2.
B
B
B
B
B
B
B
B
B
B
B
B
㪇
㪰㪼㪸㫉
Fig.10 NRF fixed data set
3.
㪈㪉㪇
㪈㪇㪇
㪛㪥
㪏㪇
㪥㪩㪝㩷㫆㫉㫀㪾㫀㫅㪸㫃
㪥㪩㪝㩷㪽㫀㫏㪼㪻
㪍㪇
4.
㪋㪇
㪉㪇
B
B
B
B
B
B
B
B
B
B
B
B
㪇
㪰㪼㪸㫉
Fig.11 NRF fixed data set
㪈㪉㪇
㪈㪇㪇
㪛㪥
㪏㪇
㪥㪩㪝㩷㫆㫉㫀㪾㫀㫅㪸㫃
㪥㪩㪝㩷㪽㫀㫏㪼㪻
㪍㪇
㪋㪇
㪉㪇
B
B
B
B
B
B
B
B
B
B
B
B
㪇
㪰㪼㪸㫉
Fig.12 NRF fixed data set
5. CONCLUSTION
It was found that a fixed effect had the influence of the sunlight
by change of a solar altitude by applying NRF processing in
this study. It's expected to estimate the gain adjusted value
which won't be published from now on.
However, when performing the monitor over a long period of
time etc., there is a point which should be further taken into
consideration in information extraction with high accuracy,
such as carrying out the compensation which fully took into
consideration the difference in the characteristic of the satellite
sensor. In order to use the past data as a series of readings and
834
Masanao Hara, Shuhei Okada, Hiroshi Yagi, Takashi
Moriyama, Koji Shigehara and Yasuhiro Sugimori,
Developing and evaluation of the noise reduction
filter for the time-series satellite images, Journal of
the Japan Society of Photogrammetry and Remote
Sensing, Vol.42, No.5, 48-59, 2003
Hesiletu, Masanao Hara, Shuhei Okada, Hiroshi Yagi,
Hironori Kotake, Kazuhiro Naoki and Fumihiko
Nishio, Extraction of stable light from DMSP/OLS
night-time image, Journal of Advanced Marine
Science and Technology Society, Vol.14, No.2,
pp.21-28, 2008
Masanao Hara, Shuhei Okada, Hiroshi Yagi, Takashi
Moriyama , Koji Shigehara, and Yasuhiro Sugimori,
Progress for Stable Artificial Lights Distribution
Extraction Accuracy and Estimation of Electric
Power Consumption by mean of DMSP/OLS
Nighttime Imagery, Remote Sensing and Earth
Sciences, 1(1), 31-42, 2004
Masanao Hara, Shuhei Okada, Masahiko Ichizuka,
Koji Shigehara, Takashi Moriyama and Yasuhiro
Sugimori, Monitoring of fishing lights intensity for
squid fishing vessels by means DMSP/OLS nightlight
imagery, Journal of Advanced Marine Science
Technology Society, Vol.9, No.25 2004