- Email: [email protected]
Hot spots and plumes: Observation by meteorological satellite
,irmorplnic Printed
,fmdr~nr
in Gnat
Vol. 21. No. 6. pp. 1421-1435.
ooo4-6981/87 s3.00+0.00 Perpmon Journals Ltd.
1987.
Britain.
HOT SPOTS AND PLUMES: OBSERVATION METEOROLOGICAL SATELLITE
BY
R. S. SCORER Imperial College, London, U.K. (First receiued 8 April 1986 and received jar publicotion 17 November 1986)
Abstract-Many hot spots can be detected by Channel 3 of the sun-synchronous meteorological satellites NOAA 6-9. This is because that channel is much more sensitive to radiation from bodies in the temperature range 500-1500 K than to scattered sunshine or earth-temperature radiation. Hot bodies occupying only a small fraction ofa pixel can be detected. The other channels are designed to receive much greater power from earth radiation (300 K) or scattered sunshine (6000 K). Plumes from chimneys are very difficult to detect because they are small and rapidly diluted and dispersed. Some flukes of illumination make them detectable occasionally; and some infrequent meteorological circumstances make them persistent and large. Ship trails are quite exceptional and very obvious, but they are very rare.
1. DETECTION OF HOT-SPOTS
red or IR filters. Cloud shadows are well displayed and it records the scene as seen by scattered sunshine.
1.1. Satellite receiving systems
satellites are designed to provide photographic images either by recording scattered sunshine (Channels 1 and 2) or by receiving infra-red emissions from clouds and objects at the earth’s surface (Chs 4 and 5). Figure 1 shows the classical relative black body emission intensity, B, f (according to Planck), as a function of AT (bottom scale), I being the wavelength in pm, and T the absolute temperature in K. The wavelength bands (channels) received are chosen to be in windows of the water vapour spectrum, and are assumed to be well represented by the Planck function, although variable pollution may interfere with the image provided by the satellite. The maximum intensity is given the value unity, and occurs where IT = 2900. This investigation is restricted to the satellites NOAA 6-9 of which there are very good records in the University of Dundee archive. The following are the channels used:Meteorological
1 2 3 4 5
0.55-0.68 0.725-1.10 3.55-3.93 10.5-11.5 11.5-12.5
pm pm pm pm pm
(yellow-orange) (very near IR) (near IR) (IR) (IR).
Ch5 is so similar to Ch4 for our purposes that it needs no separate discussion. Chl is useful in identifying haze which may not be as clear on Ch2, but it has no interest in this section on hot spots. Therefore we concentrate on the properties of Chs 2,3 and 4. Ch2 is commonly called ‘visible’ although it is actually all in the IR: but it produces pictures very similar to those obtained with ordinary film using deep
Ch3 also picks up scattered sunshine by day but the intensity is much less than in Ch2, and the scattering has directional characteristics and intensity variations in clouds of water droplets whose size range includes the wavelength band. Small droplets scatter more powerfully fowards and backwards than sideways, and in droplets larger than about 30 p and clouds composed of ice crystals (which have a much larger typical size than liquid droplet clouds) the absorption in Ch3 is so large that for practical purposes the scatter is reduced to zero and the clouds appear black when the scene is printed in photo-positive, which is called Ch3-. This is because the channel was designed to measure emitted radiation and night images were to be compared with Chs 4 and 5, and printed in photo-negative with the warmest objects black and the coldest white: at night, in this mode, the pictures have many similarities to images of Ch4. but also some essential differences. Ch4 images are printed in photo-negative, and provide a measure of temperature. The intensity of scattered sunshine in Ch3 is greater than that of the emitted radiation, and overwhelms it. At sunrise Ch3 images consist of scattered sunshine in the east, but emitted radiation showing the temperature distribution in the west where it is still dark. Sometimes where the scattered sunshine is almost zero or very uniform, as is often the case at sea, temperature variations are evident by day. 1.2. Intensity characteristics
of Chs 2,3 and 4
In Fig. 1 the positions of the central wavelengths of Chs 1.2 and 3 are shown on the upper side of the bottom scale for 6000 K representing sunshine. The position of Ch4 is shown on the scale marked 300 K
1427
1428
R. S. SCORER
Km
a3xl
3003
4cm
5030
6am XT
7aO
Qo30
rr,coo
Fig. 1.
(for earth emissions). Relevant positions are marked on the curve and on the scales for 400 K and 700 K, representing warm and hot spots. The intensity is set at unity at the maximum where T = 2900, and the curve indicates the relative intensity as a function either of wavelength for a given temperature or of temperature for a given wavelength. For the solar beam and for the earth’s emission the central wavelength of Ch3 (L = 3.75 pm) has a small relative intensity equal to 0.044 at T = 300 K and to 0.006 for T = 6000 K. As the scale for the position of this wavelength at different temperatures shows, the relative intensity is unity for T = 784 K. At the same time the temperature scale for Ch4 (A = 11 pm) shows that while the relative amplitude is unity for T = 264 K it has falkn to 0.2 at T = 700 K. The consequence is that for hot bodies Ch3 has a relative intensity which, at 500 K, is 15 times the 300 K value, and at 800 K is 23 times that value. By contrast, for Ch4 hot bodies have a relative intensity which is decreased from unity at 264 K to l/2 at 477 K and to l/S at 700 K. These factors are to be multiplied by 7” according to Stefan’s law, to obtain the relative magnitudes of the signals at different temperatures. Ch3 is therefore much more sensitive to temperatures of earthly hot spots than Ch4. Approximately the part of the curve representing B ,,&n the range 260-800 K may be represented by a sine curve, viz 13,75= PB,,,,,=
0.5+SinTi >
T4.
(1)
The rate of increase of the part of this function in the bracket is a maximum at T = 480 K.
It is found in practice that although the scattered sunshine completely obscures the emissions from the earth in Ch3 by day, hot spots of significantly higher temperature show through brightly, as shown in the next section. In the same way the part of the curve representing B,, for 1= 11 pm (Ch4) over the temperature range 300-800 K may he approximately represented by a hyperbola, and the resulting formula is I,, = B,,$-“
r&T’.
The formulas (1) and (2) are not valid outside the temperature ranges specified. 1.3. How small a hot spot can be detected? We imagine a hot spot of temperature T, to occupy a fraction a of a pixel, the temperature of the surrounding land being TO.The emitted radiation received at the satellite in wavelength 1 is proportional to II=
(l-a)BL,r,,G+aB~,r,fi.
(3)
To exemplify this we put ,I = 3.75 P. TO = 300 K, T, = 700 K, and obtain lo-* i3,75 = (1 -a) x0.044x
81 +a x 0.975 x 2401
= 3.56 + 2337a z 3.56 (1 + 66Oa).
(4)
It is clear from this expression that if a exceeds about 0.0015, I3,,5 is more than twice the value when a = 0. The hot spot would then be detected as a pixel with more than twice the brightness of the surroundings, at night.
1429
Hot spots and plumes With a pixel size of 1.1 x 1.1 km” the hot spot would be easily detected if it had an area of about 40 x 40 m* or more. To be detected by day it would be ~CCCS~~T for the emission to exceed the intensity of the scattered sunshine. This would be achieved by a value of e around 0.01 because the sunshine is only two or three times as bright as the ordinary infra-red emissions in Ch3. Thus a hot spot of 100* m* will be easily detectable at 700 K. By contrast, when the same calculation is performed for Ch2 we find, with 1= 0.875 pm 10-*&V,
=(l-a)x
10-20x81+ax
1O-4x24O1
(5) and nothing would be detected in the ‘visible’ at night because the emission at such a short wavelength is far too small, even though the radiation from the hot area is a relatively large fraction of the total. By day the scattered sunshine is completely dominant, being around lo5 times as powerful. The radiation received by Ch4 (A = 11 p) is, likewise, IO-sI 11 =(l-a)xO.965x8l+axO.2x2401 2: 78(1+ 5a)
(6)
and the relative importance of the hot spot is eight per cent of what it is in the case of Ch3. 1.4. Detection of James Flares at chemical works and on oil or gas rigs may have temperatures in the range lOOO-1500 K. They are incandescent because of the carbon particles present. There is radiation from the gases present, but since this is mostly from Hz0 and COz it is effectively absorbed by the atmosphere fairly close to the flame. The values of BArfor Chs 2,3 and 4 for a pixel with ground at 300 K and two flames occupying fractions a and Ji of the area and having temperatures HMOand 1500 K, respectively are given in Table 1. For a pixel in which the flame occupies a small fraction, a, of the area the intensities of the radiation received at the satellite in the three channels are
At 1000 K the flame is not detected by Ch2 because the signal is too small compared with scattered day&bt. Even at night the channel is not designed to respond to such a low signal, and it does not ‘see’ the illumination of cities, although it might just detect a flame at 1500 K. A Iarge flame at 1000 K might have an area of the order of 30 x 30m which, with a pixel of size 1.1 x 1.1 km2, gives a value of a of 9 x 10T4. In that case the flame would give the pixel an intensity about three times the environmental intensity in Ch3, and would be detected. A much smaller flame with dark surroundings, such as at night of at sea where there is no glint, co&d be detected at about one tenth of that size, say IO x 10 m*. A flame at 1500 K would certainly be detected by Ch3, being 2f times as bright as at lOO0K. Ch4 however, will not detect a Rame even of the larger size or higher temperature just mentioned, because the intensity would only add about lx, and this would not distinguish it from other local variations, 1.5. Large warm spots A warm spot occupying a whole pixel can obviously be detected by Ch4 which is designed to separate areas whose temperature differs from that surrounding it by 1 or 2°C. Thus lakes or rivers are easily detected on a cool night or day. Ch2 produces a very similar appearance by day with water surfaces which have no glint. By both channels lakes are depicted as dark spots, often occupying a few adjacent pixels. They could appear as light spots on a warm day when they are cooler than the ground by Ch4, or when they have sun glint by Ch2. A very warm spot, such as a volcano which is not erupting and whose top has a temperature of, say, 350 K over a whole pixel, would be much more prominent on Ch3 than on Ch4. The relative intensities of the surroundings and of the warm pixel are given in Table 2.
given by
x 81 +a x lOI3 x lo4 = 10a +/3x 0.13 x 5.06 x lo4 = 6.6 x lOJ/I Ch3: 10-813,ts = (1 -a)O.O44 x 81 +a x0.88x lo4 = 3.6+88Otla [email protected] 104=3.6+21500~ Ch4: lo-*1,,=(1-a)0.965x81 +axO.O7x10’=78+620a +/5x0.017x5.06x104=78$860~
Ch2:
10-810,s,S = (1 -a)1O-2o
Table I. The values of the intensity of the emission in channels 2, 3 and 4 relative to the maximum at various temperatures
channel a. Temperature E 1000 1500
;.*75
z.75
4 11
IO_‘0 lo-’ 10-3 0.13
0.044 0.975 0.86 0.425
0.965 0.2 0.07 0.017
for for for for for for
lOOOK 150OK lOOOK 1500 K lOOOK 1500 K.
Table 2. The relative intensities of the surroundings and of a warm pixel at the temperatures given, and their ratio in channels 3 and 4
Ch3 Ch4
Surroundings at 300°K
Warm spot at 350°K
0.044~81 0.965 x 8 1
0.15 x 150 0.84 x 150
Ch3 is clearly more sensitive.
Ratio &i,ol&oo 6.3 1.6
1430
R. S. SCORER
To be evident by day on Ch3 the intensity must exceed that of the surrounding scattered sunshine, which is around twice that of the emitted radiation. At 350 K the factor of 6.3 will give the warm spot greater prominence on Ch3 than it has in Ch4. (See Stromboli in Scorer, 1986a.)
2. DETECHON OF PLUMES
2.1. Plumes of individual chimneys To be detected by satellite a plume must extend over at least two pixels, and perhaps scarcely deserves the name of plume unless it extends for three or more. Most industrial plumes are less than 1 km in length and cannot therefore be detected by routine meteorological satellite. Indeed cases of identifiable plumes are very few. To be identified by Ch4 a plume must contain particles radiating as black bodies. This condition would be fulfilled by a plume in which water was condensed and the plume would look like low cloud. The temperature contrast with the surface background is unlikely to be large enough for clear identification except in exceptional circumstances of very cold ground. Plumes of condensed water such as from a cement works or a power station burning natural gas usually evaporate fairly quickly and are seldom more than a few chimney heights in length in the absence of natural cloud. Some plumes, thought to contain SO,,
*ML 3
are much longer because of the hygroscopic condensation nuclei. Neither Ch2 nor Ch3 is likely to make a plume from a single chimney visible except at sunrise or sunset when the plume is illuminated against a much darker ground which may be largely in shadow. Ch3 is particularly good in this respect when viewing a cloud of condensed water towards the sunrise because the forward scattering by small droplets is relatively intense. (See Scorer, 1986b, 8.9 and 3.15.) 2.2. Plumes from industrial complexes These can sometimes be detected and the best viewing conditions are provided by Ch 1 when looking towards the sun, because forward scattering is the most intense in the case of haze, and Chl has a relatively much smaller signal from typical ground than Chs 2 and 3 in sunshine. Chs 1-4 of the CZCS satellite Nimbus 7 are sensitive to haze, but because of the different viewing angle are relatively more sensitive to tenuous cirrus, and are unreliable for routine meteorological purposes. 2.3. Plumes at sea Ship trails are an exception to the general rule that pollution trails are so quickly diluted that they become invisible by satellite before they achieve a length of a few pixel widths. Plumes of this kind occur only at sea, and there only rarely because they depend on the air being exceptionally clean (Scorer, 1986b, 1987). The
s”’
Plate 1. Flare plumes and ship trails in the North Sea. 0838,21.6.83 3-.
Hot spots and plumes
143 1
Plate 2. Straw burning hot spots after harvest, with some of industrial origin, e.g. Haarlem. Dunkerque. 1357,11.9X5.3-.
1432
R.S.SCORER
Plate 3.(aii) Warm spots of the same scene-mainly lakes. 0237,19.9.85,4.
Plate 3.(bi) The scene after sunrise with a plume 22 km long from the Krakow steel works. 0800,19.9.85,3-.
Hot spots and plumes
Plate 3.(bii) The dark areas are mainly forest, and the white areas clouds when seen by the ‘visible’ wavelength. 0800,19.9.85,2.
Plate 3.(c) Even a1 midday, when the ground is most brightly illuminated. the hot spots are clearly seen. The lakes (now cool spots for Ch4) are dark, having no glint. 1232,19.9.85.3-.
1433
1434
R.
s. SCORER
1435
Hot spots and plumes
only case on record of a pollution trail having a source which was not a ship is that of an oil rig with a flare in the North Sea. Flares are sometimes detectable by Ch3, but not by Chs 2 and 4, nor by any CZCS channel.
3. ILLMTRATIONS
3.1. North Sea ship andjlare
trails
Plate 1 shows three hot spots in the middle of the North Sea which are presumed to have been on oil rigs with flares. This occasion was of special interest because it was, until I 5 October 1985, the only case on record of ship trails in the middle of the North Sea. The additional interest in this case is that at least two of the trails are produced by flares, not by ships, and so they are an accurate record of the translation of the air. 3.2. Straw burn This topic was discussed by Muirhead and Cracknell (1985) who showed that hot spots created by farmers burning straw in the fields after the grain harvest could be detected. The hot spots in the case illustrated here (Plate 2) include a few industrial ones in addition to the hundreds of straw burnings. The hot spots on farms are seen in different places on different occasions and even vary between morning and evening. Assuming this sample to be representative of the 3 weeks or so of this practice, the total number of such fires in one season must be many thousand. 3.3. Industrial hot spots and plumes The nocturnal picture by Ch3 of Poland’s great industrial area of Upper Silesia (Plate 3,ai) shows several hot spots, the most easterly of which is the steel works just east of Krakow. Also visible are some lakes
which are larger but less bright than the hot spots. The simultaneous Ch4 picture shows these water surfaces but not the hot spots. After sunrise (Plate 3, bi) the hot spots are still visible by Ch3 above the level of the solar illumination, but iess brightly. The Krakow steel works now has a plume detected by Ch3 over a length of about 22 km. Such a plume has not appeared on other occasions so far examined and it was scarcely detectable by Ch2 (Plate 3, bii) or Ch4. It was not seen at 1232 (Plate 3,~). It is thought that it might have been caused by some shortlived operation such ascoke oven quenching. The same industrial area of SW Poland provides a good example of dense industrial haze containing several individual plumes (Plate 3, d). The most prominent plume, from Gliwice, seems to have contained water cloud, Individual clouds can be seen in the industrial areas of N Bohemia,and the air there is hazy with no obvious wind direction indicated by plumes. The wind drift is from the ESE in Upper Silesia. The satellite is viewing towards the sun, and the stretching of the pixel width in the east where the view is more oblique is obvious. Krakow is at the extreme eastern edge of the scene.
REFERENCES
Muirhead K. and Cracknell A. P. (1985) ht. J. Remote
Sensing 6, 827-833. Scorer R. S. (1986a) Etna: the eruption of Christmas 1985 as seen by meteorological satellite. Wearher 41, 378-384. Scorer R. S. (1986b) Cloud Invesrjgarion by Sure~~ite.Ellis Horwood (and John Wiley), Chichester. Scorer R. S. (1987) Ship trails. Armosph~rir Enuironment Zt, 1417-1425.
,fmdr~nr
in Gnat
Vol. 21. No. 6. pp. 1421-1435.
ooo4-6981/87 s3.00+0.00 Perpmon Journals Ltd.
1987.
Britain.
HOT SPOTS AND PLUMES: OBSERVATION METEOROLOGICAL SATELLITE
BY
R. S. SCORER Imperial College, London, U.K. (First receiued 8 April 1986 and received jar publicotion 17 November 1986)
Abstract-Many hot spots can be detected by Channel 3 of the sun-synchronous meteorological satellites NOAA 6-9. This is because that channel is much more sensitive to radiation from bodies in the temperature range 500-1500 K than to scattered sunshine or earth-temperature radiation. Hot bodies occupying only a small fraction ofa pixel can be detected. The other channels are designed to receive much greater power from earth radiation (300 K) or scattered sunshine (6000 K). Plumes from chimneys are very difficult to detect because they are small and rapidly diluted and dispersed. Some flukes of illumination make them detectable occasionally; and some infrequent meteorological circumstances make them persistent and large. Ship trails are quite exceptional and very obvious, but they are very rare.
1. DETECTION OF HOT-SPOTS
red or IR filters. Cloud shadows are well displayed and it records the scene as seen by scattered sunshine.
1.1. Satellite receiving systems
satellites are designed to provide photographic images either by recording scattered sunshine (Channels 1 and 2) or by receiving infra-red emissions from clouds and objects at the earth’s surface (Chs 4 and 5). Figure 1 shows the classical relative black body emission intensity, B, f (according to Planck), as a function of AT (bottom scale), I being the wavelength in pm, and T the absolute temperature in K. The wavelength bands (channels) received are chosen to be in windows of the water vapour spectrum, and are assumed to be well represented by the Planck function, although variable pollution may interfere with the image provided by the satellite. The maximum intensity is given the value unity, and occurs where IT = 2900. This investigation is restricted to the satellites NOAA 6-9 of which there are very good records in the University of Dundee archive. The following are the channels used:Meteorological
1 2 3 4 5
0.55-0.68 0.725-1.10 3.55-3.93 10.5-11.5 11.5-12.5
pm pm pm pm pm
(yellow-orange) (very near IR) (near IR) (IR) (IR).
Ch5 is so similar to Ch4 for our purposes that it needs no separate discussion. Chl is useful in identifying haze which may not be as clear on Ch2, but it has no interest in this section on hot spots. Therefore we concentrate on the properties of Chs 2,3 and 4. Ch2 is commonly called ‘visible’ although it is actually all in the IR: but it produces pictures very similar to those obtained with ordinary film using deep
Ch3 also picks up scattered sunshine by day but the intensity is much less than in Ch2, and the scattering has directional characteristics and intensity variations in clouds of water droplets whose size range includes the wavelength band. Small droplets scatter more powerfully fowards and backwards than sideways, and in droplets larger than about 30 p and clouds composed of ice crystals (which have a much larger typical size than liquid droplet clouds) the absorption in Ch3 is so large that for practical purposes the scatter is reduced to zero and the clouds appear black when the scene is printed in photo-positive, which is called Ch3-. This is because the channel was designed to measure emitted radiation and night images were to be compared with Chs 4 and 5, and printed in photo-negative with the warmest objects black and the coldest white: at night, in this mode, the pictures have many similarities to images of Ch4. but also some essential differences. Ch4 images are printed in photo-negative, and provide a measure of temperature. The intensity of scattered sunshine in Ch3 is greater than that of the emitted radiation, and overwhelms it. At sunrise Ch3 images consist of scattered sunshine in the east, but emitted radiation showing the temperature distribution in the west where it is still dark. Sometimes where the scattered sunshine is almost zero or very uniform, as is often the case at sea, temperature variations are evident by day. 1.2. Intensity characteristics
of Chs 2,3 and 4
In Fig. 1 the positions of the central wavelengths of Chs 1.2 and 3 are shown on the upper side of the bottom scale for 6000 K representing sunshine. The position of Ch4 is shown on the scale marked 300 K
1427
1428
R. S. SCORER
Km
a3xl
3003
4cm
5030
6am XT
7aO
Qo30
rr,coo
Fig. 1.
(for earth emissions). Relevant positions are marked on the curve and on the scales for 400 K and 700 K, representing warm and hot spots. The intensity is set at unity at the maximum where T = 2900, and the curve indicates the relative intensity as a function either of wavelength for a given temperature or of temperature for a given wavelength. For the solar beam and for the earth’s emission the central wavelength of Ch3 (L = 3.75 pm) has a small relative intensity equal to 0.044 at T = 300 K and to 0.006 for T = 6000 K. As the scale for the position of this wavelength at different temperatures shows, the relative intensity is unity for T = 784 K. At the same time the temperature scale for Ch4 (A = 11 pm) shows that while the relative amplitude is unity for T = 264 K it has falkn to 0.2 at T = 700 K. The consequence is that for hot bodies Ch3 has a relative intensity which, at 500 K, is 15 times the 300 K value, and at 800 K is 23 times that value. By contrast, for Ch4 hot bodies have a relative intensity which is decreased from unity at 264 K to l/2 at 477 K and to l/S at 700 K. These factors are to be multiplied by 7” according to Stefan’s law, to obtain the relative magnitudes of the signals at different temperatures. Ch3 is therefore much more sensitive to temperatures of earthly hot spots than Ch4. Approximately the part of the curve representing B ,,&n the range 260-800 K may be represented by a sine curve, viz 13,75= PB,,,,,=
0.5+SinTi >
T4.
(1)
The rate of increase of the part of this function in the bracket is a maximum at T = 480 K.
It is found in practice that although the scattered sunshine completely obscures the emissions from the earth in Ch3 by day, hot spots of significantly higher temperature show through brightly, as shown in the next section. In the same way the part of the curve representing B,, for 1= 11 pm (Ch4) over the temperature range 300-800 K may he approximately represented by a hyperbola, and the resulting formula is I,, = B,,$-“
r&T’.
The formulas (1) and (2) are not valid outside the temperature ranges specified. 1.3. How small a hot spot can be detected? We imagine a hot spot of temperature T, to occupy a fraction a of a pixel, the temperature of the surrounding land being TO.The emitted radiation received at the satellite in wavelength 1 is proportional to II=
(l-a)BL,r,,G+aB~,r,fi.
(3)
To exemplify this we put ,I = 3.75 P. TO = 300 K, T, = 700 K, and obtain lo-* i3,75 = (1 -a) x0.044x
81 +a x 0.975 x 2401
= 3.56 + 2337a z 3.56 (1 + 66Oa).
(4)
It is clear from this expression that if a exceeds about 0.0015, I3,,5 is more than twice the value when a = 0. The hot spot would then be detected as a pixel with more than twice the brightness of the surroundings, at night.
1429
Hot spots and plumes With a pixel size of 1.1 x 1.1 km” the hot spot would be easily detected if it had an area of about 40 x 40 m* or more. To be detected by day it would be ~CCCS~~T for the emission to exceed the intensity of the scattered sunshine. This would be achieved by a value of e around 0.01 because the sunshine is only two or three times as bright as the ordinary infra-red emissions in Ch3. Thus a hot spot of 100* m* will be easily detectable at 700 K. By contrast, when the same calculation is performed for Ch2 we find, with 1= 0.875 pm 10-*&V,
=(l-a)x
10-20x81+ax
1O-4x24O1
(5) and nothing would be detected in the ‘visible’ at night because the emission at such a short wavelength is far too small, even though the radiation from the hot area is a relatively large fraction of the total. By day the scattered sunshine is completely dominant, being around lo5 times as powerful. The radiation received by Ch4 (A = 11 p) is, likewise, IO-sI 11 =(l-a)xO.965x8l+axO.2x2401 2: 78(1+ 5a)
(6)
and the relative importance of the hot spot is eight per cent of what it is in the case of Ch3. 1.4. Detection of James Flares at chemical works and on oil or gas rigs may have temperatures in the range lOOO-1500 K. They are incandescent because of the carbon particles present. There is radiation from the gases present, but since this is mostly from Hz0 and COz it is effectively absorbed by the atmosphere fairly close to the flame. The values of BArfor Chs 2,3 and 4 for a pixel with ground at 300 K and two flames occupying fractions a and Ji of the area and having temperatures HMOand 1500 K, respectively are given in Table 1. For a pixel in which the flame occupies a small fraction, a, of the area the intensities of the radiation received at the satellite in the three channels are
At 1000 K the flame is not detected by Ch2 because the signal is too small compared with scattered day&bt. Even at night the channel is not designed to respond to such a low signal, and it does not ‘see’ the illumination of cities, although it might just detect a flame at 1500 K. A Iarge flame at 1000 K might have an area of the order of 30 x 30m which, with a pixel of size 1.1 x 1.1 km2, gives a value of a of 9 x 10T4. In that case the flame would give the pixel an intensity about three times the environmental intensity in Ch3, and would be detected. A much smaller flame with dark surroundings, such as at night of at sea where there is no glint, co&d be detected at about one tenth of that size, say IO x 10 m*. A flame at 1500 K would certainly be detected by Ch3, being 2f times as bright as at lOO0K. Ch4 however, will not detect a Rame even of the larger size or higher temperature just mentioned, because the intensity would only add about lx, and this would not distinguish it from other local variations, 1.5. Large warm spots A warm spot occupying a whole pixel can obviously be detected by Ch4 which is designed to separate areas whose temperature differs from that surrounding it by 1 or 2°C. Thus lakes or rivers are easily detected on a cool night or day. Ch2 produces a very similar appearance by day with water surfaces which have no glint. By both channels lakes are depicted as dark spots, often occupying a few adjacent pixels. They could appear as light spots on a warm day when they are cooler than the ground by Ch4, or when they have sun glint by Ch2. A very warm spot, such as a volcano which is not erupting and whose top has a temperature of, say, 350 K over a whole pixel, would be much more prominent on Ch3 than on Ch4. The relative intensities of the surroundings and of the warm pixel are given in Table 2.
given by
x 81 +a x lOI3 x lo4 = 10a +/3x 0.13 x 5.06 x lo4 = 6.6 x lOJ/I Ch3: 10-813,ts = (1 -a)O.O44 x 81 +a x0.88x lo4 = 3.6+88Otla [email protected] 104=3.6+21500~ Ch4: lo-*1,,=(1-a)0.965x81 +axO.O7x10’=78+620a +/5x0.017x5.06x104=78$860~
Ch2:
10-810,s,S = (1 -a)1O-2o
Table I. The values of the intensity of the emission in channels 2, 3 and 4 relative to the maximum at various temperatures
channel a. Temperature E 1000 1500
;.*75
z.75
4 11
IO_‘0 lo-’ 10-3 0.13
0.044 0.975 0.86 0.425
0.965 0.2 0.07 0.017
for for for for for for
lOOOK 150OK lOOOK 1500 K lOOOK 1500 K.
Table 2. The relative intensities of the surroundings and of a warm pixel at the temperatures given, and their ratio in channels 3 and 4
Ch3 Ch4
Surroundings at 300°K
Warm spot at 350°K
0.044~81 0.965 x 8 1
0.15 x 150 0.84 x 150
Ch3 is clearly more sensitive.
Ratio &i,ol&oo 6.3 1.6
1430
R. S. SCORER
To be evident by day on Ch3 the intensity must exceed that of the surrounding scattered sunshine, which is around twice that of the emitted radiation. At 350 K the factor of 6.3 will give the warm spot greater prominence on Ch3 than it has in Ch4. (See Stromboli in Scorer, 1986a.)
2. DETECHON OF PLUMES
2.1. Plumes of individual chimneys To be detected by satellite a plume must extend over at least two pixels, and perhaps scarcely deserves the name of plume unless it extends for three or more. Most industrial plumes are less than 1 km in length and cannot therefore be detected by routine meteorological satellite. Indeed cases of identifiable plumes are very few. To be identified by Ch4 a plume must contain particles radiating as black bodies. This condition would be fulfilled by a plume in which water was condensed and the plume would look like low cloud. The temperature contrast with the surface background is unlikely to be large enough for clear identification except in exceptional circumstances of very cold ground. Plumes of condensed water such as from a cement works or a power station burning natural gas usually evaporate fairly quickly and are seldom more than a few chimney heights in length in the absence of natural cloud. Some plumes, thought to contain SO,,
*ML 3
are much longer because of the hygroscopic condensation nuclei. Neither Ch2 nor Ch3 is likely to make a plume from a single chimney visible except at sunrise or sunset when the plume is illuminated against a much darker ground which may be largely in shadow. Ch3 is particularly good in this respect when viewing a cloud of condensed water towards the sunrise because the forward scattering by small droplets is relatively intense. (See Scorer, 1986b, 8.9 and 3.15.) 2.2. Plumes from industrial complexes These can sometimes be detected and the best viewing conditions are provided by Ch 1 when looking towards the sun, because forward scattering is the most intense in the case of haze, and Chl has a relatively much smaller signal from typical ground than Chs 2 and 3 in sunshine. Chs 1-4 of the CZCS satellite Nimbus 7 are sensitive to haze, but because of the different viewing angle are relatively more sensitive to tenuous cirrus, and are unreliable for routine meteorological purposes. 2.3. Plumes at sea Ship trails are an exception to the general rule that pollution trails are so quickly diluted that they become invisible by satellite before they achieve a length of a few pixel widths. Plumes of this kind occur only at sea, and there only rarely because they depend on the air being exceptionally clean (Scorer, 1986b, 1987). The
s”’
Plate 1. Flare plumes and ship trails in the North Sea. 0838,21.6.83 3-.
Hot spots and plumes
143 1
Plate 2. Straw burning hot spots after harvest, with some of industrial origin, e.g. Haarlem. Dunkerque. 1357,11.9X5.3-.
1432
R.S.SCORER
Plate 3.(aii) Warm spots of the same scene-mainly lakes. 0237,19.9.85,4.
Plate 3.(bi) The scene after sunrise with a plume 22 km long from the Krakow steel works. 0800,19.9.85,3-.
Hot spots and plumes
Plate 3.(bii) The dark areas are mainly forest, and the white areas clouds when seen by the ‘visible’ wavelength. 0800,19.9.85,2.
Plate 3.(c) Even a1 midday, when the ground is most brightly illuminated. the hot spots are clearly seen. The lakes (now cool spots for Ch4) are dark, having no glint. 1232,19.9.85.3-.
1433
1434
R.
s. SCORER
1435
Hot spots and plumes
only case on record of a pollution trail having a source which was not a ship is that of an oil rig with a flare in the North Sea. Flares are sometimes detectable by Ch3, but not by Chs 2 and 4, nor by any CZCS channel.
3. ILLMTRATIONS
3.1. North Sea ship andjlare
trails
Plate 1 shows three hot spots in the middle of the North Sea which are presumed to have been on oil rigs with flares. This occasion was of special interest because it was, until I 5 October 1985, the only case on record of ship trails in the middle of the North Sea. The additional interest in this case is that at least two of the trails are produced by flares, not by ships, and so they are an accurate record of the translation of the air. 3.2. Straw burn This topic was discussed by Muirhead and Cracknell (1985) who showed that hot spots created by farmers burning straw in the fields after the grain harvest could be detected. The hot spots in the case illustrated here (Plate 2) include a few industrial ones in addition to the hundreds of straw burnings. The hot spots on farms are seen in different places on different occasions and even vary between morning and evening. Assuming this sample to be representative of the 3 weeks or so of this practice, the total number of such fires in one season must be many thousand. 3.3. Industrial hot spots and plumes The nocturnal picture by Ch3 of Poland’s great industrial area of Upper Silesia (Plate 3,ai) shows several hot spots, the most easterly of which is the steel works just east of Krakow. Also visible are some lakes
which are larger but less bright than the hot spots. The simultaneous Ch4 picture shows these water surfaces but not the hot spots. After sunrise (Plate 3, bi) the hot spots are still visible by Ch3 above the level of the solar illumination, but iess brightly. The Krakow steel works now has a plume detected by Ch3 over a length of about 22 km. Such a plume has not appeared on other occasions so far examined and it was scarcely detectable by Ch2 (Plate 3, bii) or Ch4. It was not seen at 1232 (Plate 3,~). It is thought that it might have been caused by some shortlived operation such ascoke oven quenching. The same industrial area of SW Poland provides a good example of dense industrial haze containing several individual plumes (Plate 3, d). The most prominent plume, from Gliwice, seems to have contained water cloud, Individual clouds can be seen in the industrial areas of N Bohemia,and the air there is hazy with no obvious wind direction indicated by plumes. The wind drift is from the ESE in Upper Silesia. The satellite is viewing towards the sun, and the stretching of the pixel width in the east where the view is more oblique is obvious. Krakow is at the extreme eastern edge of the scene.
REFERENCES
Muirhead K. and Cracknell A. P. (1985) ht. J. Remote
Sensing 6, 827-833. Scorer R. S. (1986a) Etna: the eruption of Christmas 1985 as seen by meteorological satellite. Wearher 41, 378-384. Scorer R. S. (1986b) Cloud Invesrjgarion by Sure~~ite.Ellis Horwood (and John Wiley), Chichester. Scorer R. S. (1987) Ship trails. Armosph~rir Enuironment Zt, 1417-1425.