El chichon volcanic ash effects on atmospheric haze measured by NOAA7 AVHRR data

El chichon volcanic ash effects on atmospheric haze measured by NOAA7 AVHRR data

REMOTE SENSING OF ENVIRONMENT 16:157-164 (1984) 157 SHORT COMMUNICATION E1 Chichon Volcanic Ash Effects on Atmospheric Haze Measured by NOAA7 AVHRR ...

444KB Sizes 0 Downloads 9 Views

REMOTE SENSING OF ENVIRONMENT 16:157-164 (1984)

157

SHORT COMMUNICATION E1 Chichon Volcanic Ash Effects on Atmospheric Haze Measured by NOAA7 AVHRR Data

ARTHUR J. RICHARDSON Remote Sensing Research Unit, Agricultural Research Service, U.S. Department of Agriculture Weslaco, TX 78596

The increase of atmospheric haze caused by volcanic eruptive products, as measured by the NOAA7 advanced very high resolution radiometer (AVHRR) using visible (0.58-0.68 txm) and reflective infrared (0.725-1.10 /~rn) digital count data, was demonstrated. Prevailing nadir atmospheric transmission, measured on the ground at Weslaco, TX before the E1 Chiehon volcano eruptions from 28 March through 4 April 1982, was 0.69 and afterward 0.62, a statistically significant decrease of 10.1%. The decrease of atmospheric transmission measured on the ground was significantly correlated (r 2 = 0.53) with increases of NOAA7 AVHRR minimum digital count values (DCm) in the infrared band obtained over the Gulf of Mexico. The AVHRR DCm's in both visible and infrared bands ranged fl'om 17 to 37 before as compared to ranges of 45 to 116 after the eruption.

Introduction The monitoring of earth resources is dependent on analyzing reflected and emitted radiance from the ground as received by satellite based remote sensors. Transient atmospheric effects, such as clouds, and haze (man-made and natural pollutants), interfere with earth resource monitoring activities by altering the detected values of reflected and emitted radiance. The series of E1 Chichon volcanic eruptions (Fig. 1) affected atmospheric conditions for monitoring reflected radiance in south Texas using the NOAA7 advanced very high resolution radiometer (AVHRR). The series of volcanic eruptions began on 28 March, with the most violent eruption being on 4 April 1982, producing a total volume of eruptive products of slightly less than 0.5 km 3 (SEAN Bulletin; 31 May 1982). The goal of this experiment was to demonstrate and document the effects of ©Elsevier Science Publishing Co., Inc., 1984 52 Vanderbilt Ave., New York, NY 10017

El Chichon volcanic ash on atmospheric haze as measured by NOAA7 AVHRR data over south Texas. An attempt was not made to evaluate atmospheric aerosols at multiple locations around the globe as did Griggs (1983). Measurements at several times over 1981 and 1982, at Weslaco, TX provided statistics to document the before and after atmospheric haze effects of the El Chichon eruptions over south Texas. The objective was to relate ground measurements of atmospheric optical depth and radiance values of the NOAA7 AVHRR to increases in atmospheric haze conditions in south Texas before and after the E1 Chichon volcanic eruptions.

Experimental Procedures South Texas was chosen as the site for obtaining NOAA7 data for this study because of the existing monitoring station at Weslaco, TX that was shown by Robock 00344257/84/$3.00

A. J. RICHARDSON

158

FIGURE 1. E1 Chichon volcano dust cloud disposition on 8 April 1982 (Robock and Matson, 1983) over south Texas (...). Point A shows location of Eppley solar tracker station at Weslaco, Texas. Point B shows location of NOAA7 AVHRR minimum digital count samples. Point C is approximate location of the E1 Chichon volcano. The solid straight line shows where an 8 April 1982 AVHRR scan line profile was developed.

and Matson (1983) to be covered by the El Chichon volcanic dust cloud (Fig. 1). The south Texas area extends from just north of San Antonio south to Mexico. Direct (Ed) irradiation were measured at Weslaco, TX using an Eppley normal incident pyranometer t mounted on an Eppley solar tracker 1 (Point A, Fig. 1). The pyranometer is sensitive over the 0.3-2.8 /~m range. Data were obtained from 30 December 1980, through 16 December 1982, from 38 sampling dates. Measured irradiance levels are accurate to 2% (Rao and Bradley, 1983). Direct irradiance was converted to optical depth (-r) using a variation of Langley's (1881) method for computing the solar constant; = ln(E o / E d ) / m , where E o is the solar constant, E d is the direct irradiance measured by the normal incident pyranometer, and m is the air

mass (m = sec Z, Z is the solar zenith angle). Up to 140 measurements of E a per sample day were obtained at 3-rain intervals using a Polycorder 1 automatic data logger. Histograms of the derived optical depth were used to determine the prevailing daily optical depth and nadir atmospheric transmission. These daily values were plotted over time (day of year D) to show the seasonal variation in atmospheric optical depth and nadir atmospheric transmission for the south Texas area. Regression fits of cosine functions were used to approximate the seasonal change in nadir atmospheric transmission as

T = a o + alcos w t , 1Mention of a company name or trademark is for the readers' benefit and does not constitute endorsement of a particular product by the USDA over others that may be commercially available.

VOLCANIC ASH EFFECTS

where wt = 2~rD/365 and D is the day of year. The a 0 term corresponds to the prevailing T over the season and the a 1 term indicates the range/2 of T over the season. The values of the linear regression coefficients a 0 and a 1 were compared statistically before and after the volcano erupted (Steel and Torrie, 1960). The nadir atmospheric transmission can be converted to optical depth by using • = ln(1/T). Imagery from the NOAA7 AVHRR over south Texas for eight dates between October 1981 and July 1982 were analyzed on an I2S Model 70 Image Analysis System. 1 The NOAA7 satellite operates at an altitude of 850 km with a local standard equatorial crossing time of 0230 and 1430. It has an orbital period of 102 min, which produces 14.1 orbits per day. Its five-band AVHRR has an instantaneous field of view (IFOV) of 1.4 mrad that yields a resolution of 1.1 km at nadir. The five spectral band intervals of the AVHRR are 0.58-0.68, 0.725-1.1, 3.55-3.93, 10.5-11.5, and 11.5-12.5/zm (Schneider and McGinnis, 1982). Only bands 1 and 2 were used for this study. The digital count values in bands 1 and 2 were converted to eight-bit precision (256 levels) by extracting digital count values in the 33-288 range from the original ten-bit 0-1023 range. Based on calibration coefficients recorded on the NOAA7 computer tapes and using procedures given in the NOAA Polar Orbiter Data Users Guide (December, 1981) the extracted digital count range corresponds to a reflectance range of 0-27.2% at the top of the atmosphere, where each digital count equals 0.1069% reflectance. Radiance (L) as measured by the NOAA7 AVHRR at the top of the atmo-

159

sphere is given by (Turner, 1972) L = E T R / I r + Lp.

where E is total irradiance at the ground, T is nadir atmospheric transmission, R is ground reflectance, and Lp is path radiance. The ETR/~r term contains the useful information needed to monitor surface conditions. The Lp term is undesirable noise that increases with increasing haze. Minimum digital count values from clear water bodies in a NOAA7 AVHRR scene are a measure of path radiance and provide an index of the haziness of the atmosphere (Ahern et al., 1977; Richardson et al., 1980). Digital count scan line profiles for the Gulf of Mexico (Fig. 1) across the NOAA7 orbital path were developed, for one scene (8 April 1982), to show the effect of volcanic haze in the atmosphere. The minimum digital count values for the Gulf of Mexico (Point B, Fig. 1) were determined from NOAA AVHRR data distributions in scatter diagrams of bands 1 and 2 for each of the eight south Texas scene dates. The minimum digital count values in bands 1 and 2 were normalized to a solar zenith angle of 39 ° using procedures described by Richardson (1982). These minimum digital count values, corrected for solar zenith angle, were related to the optical depth seasonal variation measured at Weslaco, TX to study predictive relations of atmospheric transmission given minimum digital count values from NOAA7 AVHRR data.

Results and Discussion Figure 2 presents the optical depth measurements of the atmosphere as measured at the Weslaco research farm for 38

160

.8

,

,

,

,

,

,

,

,~

,

,

,

,

,

,

,

? ~

,

,

,

; ,~

x CLEAR 0 CLOUDY

,

,

,

:: iS ?¢:

o

.50

.6

: o

I I--O_ ILl C3

.4

_.J .< (3

i

:

o

i

.45

.55

z O co

.61

Z

(/3

~ ,67

i

,< rY E--o rY

,74

rl

0

o

:~2

i i

uJ -1fl_ (/)

.82 0

.2

.90 0.0

i

i

i

I

I

t

I

JAN MAR MAY JUL

I

I

I

I

SEP NOV

I

I

'l '

MAR

JAN

I

I

MAY

i

"l'

1.00

I

JUL

SEP NOV

1982

1981

FIGURE 2. Atmospheric translnissiou and optical depth ineasurements for 38 sample dates in 1981 and 1982 as measured at the Weslaco research farm: ( × ) clear; (rq) cloudy.

APRIL

8,1982

250

CLEAR

HAZE

CLOUD

200

!;;150

g O

100

50

2 0

, 0

EAST

1 512

,

I 1024

,

SCAN LINE PIXEL LOCATION

I 1536

,

I 2048

WEST

FIGURE 3. Scan line profile (see Fig. I) taken on 8 April 1982, from the Gulf of Mexico showing the effects of atmospheric haze on all five NOAA7 AVHRR bands due to dust from the E1 Chichon volcano eruptions on 4 April 1982.

161

VOLCANIC ASH EFFECTS

sample dates in 1981 and 1982. These measurements were obtained from direct irradiance ( E d) data using an Eppley normal incident pyrometer mounted on an Eppley solar tracker. The seasonal cosine correlation of the nadir atmospheric transmission (T) to day of year (D) before and after the eruption were:

0.69 and afterward was 0.62, a statistically significant decrease of 10.1% at the 0.01 probability level in the nadir atmospheric transmission. The range of T before the eruption was 0.106 and afterward was 0.092, a decrease of 0.014 in the variation of atmospheric transmission that was not significantly different. Thus, these results appear to show that a higher T = 0.686 + 0.106cos w t , before, r e = 0.69 atmospheric aerosol content was apparent after the eruptions. (significant at 0.01 probability level), Figure 3 is an 8 April 1982, NOAA7 T = 0.617 + 0.092cos w t , after, r 2 = 0.56 AVHRR scan line profile over the Gulf of Mexico (Fig. 1). The western half of the (significant at 0.01 probability level). profile was contaminated by haze created The prevailing T before the eruption was by atmospheric dust from the El Chichon December 18, 1981

O c t o b e r 26, 1981 250

2

200

2

150

1

• *t I

..: !: .. •

Q

"~

,oo

50

it,"

0

50

0

100

150

200

50

250

100

150

200

250

BAND 2

BAND 2

February 16, 1 9 8 2

March 12, 1 9 8 2

250

250

-.- ..

L"

20O

2!00

"' ': " .~':"

150

150

E3 z ;~ 1001

m < 100

".' •

=

50

.,

°

t.

~'~ "

~:

50

.-

0 50

100

150

BAND 2

200

250

0

50

100

150

200

250

BAND 2

FIGURE 4. Scatter diagrams of NOAA7 AVHRR data, Band 1 (0.58-0.68 gin) and Band 2 (0.725-1.10 /xm), before E1 Chichon volcanic eruptions began on 28 March 1982.

162

A.J. RICHARDSON April 8, 1 9 8 2

Apdl 25, 1 9 8 2

250

250

.,,..

. :5:

2O0

,... "o:.

2OO •

s - : r'



150

, • C".j ~..

1:3

150

,z~ ~t.:': .R. :.'~ ..

-~

loo

~..

100

5O

5O

0

0 0

50

150

150

2~

250

0

50

100

BAND 2

150

June 14, 1982

July 1, 1 9 8 2

250

250

,"

2OO

. ~_-~ • .. 4 . ! :

20O

:':'..

Z..-

: . - . ~ : 7." 150

" "," ,, ;." -" ,"-~ -, ~

150

"-~ f' ~

100 •

.

2,50

200

BAND 2

• .°U, ~I¢.~I

1oo

:"

-

.7

:-

50

"

~-

...~:. v

50

0

0

0

50

100

150

200

250

BAND 2

0

50

100

150

200

250

BAND 2

FIGURE 5. Scatter diagrams of NOAA7 AVHRR data, of Band 1 (0.58-0.68 ~m) and Band 2 (0.725-1.10 ~m), after El Chichon volcanic enlptions began on 28 March 1982.

volcano as delineated in 8 April 1982, by Robock and Matson (1983). One effect of the haze was to decrease the digital count values of band 3. Bands 1 and 2 digital count values were increased; this effect is a well-known measure of path radiance (Ahem et al., 1977; Richardson et al., 1980). Bands 4 and 5 were relatively unaffected by the haze conditions. The restdts indicate that the NOAA7 AVHRR measured higher haze content over the western half of the Gulf than over the eastern half on this scene date. Figures 4 and 5 present the NOAA7 AVHRR band 1 and 2 scatter diagrams

before (26 October and 18 December 1981, 16 February and 12 March 1982) and after (8 April, 25 April, 14 June, and 1 July 1982) the E1 Chichon volcano eruptions, respectively. These scatter diagrams were developed by regular subsampiing of 0.4% of a 512×512 pixel scene (262, 144 pixels), covering the south Texas area, at the nominal 1 km resolution pixel size that is originally recorded on the NOAA7 AVHRR data tapes; thus, there are about 1000 pixels plotted in each scatter diagram for each scene. The two main clusters of data in each diagram correspond to water from the Gulf of

VOLCANIC ASH EFFECTS TABLE 1

163

NOAA7 AVHRR Scene Dates for South Texas a DIGITAL COUNT

NOAA7 SCENE DATES

LOCAL STANDARD TIME

10/26/81 14:27 12/18/81 14:17 02/16/82 14:19 03/12/82 14:38 E1 Chichon Volcanic Eruptions 04/08/82 14:21 04/25/82 14:21 06/14/82 14:30 07/01/82 14:29

SOL~ ZENITH ANGLE

50.0 57.0 46.1 42.0

UNCORRECTED CORRECTED VIEW ANGLE 1 2 1 2 32,0 43.0 46.0 24.0

T

r

27 26 26 26

21 20 15 20

33 37 29 27

25 29 17 21

73 79 76 72

0.31 0.23 0.27 0.33

126 125 75 70

75 90 55 51

116 112 65 62

70 81 49 45

60 58 53 53

0.51 0.54 0.63 0.63

(3/28/82 to 4/4/82) 33.0 29.9 27.0 28.0

41.0 42.0 40.0 40.0

~The median LST solar zenith angle over south Texas for each scene date are also given along with general atmospheric conditions. Tile median longitude and latitude for the south Texas study area was taken as 98.0 ° and 27..5°i Tile minimum NOAA7 AVHRR digital counts (DCm) for bands 1 and 2 are given for water (Gulf of Mexico). The minimum digital count are shown uncorrected and corrected for a solar zenith angle of 39 °. Nadir atmospheric transmission (T) and optical depth ( r ) estimated from ground observations are given also.

Mexico, lakes, and reservoirs and land surfaces for agricultural and rangeland areas in south Texas. The water cluster is smaller than the land cluster. The clusters are closer to the origin on dates before as compared to after the eruption because increased haze increases the digital count values in both bands 1 and 2. The minimum digital count in bands 1 and 2 of the water clusters, from Figures 4 and 5, can be used as an estimate of the path radiance produced by atmospheric haze. These minimum digital count values for the water cluster in bands 1 and 2 are recorded in Table 1 for all eight NOAA7 scene dates. The correction to a solar zenith angle of 39 ° are also given. The digital count values are much lower before (17-37 range) the eruption dates than after (45-116 range), indicating lower atmospheric haze before compared to after the eruption dates. The optical depth and atmospheric transmittance, as estimated from the seasonal cosine regression equations obtained

from ground observations, are listed in Table 1 for all eight NOAA7 scene dates. The correlation of the NOAA7 corrected minimum digital count for band 1 (r 2 = 0.47) to atmospheric transmittance was not significant. The correlation with band 2 ( r 2 = 0 . 5 3 ) was significant at the 0.05 probability level. These results indicate that predictive relations of atmospheric transmission to minimum NOAA7 AVHRR digital count data could be developed. Additional AVHRR scene dates are needed to more htlly develop these relations. These relations should be slightly different to similar studies developed for the NOAA6 AVHRR by Griggs

(1983). Conclusions Ground-collected direct irradiance measurements and visible and infrared radiance from the NOAA7 AVHRR showed that the E1 Chichon volcanic

164 eruptive products increased the atmospheric haze over south Texas. I thank W a y n e Swanson for his optical depth data collection efforts and computer data processing and graphic work.

References Ahem, F. J., Goodenough, D. G., Jain, S. C., Rao, V. R., and Rochon, G. (1977), Use of clear lakes as standard reflectors for atmospheric measurements, Proc. Eleventh Int. Symp. Remote Sens. Environ. I: 731-755. Griggs, M. (1983), Satellite measurement of tropospheric aerosols, Adv. Space Res. 2(5): 109-118. Kidwell, K. B. (1981), NOAA Polar Orbiter Data (TIROS-N, NOAA6, and NOAA7) Users Guide, revised December, Satellite Data Services Division, Washington, DC. Langley, S. P. (1881), The bolometer and radiant energy, Proc. Am. Acad. Sci. 8: 342. Rao, Nagaroja, C. R., and Bradley, W. A. (1983), Effects of the El Chichon volcanic dust cloud on insolation measurements at Corvallis, Oregon, Geophys. Res. Lett. 10(5): 389-391.

A. J. RICHARDSON Richardson, A. J., Escobar, D. E., Gausman, H. W., and Everitt, J. H. (1980), Comparison of Landsat-2 and field spectrometer reflectance signatures of south Texas rangeland plant communities, Machine Processing of Remotely Sensed Data, June, pp. 88-91. Richardson, A. J. (1982), Relating Landsat digital count values to ground reflectance for optically thin atmospheric conditions, Appl. Opt. 21(8): 1457-1464. Robock, Alan, and Matson, Michael (1983), Circumglobal transport of the E1 Chichon volcanic dust cloud, Science 221(7): 195-197. Schneider, S. R., and McGinnis, D. F., Jr. (1982), The NOAA/AVHRR: A new satellite sensor for monitoring crop growth, Machine Processing of Remotely Sensed Data Symposium, pp. 281-297. Steel, R. G. D., and Torrie, J. H. (1960), Principles and Procedures of Statistics, McGraw-Hill, New York, 481 pp. Turner, R. E. (1972), Atmospheric model for correction of spacecraft data, Proc. Eighth Int. Symp. Remote Sens. Environ. II: 895-934. Received5 October1983;revised10 March1984.