Mapping Burns and Natural Reforestation Using Thematic Mapper Data

(1)

This article was downloaded by: [University of Waterloo]

On: 08 February 2015, At: 07:08 Publisher: Taylor & Francis

Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Geocarto International

Publication details, including instructions for authors and subscription information:

http://www.tandfonline.com/loi/tgei20

Mapping burns and natural reforestation using thematic Mapper data

M.J. López García ^a & V. Caselles ^b

a Departament de Geografia, Facultat de Geografic i Història , Universitat de València , Ap, 22060, València, 46080, Spain

b Departament de Termodinàmica, Facultat de Física , Universitat de València , Doctor Moliner 50, Burjassot, 46100, Spain

Published online: 17 Sep 2008.

To cite this article: M.J. López García & V. Caselles (1991) Mapping burns and natural reforestation using thematic Mapper data, Geocarto International, 6:1, 31-37, DOI: 10.1080/10106049109354290

To link to this article: http://dx.doi.org/10.1080/10106049109354290

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or

systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

(2)

Mapping Burns and Natural Reforestation Using Thematic Mapper Data

M.J. López García

Departament de Geografia, Facultat de Geografic i Història, Universitat de València,

Ap, 22060, 46080 València, Spain.

V. Caselles

Departament de Termodinàmica, Facultat de Física,

Universitat de València,

Doctor Moliner 50, 46100 Burjassot, Spain.

Abstract

Remote sensing techniques are specially suitable to detect and to map areas affected by forest fires. In this work, Landsat 5 Thematic Mapper (TM) data has been used to study a number of forest fires that occurred in the province of Valencia (Spain) and to monitor the vegetation regeneration over burnt areas.

A reference area (non-burnt forest) was established to assess the change produced by fire. The radiance in the thermal band (10.4-12.5 μm) and the normalized difference in reflectance between near 1R (0.76-0.90 μm) and middle IR (2.08-2.35 μm) were the most suitable parameters to map burnt areas. This index can also be used for monitoring vegetation regeneration in burnt areas. About a month after the fire, the burns show temperatures of 5-6 °C higher than those found in the reference area, and the vegetation index shows negative values whereas the reference area values remain positive. The differences between the burns and the reference area for the vegetation index decrease with time as vegetation regenerates.

1. Introduction

Since the launching of ERTS 1 in 1972, later renamed Landsat 1 (NASA, 1972), different studies have been carried out with the purpose of evaluating the size of the areas affected by forest fires using data from the Multispectral Scanner (MSS) sensor onboard this satellite.

Husson (1980, 1985), using supervised classification techniques, studied forest fires that occurred in Corsica (France) from 1973 to 1980. Richards and Milne (1983) and Richards (1984) studied the potential of principal components transformation to map burns and to study vegetation regeneration. They analysed two forest fires that occurred near Sydney (Australia). On the other hand, Bermudez (1983), using Heat Capacity Mapping Mission (HCMM) data (NASA, 1978) showed the possibilities of using thermal images to map burns. She studied the forest fires that occurred in the province of Valencia (Spain) in

1978.

Since 1982, Landsat satellites have incorporated the Thematic Mapper (TM) sensor which has greatly improved spatial, spectral and radiometric resolution. The major design improvements of TM over the MSS include: 30 m versus 80 m ground resolution in the visible and reflected infrared bands; seven bands versus four bands, among these are a new band in the blue wavelength region, two new middle infrared bands and a high resolution (120 m) thermal infrared band; and 8 bits versus 6 bits radiometric resolution data, i.e., 256 levels compared to 64. (Spectral characteristics of the TM bands are given in Table 1).

This work has a twofold objective: firstly, to study the

possibilities of the Thematic Mapper sensor, especially the newly incorporated bands, for mapping forest fires;

secondly, to develop a procedure for monitoring vegetation recovery over burnt areas based on the use of multidate data.

2. Study Area

The study area is situated in the province of Valencia (Spain). This province is on the Spanish Mediterranean coast, between 39° and 40° North latitude, with a surface of about one million hectares, of which 212,379 hectares are forest and over 100,000 hectares are scrub or bush. In the last few years, each summer a great number of forest fires have destroyed a large part of the forest area of the Valencian region. Forest fires are becoming one of the main environmental problems in this zone.

From a climatic point of view, the province of Valencia is included in the so-called "Dry Iberia" and within this zone, in the eastern part of the semi-arid dominium, the climate tends towards mild-warm or subtropical. There- fore, the main climatic characteristics are a tendency to aridity and to climatic irregularity (Lines, 1970). As a result, the vegetation is adapted to the aridity and, consequently, there is a high risk of fire.

In the province of Valencia (Figure 1), the zone of Bun ol has been selected as a test area to carry out the study of vegetation regeneration because there were forest fires on three different dates within this area: October 18,1978 (4600 ha), April 27, 1982 (55Oha) and April 22, 1984(550 ha). The vegetation of this area is a typical Mediterranean

Downloaded by [University of Waterloo] at 07:08 08 February 2015

(3)

forest growing on limestone substratum. The Mediterra- nean forest is a heterogeneous formation composed mainly of oaks (Quercus rotundifolia), considered to be the "climax vegetation", which have given way, in the present, to coniferous species (Pinus halepensis, Pinus pinaster), and to secondary formations, originating from the degradation of the climax vegetation, composed of:

Quercus coccifera, Ulexparviflora, Rosmarinusofficinalis, Thymus vulgaris, Stipa tenacissima, etc. (Costa, 1982).

3. Methods

The assessment of the areas affected by forest fires and vegetation regeneration studies involve the use of multidate images. In such studies, we must distinguish in the total information contained in the scene between that which relates to the particular subject studied (forest fires in this case) and that which refers to other factors such as:

viewing and illumination geometry; radiometric sensor calibration; atmospheric, topographic and climatic effects;

and detector gains or offsets. These latter factors cause artificial differences that are added to those produced by the forest fire. Viewing geometry is not relevant in the case of the Thematic Mapper sensor because its scan angle is only +7.5° (Staenz et al, 1986; Sobrino et al, 1988). Therefore, this effect can be neglected in this study.

In order to avoid sensor and sun angle dependence, digital counts of the images have been converted into the corresponding physical magnitudes, as Price (1987) suggested. So, apparent spectral reflectance values px

have been obtained through the expression:

The radiance values for the seven bands are obtained from the digital counts, DC, through the expression:

KnL, Excos z

(1)

where k is the correcting factor for the variable distance Earth-Sun, Lx is the radiance which reaches the satellite (W m'2sr'), Ex is the solar irradiance (W nv2) given in Table

1, and z is the Sun zenith angle.

The expressions for k and z calculations can be found inlqbal (1983).

And the brightness temperature has been computed by inverse analysis of Planck equation, using a relation developed by Lansing and Barker (1983) forTM6, as:

(K) (2)

L n I + 1)

where kl = 1276.296 (W nr2 sr1), and k2 = 1260.56 K (NASA, 1984).

k = a, DC + a0 (3)

where a0 and a, are the calibration coefficients for each band (Table 1).

In order to minimize the effect of the remaining scene- dependent factors (atmospheric, topographic, and climatic effects and changes in detector gain or offset), a vegetated control area, "the reference area" was chosen. This area was selected from the non-burnt forest because it was spatially near burns and because its general characteristics (soil, humidity, topography and type of vegetation) were representative of pre-burnt conditions. Satellite data can be acquired nearly simultaneously over both the test and reference area; thus, the existing differences between burnt and unburnt forests may be obtained. In this way, the scene-dependent effects will not affect the results because they could be considered to be the same for both area.

The areas affected by forest fires were located on the digital image starting from the information on size and site provided by the Instituto para la Conservation de la Naturaleza (ICONA). Georeferenced registration was not required since the spatial resolution of Thematic Mapper data facilitated the location of affected areas by using the identification of local features (streams, roads, reservoirs, villages...). Band 4 data was particularly useful. However, a scene-to-scene registration was used in the natural reforestation study with multidate images.

4. Results and Discussion

4.1 Mapping of The Burnt Area:

Mapping of the areas affected by forest fires has been based on the analysis of the spectral signatures obtained both in the reference and burnt areas, and likewise with the evaluation of the thermal differences generated between them.

4.1.1 Analysis of the Spectral Slngatures:

Spectral signatures of several forest fires have been obtained from TM data acquired one or two months after the times of the fires. All have characteristics similar to that shown in Figure 2, which corresponds to the fire on April 22,1984. In this Figure, the spectral signature of the reference area has also been represented. This is the signature that the burnt area had before the fire occurred, as can be observed from Figure 3, in which the signatures of burnt and reference areas are obtained from data acquired before the fire date.

The heterogeneity of the control and test areas — due to variation in tree spacing, height and species composi- tion, orientation of slopes, etc. — explains the high values

32

(4)

J

^ i BURNT ^

^ / 22-4-84

/ i s RKFRRKNCr1!

BUPOT

18-10-78

0

R.Magro

5

BUNOL

\

10Km

Figure 1 Map or the study area.

TABLE 1

CALIBRATION COEFFICIENTS AND SOLAR IRRADIANCE FOR LANDSAT 5 TM BANDS

BANDS

TM1 TM2 TM3 TM4 TM5 TM6 TM7

SPECTRAL RANGE (jim)

0.45 - 0.52 0.52 - 0.60 0.63 - 0.69 0.76 - 0.90 1.55 - 1.75 10.4 -12.5

2.08 - 2.35

CALIBRATION COEFFICIENTS ( W m "2 sr •

a0 -0.067 -0.16 -0.11 -0.23 -0.086

2.6 * -0.0511

! )

ai 0.0420 0.103 0.0650 0.117 0.0273 0.118 ' 0.0169

SOLAR IRRADIANCE

(Wirr2)**

1 1 9 1 3 0 1 0 4 1 3 3

44 - 18

after Metzler and Malila (1985) after Markam and Barker (19851

2 5 - 20H

1 5 - ixi

• REFERENCE

•o- FIREBURN

TM1 TM2 TM3 TM4 TM5 TM7

Figure 2 Special signature corresponding to the reference area and to the fire burn of April 22, 1984, derived from Landsat 5 (TM) data of June 1. 1984 (9.30h. GMT). Values n* the mean ± standard deviation.

oLU

z

LU

REFL

25- 20- 15- 10- 5 -

- » REFERENCE -a- BURNT

/ \

i • i • i • i • i • i

TM1 TM2 TM3 TM4 TM5 TM7

Figure 3 Spectral signatures corresponding to the reference area and to the area burnt on April 22, 1984, derived from Landsat 5 (TM) data of April 7, 1984 (9.3Oh. GMT).

(5)

of error bars in Figure 2, since several spectral classes are included within these areas. The bands 4, 5 and 7 showed the greatest differences between the control and the affected areas. The burnt area shows a lower reflectance value in the near infrared region (band 4) and a higher response in the middle infrared (bands 5 and 7), as a consequence of vegetation disappearance.

In order to define the most suitable "vegetation index"

for mapping burnt areas, statistical analysis of spectral reflectances was employed. Thus, the correlation matrix of the reference and burnt area is provided in Table 2. The most uncorrelated pair of bands is TM4-TM7, followed by the pairs TM4-TM5 and TM4-TM1. From these results, the most uncorrelated bands have been chosen to define a vegetation index suitable for evaluating surfaces affected by fire. Therefore, this index is the normalized difference between reflectances obtained from TM4 and TM7: p4-p7/

p4+p7. According to the characteristics of the spectral signature of vegetation and soil (Horlerand Barber, 1981).

this index will present negative values for areas without vegetation, such as burns, and positive values for areas with vegetation, these values being higher when denser vegetation exists.

It has been verified that the use of this index allows mapping of burnt areas from an image acquired one or two months after the fire date. As an example. Figure 4 shows the area affected by the fire on April 22, 1984 mapped from the vegetation index obtained from the Thematic Mapper data of June 1, 1984.

4.1.2 Thermal Band Analysis:

The Thematic Mapper thermal band, TM6, has also been examined in order to determine its possibilities for mapping burnt areas. At the time of satellite overpass (9.30 h. GMT), we found substantial differences between brightness temperature of burnt and reference areas.

Temperature mean values obtained in the affected area are 5— 6°C higher than those obtained in the reference one, on an image acquired one or two months after the fire date. Similar results were found by Bermudez (1983) who studied with HCMM satellite data the fire that has affected the area in 1978. This satellite overpassed the zone at the time of maximum temperature, when differences between bare soil and vegetated areas are the largest.

Figure 5 corresponds to the thermal band of the image of 1st June 1984. The brightest tones show the highest temperature values and enable the identification of the burnt area. Since emissivity corrections have not been considered, the temperature differences observed do not correspond to the absolute temperature differences.

A change of surface emissivity may be expected when a forest is burnt. According to the magnitude of the errors introduced by emissivity changes, 0.6°C by each 1 % in emissivity (Becker, 1978), and because the disappear-

ance of vegetation in our zone means a decrease of less than 2% in emissivity (Caselles, 1984), the emissivity correction could only add about 1 °C to the differences obtained.

4.2 Monitoring Natural Reforestation of Burnt Areas

The three fires that affected the Bufiol region since 1978, (October 18, 1978, April 27, 1982, and April 22, 1984,) (Figure 1) have been studied. We have used the TM images of 1 June 1984, 19 July 1984 and 16 March 1985, which supply data approximately 1 month, 3 months and

1 year respectively after the fire on April 22, 1984. The fires on April 27, 1982 and October 18, 1978 have been studied with those same images and, consequently, we have also information for about 2, 3 and 6 years after a fire.

Figure 6 represents the spectral signature of the reference areas obtained from the three images used. The amplitute of reflectance variations is limited for the reference area and we are only interested in analysing the relative variations.

Figure 7 shows the spectral signature of the 3 fires studied and that of the reference are obtained from the images of 1 June 1984 (Fig. 7a), 19 July 1984 (Fig. 7b) and 16March 1985 (Fig. 7c). The spectral curve of a burnt area has the same shape during about 1 year after the fire date (low values in TM4 and high values in TM7). After about 2 years, it changed, and progressively becomes similar to the spectral curve of the vegetated reference area. Thus, the shape of the curve 6 years after the fire is just a typical vegetation curve, similar to that of the reference area.

However, there is a shift between the two curves, since the regenerated areas show higher reflectance values than the reference area. This could be explained by the fact that the vegetation regenerated over the burnt area is not composed of exactly the same species as that which were growing before the fire. As we mentioned in § 2, the vegetation existing over the reference area is a well developed forest mainly composed of oaks and pines.

When a forest is degraded, as is the case of a forest affected by fire, and if the conditions make regeneration possible, the original vegetation will be replaced by the species corresponding to the first stages in the "vegetation series". Over this area, a scrub vegetation (Phlamido- Brachypodium retusi, Saxifaga-Hornungietum petrae) composed mainly of species such as: Rosmarinus officinalis. Erica multiflora, Ulexparviflora, etc. forms the first stage in the vegetation recovery to climax forest conditions. This is the vegetation that is growing in the area burnt in 1982. However, in the area affected by fire in 1978, the vegetation has reached a further serial stage.

We can find a developed scrub (Rhamo-Quercetum cocciferae ulicetosum parvi florae) that corresponds to a second stage in the vegetation series, but differs from that

34

(6)

TABLE 2

CORRELATION MATRIX OF BURNT AND

REFERENCE AREA DATA 25

TM TM TM TM TM TM

1 2 3 4 5 7

TM 1

1.0000 0 1

TM 2

.9291 .0000

0 0 1

TM 3

.9228 .9741 .0000

TM 4

0 0 0 1

6386 7892 7247 0000

TM 5

0.8210 0.8239 0.8399 0.6119 1.0000

TM 7

0.7738 0.7330 0.7757 0.3901 0.9231 1.0000

Figure 4 Image of the vegetation index: p4-p7/p4+p7 made Landsat 5 (TM) acquired on June 1. 1984 (9.3Oh TMG). The dark correspond to negative values and show the fire burn that occurred April 22

1984.

Figure 5 Landsat 5 (TM) thermal band of June 1, 1984 (9.3Oh. GMT).

Brightest tones correspond to the highest temperature values, and identify the burn area.

UJ

o I

UJ UL UJ

rr

2 0 - 15- 1 0 - 5

1 JUNE 84 19 JULY 84 16 MARCH 85.

TM1 TM2 TM3 TM4 TM5 TM7

Figure 6 Seasonal variation of the spectral signature of reference area obtained from Landsat 5 (TM) images corresponding three different dates.

30

•9- REFERENCE

TM1 TM2 TM3 TM4 TM5 TM7

Co"

U l

aos "

Li.

UJ

rr 2 5 -

•

2 0 - 15- 10- 5-

* REFERENCE

•o- 3 MONTHS ^

-•- 2 YEARS f\

-o- 6 YEARS /jT

• • i • i • i

(b)

V\

1 • i

TM1 TM2 TM3 TM4 TM5 TM7 30

Ul

oz

LJJ UlLL DC

2 5 - 2 0 - 1 5 - 1 0 -

REFERENCE 1 YEAR 3 YEARS 6 YEARS

TM1 TM2 TM3 TM4 TM5 TM7

Figure 7 Spectral signatures of the reference area and the three fire burns studied, derived from an image of June 1, 1984 (7a), July 10.

1984 (7b) and March 10, 1985 (7c). It is shown in each figure the period of time passed after the fire took place.

(7)

of the reference area (climax conditions) in its degree of succession and because coniferous species are not present (R. Boluda, 1986, personal communication).

In order to quantify the vegetation regeneration on burnt areas. Table 3 shows the values of temperature and index p4-p7 / p4+p7 obtained on the reference and affected areas. The thermal band, which distinguishes burnt areas one or two months after the fire date, is not suitable to study vegetation regeneration because differences between both areas quickly decrease as the vegetation reappears (the temperature differences between the burnt and reference areas are of the same order as the standard deviation of these measurements). However, the difference between the burnt and reference area vegetation indices decreases with time more slowly than the temperature. In fact, by plotting the values of the vegetation cover estimated in situ, CE, versus the vegetation index differences, VID, (Figure 8), the following linear regression model is obtained:

CE=67.2-2.1 VID (4)

with a linear correlation coefficient of -0.97. In these statistical analysis we have used raw data not corrected for atmospheric effects. So systematics errors are introduced into the vegetation index. Moreover the number of data points is limited. For these reasons this regression model is valid whenever the atmospheric conditions be the same that in the images used, and for vegetation cover lower than 50-60%. These results show that the mean difference between the calculated and in situ estimated values is about 20%. However, we consider vegetation index p4-p7/p4+p7a good parameter for study- ing vegetation regeneration on burnt areas. Figure 9 is the index image of the Bunol area, which illustrates the vegetation regeneration on the firebum on October 18,

1978.

HI

o

10 20 30 40

VID (%)

Figure 8 Vegetation cover estimated in situ (CB) versus the difference between the burnt and reference area vegetation indices (VID).

Figure 9 Image of the vegetation index: 4-7/4+7 obtained from Landsat 5 (TM) taken on June 1, 1984. It shows the three fire burns studied and the reference area. Bright tones correspond to a highest index value.

TABLE 3

VEGETATION INDEX AND TEMPERATURE (mean and standard deviation) OBTAINED ON THE REFERENCE AREA AND EVERY FIRE BURN

ZONE

IMAGE ' TIME AFTER THE FIRE

T(°C)

P4-P7/P4+P7

<%) COVER (%)*

REFERENCE AREA

1june84 19july 84 16march85

18.1 (0.9) 33.7 (1.0) 8.5 (1.5) 30 (10) 28 ( 8) 32 (15)

50 - 60

FIRE BURN ON 22 APRIL 1984

1 June 84 19July 84 16 march 85 1 month 3 month 1year

24.4 (1.8) 39.7 (1.6) 10.0 (2.5) -19 (11) 2 ( 8 ) 0 (18)

0 - 10

FIRE BURN ON 27 APRIL 1982

1 June 84 19July 84 16 march 85 2 years 2 years 3 years

21.3 (1.3)37.3 (1.6) 9.0 (2.0) 16 ( 9 ) 17 ( 7) 16 (15)

4 0

FIRE BURN ON 18 OCTOBER 1978

1 June 84 19July 84 16 march 85 6 years

19.9 (0.9) 36.9 (1.2)12.4 (1.5) 27 (11) 21 ( 8) 23 (15)

50 - 60

* Ground data (R. Boluda, personal communication)

36

(8)

5. Conclusions

From the results obtained, we consider that the thermal band and index 4-7/4+7 are suitable parameters to evaluate the area affected by forest fires. This index is also appropiate for monitoring vegetation regeneration in burnt areas.

In this work, it has been also demostrated the useful- ness of using a reference area in order to minimize the influence of scene-dependent factors (atmospheric, radiometric, viewing and topographical effects). This is very interesting in multitemporal studies, such as vegetation monitoring, because time computer can be considerablely reduced.

Acknowledgements

The authors wish to thank the "Instituto para la Conservacion de la Naturaleza (ICONA)" for providing the data on forest fires and to the EARTHNET-NPOC Spain for supplying the Landsat images used in this work.

We also acknowledge R. Boluda from Department of Edaphology (University of Valencia, Spain) for his assist- ance and the field data provided. This work was partly supported by the CAICyT (Project no. A-172/85)

References

Becker, F. (ed.), 1978. Principes Physiques et Mathématiques de la Télédétection, CNES Publ., Tou- louse (France).

Bermudez, C., 1983. Determinatión de las Alteraciones Microclimáticas Producidas por Incendios Forestales Mediante Técnicas de Teledetecciíon, Tesis of Licenciatura. Universidad de Valencia (Spain).

Caselles, V., 1984. Estudio de las Medidas Radiométricas Realizadas desde Satélites Artificiales. Aplicación a la Agrometeorología. Tesis doctoral. Universidad de Valencia (Spain).

Costa, M., 1982. Pisos Bioclimáticos y Series de Vegetación en el Area Valenciana, Cuadernos de Geografía Vol.

31. pp. 129-142.

Horler, D.N. and J. Barber, 1981. Principles of remote sensing of plants, in Plants and the Daylight Spectrum, (Ed. Smith, H.), Academic Press, London, pp.43-63.

Husson, A., 1980. Télédétection des Incendies de Forêt en Région Mediterranéene. Cahiers de l'OPIT, No. 3, pp. 17-28.

Husson, A., 1985. Télédétection des Incendies de Forêt en Corse entre 1973 et 1980, Méditerrané, Vol. 54, No.

1-2 Télédétection III, pp. 53-59.

Iqbal, M., 1983. An Introduction to Solar Radiation.

Academic Press, Toronto.

Lansing, J.C. and J.L. Barker, 1983. Thermal Band Char- acterization of the Landsat 4 Thematic Mapper: Landsat 4 Science Investigation Summaries, NASA CP-2326.

Lines, A., 1970. The Climate of the Iberian Peninsula, World Survey of CIimatoIogy(Vol. 5), Carl, Chr. Wallen,

Amsterdam.

Markham, B.L. and J.L. Barker, 1985. Spectral Characteri- zation of the LANDSAT Thematic Mapper Sensors, Int.

J. Remote Sensing, Vol. 6, pp. 697-716.

Metzler, M.D. and W.A. Malila, 1985. Characterization and Comparison of Landsat-4 and Landsat-5 Thematic Mapper Data, Photogrammetric Engineering and Re- mote Sensing, Vol. 51, No 9, pp. 1315-1330.

NASA. 1972. ERTS Document No. 71SD4249, Goddard Space Flight Center, Greenbelt MD.

NASA, 1978. Heat Capacity Mapping Mission (HCMM) Data Users' Guide, Goddad Space Flight Center, Greenbelt, MD. (revised 1980).

NASA, 1984. Thematic Mapper Design Through Flight Evaluation, Final Report (revised) Santa Barbara Re- search Center RPT 41741. NASA Contract NAS5-24200,

12/1984.

Price, J.C., 1987. Calibration of Satellite Radiometers and The Comparison of Vegetation Indices, Remote Sens- ing of Environment Vol. 21, pp. 15-27.

Richards, J.A. and A.K. Milne, 1983. Mapping Fire Burns and Vegetation Regeneration Using Principal Compo- nents Analysis. Proceedings of the International Geoscience and Remote Sensing Symposium, San Francisco, CA.

Richards, J.A., 1984. Thematic Mapping from Multitemporal Image Data Using the Principal Compo- nents Transformation, Remote Sensing of Environ- ment, Vol. 16, pp. 35-46.

Sobrino, J.A., V. Caselles and F. Beckecr, 1988. A Theo- retical Model for Interpreting Remotely Sensed Ther- mal Infrared Measurements Obtained Over Agricul- tural Areas. Proceedings of the 4th Int. Coll. Spectral Signatures of Objects in Remote Sensing, Aussois (France) 18-22 January 1988. (ESA SP-287). pp. 217- 220.

Staenz, K., P. Meyer, K.I. Itten, D.G. Goodenough and P.M. Teillet, 1986, Viewing Angle Corrections of Air- borne Multispectral Scanner Data Acquired Over Forested Surfaces, Proceedings of the International Geoscience and Remote Sensing Symposium, Zürich (Switzerland), pp. 671-676.