Appendix 2: Additional Data on Remote Sensing Instruments

by
Kai Mäkisara

1. Introduction

This Appendix contains a review of the available remote sensing instruments supporting the discussion in Chapter 3 of this report.

2. Instruments by Class

2.1 Introduction

Based on the factors discussed in the previous section, the existing and future remote sensing instruments can be divided into several categories. The categories most useful for forest analysis are discussed below. The information presented is from numerous sources. One or two WWW addresses has been mentioned as reference where applicable. The WWW pages at these addresses contain further references.

2.2 Low-Resolution and Medium-Resolution Satellite Images (e.g., MODIS, MERIS, WIFS, AVHRR)

2.2.1 AVHRR

The best known low-resolution optical imager is probably AVHRR carried on-board the NOAA weather satellites. Depending on the model, AVHRR collects data from four or five spectral channels (see Table 25). URL http://www.ngdc.noaa.gov/seg/globsys/avhrr7.shtml

Table 25 The AVHRR spectral channels (µm).

Band number	Satellites: NOAA-6,8,10	Satellites: NOAA-7,9,11,12,14,15	Satellites: NOAA-KLM, METOP
1	0.58 - 0.68	0.58 - 0.68	0.58 - 0.68
2	0.725 - 1.10	0.725 - 1.10	0.725 - 1.0
3	3.55 - 3.93	3.55 - 3.93	1.58 - 1.64 (day) 3.55 - 3.93 (night)
4	10.50 - 11.50	10.3 - 11.3	10.3 - 11.3
5	band 4 repeated	11.5 - 12.5	11.5 - 12.5

AVHRR is a whiskbroom scanner. The pixel size at nadir is 1.1 km and the swath width 2399 km. Because of the scanning geometry, the pixel size increases with increasing scanning angle The quantisation of the pixel data is 10 bits.

AVHRR data is widely available because it can be received from the satellite using a fairly inexpensive system. Besides meteorological purposes, AVHRR data has been used in land and water applications. The continuity of AVHRR data is very good. Satellites with AVHRR have been orbiting from 1978 and the next generation of polar-orbiting weather satellites (NOAA satellites and METOP satellites) will be carrying a new version of AVHRR.

2.2.2 MODIS

The US MODIS instrument was launched on-board the Terra satellite in 1999. This satellite is a research satellite. MODIS is a very versatile instrument collecting data in 36 spectral bands (seeTable 26). The ground pixel size at nadir varies from 250 meters to 1 km depending on the channel. The swath width is 2230 km. The instrument is a whiskbroom scanner and the pixel size increases off-nadir (factor 2 along track, 5 across track). The quantisation of the data is 12 bits.URL http://modis.gsfc.nasa.gov/

Table 26 The MODIS spectral channels

band	wavelength	band	wavelength	pixel
1	620 - 670	2	841 - 876	250 m
3	450 - 470	4	545 - 565	500 m
5	1230 - 1250	6	1628 - 1652
7	2105 - 2155
8	405 - 420	9	438 - 448	1000 m
10	483 - 493	11	526 - 536
12	546 - 556	13	662 - 672
14	673 - 683	15	743 - 753
16	862 - 877	17	890 - 920
18	931 - 941	19	915 - 965
20	3660 - 3840	21	3929 - 3989
22	3929 - 3989	23	4020 - 4080
24	4433 - 4498	25	4482 - 4549
26	1360 - 1390	27	6535 - 6895
28	7175 - 7475	29	8400 - 8700
30	9580 - 9880	31	10780 - 11280
32	11770 - 12270	33	13185 - 13485
34	13485 - 13785	35	13735 - 14085
36	14085 - 14385

The most interesting channels for forest applications are 1 - 7. MODIS includes channels that can be used to estimate the characteristics of the atmosphere. These allow direct atmospheric correction of the MODIS data. The MODIS ground segment defines 44 different data products ranging from radiance counts to atmospherically corrected radiance to products like the leaf area index (LAI) and even land cover type (17 IGBP classes).

The MODIS design life is six years. Another MODIS instrument has been launched on the Aqua spacecraft in May 2002. The future of the MODIS instruments beyond this is not defined yet. Continuation for the MODIS type of instruments is planned in the NPP mission in 2005.

The MODIS data costs only the delivery costs if acquired from USGS or NASA. Other receiving entities may charge a price covering the costs.

2.2.3 MERIS

One of the instruments on-board the ESA Envisat research satellite is the MERIS imaging spectrometer. Envisat was launched in 2002. In principle, MERIS has 520 1.25 nm wide channels between 390 and 1040 nm. All of this data can’t be downlinked to earth. The instrument is programmable and the set of channels to be downlinked can be selected. It has been decided that in normal operation MERIS will use the channel selection shown in Table 27. Other configurations may be used experimentally for short times. URL http://envisat.esa.int/instruments/meris/

Table 27 The 15 MERIS standard spectral channels.

Channel	Band centre nm	Bandwidth Nm	Potential applications
1	412.5	10	Yellow substance and detrital pigments
2	442.5	10	Chlorophyll absorption maximum
3	490	10	Chlorophyll and other pigments
4	510	10	Suspended sediment, red tides
5	560	10	Chlorophyll absorption minimum
6	620	10	Suspended sediment
7	665	10	Chlorophyll absorption and fluo. reference
8	681.25	7.5	Chlorophyll fluorescence peak
9	708.75	10	Fluo. Reference, atmospheric corrections
10	753.75	7.5	Vegetation, cloud
11	760.625	3.75	Oxygen absorption R-branch
12	778.75	15	Atmosphere corrections
13	865	20	Vegetation, water vapour reference
14	885	10	Atmosphere corrections
15	900	10	Water vapour, land

MERIS is a pushbroom scanner with pixel size 300 m (full resolution mode) or 1200 m (reduced resolution mode). (The pixel size in a pushbroom scanner is the same across the field of view if the curvature of the earth is ignored.) The swath width is 1150 km.

The ERS/Envisat data policy divides the users into two categories. Category 1 includes research and long term monitoring programmes. For this category, the data price will be set at or near reproduction cost. Other users belong to Category 2 where pricing is set by the distributors but ESA reserves the right to set a maximum price.

2.2.4 WIFS

The WIFS image on-board Indian IRS-1C/D collects data from two spectral channels (see Table 31) with 188 m pixel size. The imager consists of two cameras for each wavelength. The cameras are tilted 26 degrees. The swath width is 728 - 818 km. The cameras use linear detectors and this is why the pixel size changes more slowly than in a whiskbroom scanner. Quantisation is 7 bits. URL http://www.euromap.de/

The WIFS data is sold commercially. The price is about USD 900 for each image (774 by 774 km).

2.2.5 MOS-1/1b

The Japanese MOS-1 satellite was launched in 1987 and the MOS-1b satellite in 1990. Operation of MOS-1 finished in 1995 and operation of MOS-1b finished in 1996. Both satellites carried a VTIR imager. It collected data on four spectral channels (0.5-0.7 µm, 6.0-7.0 µm, 10.5-11.5 µm, 11.5-12.5 µm). The pixel size at nadir was 900 m in the visible channel and 2700 m in the infrared channels. The swath width was 1500 m. URL http://www.nasda.go.jp/sat/mos1/index_e.html

2.2.6 The Spot Vegetation Instrument

In addition to the high-resolution imagers, the Spot-4 satellite (see Section 2.3.2) is carrying a low-resolution imager, the Vegetation Instrument. Spot-5 is carrying a similar instrument, named Vegetation 2. The Vegetation Programme is developed jointly by France, the European Commission, Belgium, Italy and Sweden. The instruments collect data in four spectral channels (0.43-0.47, 0.61-0.68, 0.78-0.89, and 1.58-1.75 µm). The pixel size of these whiskbroom scanners is 1 km and swath width 2250 km.

The Vegetation data is available either as a zone extracted from a single data strip or a zone extracted from a synthesis of data strips from one day or from a 10 day period. The pricing is according to the area. A typical price for a 4 million km² of data is between USD 150 and USD 350.

2.2.7 Future low- and medium-resolution instruments

ADEOS-II is planned to be launched in 2002. It carries the GLI imager with 36 spectral channels. The pixel size is 1 km at nadir (250 m on six channels corresponding to the Landsat TM channels) and swath width is 1600 km. Quantisation is 12 bits. URL http://www.nasda.go.jp/sat/adeos2/index_e.html

2.3 High-Resolution Optical Satellite Image Data (e.g., Landsat TM, ETM+, Spot, IRS-1)

The high-resolution optical sensors have pixel size in the range of tens of meters. The light from the sun reflected by the targets is collected using several wavelength channels. Some sensors include also one or more channels sensing the thermal radiation emitted by the targets. In addition to these limited wavelength channels, some satellites have one panchromatic (wide channel in the visible/near infrared region) channel, which may have smaller pixel size than the narrower channels.

2.3.1 Landsat

The best known remote sensing satellites may be the Landsat satellites from the USA. The currently operating Landsat satellites have the numbers 5 and 7. Landsat 5 was launched in 1984 and it is still operating. Landsat 7 was launched in 1999 and it is currently collecting most of the Landsat images. A list of all the Landsat satellites up to now is in Table 27. The early Landsat satellites carried a multispectral scanner called MSS (Multi-Spectral Scanner). Landsat 4 introduced an improved sensor, called Thematic Mapper and is best known by the acronym TM. This is the main instrument on-board Landsat 5. Landsat 7 carries an improved version of this instrument, called Enhanced Thematic Mapper, ETM+. Both of these sensors collect data from seven spectral channels. In addition to the multispectral channels, ETM+ collects data also from one panchromatic channel with nominal pixel size of 15 meters. The wavelength ranges of the sensor channels are summarised in Table 28. URL http://landsat7.usgs.gov/, http://landsat.gsfc.nasa.gov/

Table 27 The Landsat missions.

Table 28 The Landsat MSS, TM, and ETM+ instrument chracteristics.

MSS		TM		ETM+
Micrometers	Resolution	Micrometers	Resolution	Micrometers	Resolution
0.5 - 0.6	80 m	0.45 - 0.53	30 m	0.450 - 0.515	30 m
0.6 - 0.7	80 m	0.52 - 0.60	30 m	0.525 - 0.605	30 m
0.7 - 0.8	80 m	0.63 - 0.69	30 m	0.630 - 0.690	30 m
0.8 - 1.1	80 m	0.76 - 0.90	30 m	0.750 - 0.900	30 m
10.41 - 12.6	237 m	1.55 - 1.75	30 m	1.55 - 1.75	30 m
		10.40 - 12.50	120 m	10.40 - 12.50	60 m
		2.08 - 2.35	30 m	2.09 - 2.35	30 m
				0.52 - 0.90	15 m

The pixel data is digitised using eight bits. This limits the dynamic range and may be a problem in visible wavelength channels with forest targets.

The nominal width on ground of the Landsat images is about 185 km. The orbit of the Landsat satellites permit imaging of the same size every 16 days. The orbits of Landsat 5 and Landsat 7 are interleaved so that the same location can be imaged every 8 days while Landsat 5 operation continues. Note that the availability of useful images is limited by the clouds.

The Landsat data has several advantages. The data has been collected using nearly similar sensors for a long time. The price of the images is currently between USD 500 and USD 2000, depending on the location and receiving station.

New satellites in the Landsat series are in early phase of design. The sensors will have channels matching the TM and ETM channels to maintain data continuity. In addition to this, some new channels may be added. The next Landsat launch has been planned for year 2006. (URL http: //ldcm.usgs.gov). An experimental Landsat-like sensor ALI is currently orbiting on the EO-1 satellite. URL http: //eo1.gsfc.nasa.gov/

2.3.2 Spot

The French SPOT satellites have slightly different characteristics than the Landsat satellites. The pixel sizes in the HRV sensors on the satellites are smaller than in the TM/ETM sensors but the number of spectral channels is smaller and the image size is smaller.

Table 29 shows the basic characteristics of the HRV instruments on-board Spot 1 – 4. URL http://www.spotimage.fr/

Table 29 The Spot sensor characteristics.

	Multispectral	Panchromatic
Spectral bands (nm)	500 - 590
	610 - 680
	790 - 890	510 - 730
	1580 - 1750 (Spot 4)	610 - 680 (Spot 4)
Pixel size	20 x 20 m	10 x 10 m
Nbr pixels per line	3000	6000

The width of the Spot images is 60 - 80 km. The orbit of the satellite passes over the same location every 26 days. The sensor is pointable (+/- 27 degrees) and this allows imaging of the same target each 2.4 days. The pointable sensor allows also stereo imaging.

The currently orbiting satellites in the Spot series are 2, 3, 4 and 5. The latest, Spot-5 was launched in May 2002. The sensor on this satellite has pixel size of 10 m in the visible and near-infrared channels and 5 or 2.5 m in the panchromatic channel. The numbers of pixels are increased so that the image width of about 60 km is maintained. The satellite can acquire two images simultaneously using along-track pointing.

The cost of the Spot images is considerably higher than the cost of the Landsat images. A typical price for a 60 by 60 km scene is between USD 2800 - 5500 (archived images may be available for lower price).

2.3.3 IRS

India has launched up to now five IRS-1 remote sensing satellites (1A in 1988, 1B in 1991, 1E in 1993, 1C in 1995, and 1D in 1997). The satellites 1b, 1C and 1D are currently producing data (IRS-1B data not available worldwide). The instruments on board IRS-1C/D include the multispectral scanner LISS-III and the panchromatic imager PAN. The characteristics of the instruments are summarised in Table 31. URL http: //www.euromap.de/

The 7-bit quantisation in the multispectral data limits the radiometric resolution of the images

Table 31 The IRS-1C/D instrument characteristics.

	PAN		LISS-III		WIFS
Spectral channels	500 - 750 nm 5.8 m		520-590 nm	23 m
			620-680 nm	23 m	620-680 nm	188 m
			770-860 nm	23 m	770-860 nm	188 m
			1550-1700 nm	70 m
Swath width	70 km		142 km		810 km
Quantisation	6 bit		7 bit		7 bit

The next satellite in the series carrying multispectral instruments will be IRS-P6 (Resourcesat-1). Its launch is planned for year 2002 or 2003. It is carrying improved versions of the PAN and LISS instruments. The characteristics of these new instruments are shown in Table 32. The satellite carries also a LISS-III scanner in addition to these new instruments.

Table 32 The IRS-P6 instrument characteristics.

		LISS-	IV		AWIFS
	PAN		MSS
Spectral channels	500 - 750 nm 5.8 m		520-590 nm	5.8 m	520-590 nm	60-70m
			620-680 nm	5.8 m	620-680 nm	60-70 m
			770-860 nm	5.8 m	770-860 nm	60-70 m
					1550-1700 nm	60-70 m
Swath width	70 km		23.9 km		700 km
Quantisation	7 bit		7 bit		10 bit

The price of one IRS-1C/D LISS-III or PAN image is about USD 2500.

2.3.4 Other High-Resolution Optical Satellites

There are several other past, present, or future satellites carrying high-resolution optical instruments. The data from these satellites can be used to complement/supplement data from the satellites mentioned above. These satellites are either experimental, or belong to short series of satellites. The availability of images from these satellites is limited, especially on a long-term basis, and this is why these satellites are not very useful for global forest monitoring.

The Russian Resurs series currently includes two satellites with high-resolution imagers, Resurs O1-3 and Resurs O1-4. The satellites were launched in 1994 and 1998. Both carry a three-channel (0.5 - 0.67 µm, 0.65 - 0.8 µm, and 0.8 - 1.0 µm) pushbroom scanner with 29 - 45 meter pixel size across track. The swath widths are 45 km and 58 km, respectively (the scanners are similar except for the number of elements). The orbit heights are different. URL http://www.scanex.ru/stations/resurs.htm

The first Japanese earth observation satellites were MOS-1 and MOS-1b. The satellites were launched in 1987 and 1990 and operated until 1995 and 1996, respectively. One of the instruments on-board these satellites was MESSR. It consisted of four modules, two for dual channel (0.51 - 0.59 µm and 0.61 - 0.69 µm) visible range monitoring and two for dual channel (0.72 - 0.80 µm and 0.80 - 1.0 µm) near-infrared region. The pixel size was 50 meters and maximum swath width 185 km. URL http://www.nasda.go.jp/sat/mos1/index_e.html

The Japanese JERS-1 satellite carried the optical OPS sensor from 1992 to 1998. It included three visible channels, four shortwave infrared bands, and one band for stereoscopic viewing. The pixels size was 18 meters and swath width 75 meters. URL http://www.nasda.go.jp/sat/jers1/index_e.html

The Japanese ADEOS satellite carried the AVNIR high-resolution imager. Unfortunately the satellite operated only 10 months. URL http://www.nasda.go.jp/sat/adeos/index_e.html

The launch of the Japanese ALOS satellite is planned for 2004. One of the ALOS instruments is AVNIR-2. It is a multispectral imager with four channels (0.42 - 0.50 µm, 0.52 - 0.60 µm, 0.61 - 0.69 µm, 0.76 - 0.89 µm). The pixel size at nadir will be 10 meters and the swath width will be 70 km. The imager can be pointed up to 44 degrees off nadir. URL http://www.nasda.go.jp/sat/alos/index_e.html

2.4 Very High Resolution Satellite Image Data (e.g., Ikonos-1, QuickBird)

The very-high resolution instruments have pixel size around one meter. This resolution has been introduced to the non-military users only a few years ago. Most of these satellites have been designed to complement and/or to replace aerial photography in many applications. The satellites operate usually on a commercial basis. All of the satellites include a panchromatic band with high spatial resolution. Many satellites also include multispectral capacity with slightly larger pixel size than is used in the panchromatic channel.

A typical swath width of the high-resolution satellites is about 10 kilometres. This is limited basically by the downlink capacity. The satellites include pointing capability and this enables obtaining imagery from every location on earth within a reasonable time. The area covered by the satellites is constrained by the small image size.

The pricing of the very-high resolution data can be expected to change according to supply and demand.

The characteristics of the currently orbiting and some planned high-resolution satellites are summarised below.

2.4.1 Ikonos-2

Ikonos-2 (Space Imaging) was the first operational high-resolution satellite with pixel size in the one meter region. It was launched in 1999. The imager on board the satellite can collect data with one panchromatic channel (450-900 nm) and four multispectral channels (445-516 nm, 506-595 nm, 632-698 nm, 757-853 nm). The pixel size on the panchromatic channel is 1 meter and on the multispectral channels 4 meters. The swath width is 13 km and the sensor is pointable both on-track and across-track. The quantisation is 11 bits. URL http://www.spaceimaging.com/

The Ikonos-2 data is currently priced per square kilometre. The minimum order is 100 km². The price varies between USD 7 and USD 235 per square km depending on the geometric processing, location, included channels, and acquisition type (archive, on-demand).

2.4.2 QuickBird

The second operational one-meter satellite is QuickBird 2 (DigitalGlobe). It was launched in 2001. The imager on board this satellite can also collect data with one panchromatic channel (450-900 nm) and four multispectral channels (450-520 nm, 520-600 nm, 630-690 nm, 760-900 nm). The pixel size on the panchromatic channel at nadir is 61 cm and on the multispectral channels 2.44 meters. The nominal swath width is 16.5 km (nadir pointing). Also this instrument is pointable both along-track and across-track direction. The quantisation of the data is 11 bits. URL http://www.digitalglobe.com/

The QuickBird imagery is currently sold as single frames (no geometric correction) or data covering a defined area. The basic price of a single image is USD 6000-8000 and the price of imagery with standard geometric processing is USD 22.5-30 per square km. Additional costs are incurred for different licensing and acquisition options.

2.4.3 Eros-A1

The first satellite (A1) in the EROS series (ImageSat International) was launched in year 2000. The series is planned to consist of six satellites. The remaining five satellites will be of the B series that have slightly different specifications. Launch of EROS-B1 is planned for year 2003 and the constellation will be completed in 2005. URL http://www.imagesatintl.com/

The EROS-A1 contains a panchromatic imager (0.5-0.9 µm) that can collect data using either 1.8 meter (“normal”) or 1.0 meter (“over-sampled”) pixel size. The swath width is 12.5 km and quantisation 11 bits.

The B series satellites will carry an improved panchromatic imager with pixel size 0.82 meters and swath width 16 km. The quantisation is 10 bits. These satellites can also provide multispectral imagery with four spectral channels (480-520 nm, 540-580 nm, 640-680 nm, 820-900nm). Pixel size in the multispectral images is 3.28 m.

The satellites are pointable both in the along-track and the across-track direction up to 45 degrees. The visibility of receiving stations limits the availability of images worldwide.

The current price of one EROS-A1 scene is USD 1500-2500, depending on the geometric processing. Stereo pairs without geometric correction are sold for USD 3000.

2.4.4 SPOT-5

The SPOT-5 satellite is able to produce 2.5 meter effective pixel side with the so called Supermode. It uses two overlapping sensor arrays with 5 meter pixel size. The resulting image is computed from these two images. The swath width is 60 km. URL http://www.spotimage.fr/

2.4.5 Future systems

The high-resolution satellites have been made using low-cost technology and procedures. This may be the reason why several launch attempts have failed. The demand for very-high resolution data seems to be proven. This ensures that new satellites will be launched. However, whether a certain planned satellite will become operational is not certain.

Space Imaging is planning a successor for Ikonos but no specifications have been released. The same applies to DigitalGlobe and QuickBird.

The Orbview-3 satellite will carry a panchromatic imager (450-900nm) with 1 m pixel size and a multispectral imager with 4 channels (450-520 nm, 520-600 nm, 625-695 nm, 760-900 nm) and 4 m pixel size. The swath width will be 8 km and the imagers will be pointable across-track up to 45 degrees. The pixel data quantization is 11 bits. The launch is planned for late 2002. URL http://www.orbimage.com/

The Indian IRS-P5 (Cartosat-1) to be launched in 2003 - 2004. It will be carrying two panchromatic (500 - 750 nm) cameras, one pointing 26 degrees forward and one pointing 5 degrees backwards. These enable stereo imaging from a single orbit. The pixels size will be 2.5 meters and swath width 30 km. Quantisation is 10 bits. URL http://www.euromap.de/

The Japanese ALOS satellite will include the PRISM imager. It is able to collect simultaneously three panchromatic (520 - 700 nm) images with 2.5 meter pixel size, one 24 degrees forward, one nadir, and one 24 degrees backwards. The swath width is 35 km when all three imagers are used and 70 km for nadir only viewing. URL http://www.nasda.go.jp/sat/alos/index_e.html

2.5 SAR Data

SAR is an acronym for Synthetic Aperture Radar. It is an active microwave imager. SAR uses a side-looking antenna to achieve good resolution across track with high-enough pulse power levels. The resolution along track is sacrificed but it is enhanced by computer processing afterwards.

The parameters relevant to SARs include the wavelength range (band), transmitted and received polarisation, incidence angle, and pixel size on ground. In the microwave region the wavelength ranges have traditionally been denoted by letters. The definitions of the bands can be seen in

Table 33 (only some of these bands are currently used in imaging radars).

Table 33 The radar band names and frequencies.

Name	Nominal frequency range	Wavelength range	Specific bands used in radars
VHF	30-300 MHz	10-1 m	138-144 MHz 216-225 MHz
P (UHF)	300-1000MHz	100-30 cm	420-450 MHz 890-942 MHz
L	1-2 GHz	30-15 cm	1.215-1.4 GHz
S	2-4 GHz	15-7.5 cm	2.3-2.5 GHz 2.7-3.7 GHz
C	4-8 GHz	7.5-3.75 cm	5.25-5.925 GHz
X	8-12 GHz	3.75-2.5 cm	8.5-10.68 GHz
Ku	12-18 GHz	2.5-1.67 cm	13.4-14.0 GHz 15.7-17.7 GHz
K	18-27 GHz	1.67-1.11 cm	24.05-24.25 GHz
Ka	27-40 GHz	1.11-0.75 cm	33.4-36.0 GHz
V	40-75 GHz	0.75-0.40 cm	59-64 GHz
W	75-110 GHz	0.40-0.27 cm	76-81 GHz 92-100 GHz
millimeter	110-300 GHz	2.7-1.0 mm

The radar is measuring the fraction of the transmitted pulse that is received at each time. The time specifies the distance between the the target reflecting the pulse and the receiving antenna. In this way the reflection (backscatter) capabilities of different targets can be measured. In addition to the target itself, the backscattered power depends on the wavelength, the geometry of the backscattering, and the polarisations (transmitted pulse, receiver). Some combinations of these auxiliary parameters are better with some targets than others. In this way different aspects of the forest can be measured using different parameters. Best results can be obtained if several parameter combinations can be used at the same time (e.g., different polarisation combinations in a polarimetric SAR).

There are a few spaceborne SARs orbiting at the moment. These operate either in the C band or in the L band and use a limited set of polarisation combinations or incident angles. The new SARs being launched in the near future widen the choice of the polarisations and imaging geometries. Unfortunately, the bands remain the same although SARs with longer wavelength would be better for forest analysis.

The SAR processing is coherent. This can be utilised in the interferometric SAR processing with two measurements from slightly different orbits. The interferometric processing allows determination of the 3-dimensional structure of the targets to some extent. A popular application is making Digital Elevation Models (DEM). The forest is a very complicated target and the elevation model of the forest can’t be determined by the interferometric processing. However, the coherence of the responses from forest can be measured and this can be used as a feature differentiating, e.g., forest types. Coherence and interferometric SAR is currently a research topics and not suitable for operational forest analysis.

The advantage of SAR against optical instruments is that the microwaves propagate through clouds. Because of this, SAR images can be obtained at any time. The drawback is that the backscattering signal from forest is very noisy and does not allow detailed analysis of forests. SAR may be useful in separating forested and non-forested areas at some locations in the world.

The SAR images are strongly influenced by the moisture of the imaged targets. This can be used as an advantage by using several images from one area from different times. In many forest application this is a necessity and increases the number of images necessary for the application.

2.5.1 ERS-1/2

The ESA ERS-1 satellite was launched in 1991 and it is still operational. The ERS-2 satellite was launched in 1995 and it is also operational. Both include a C-band SAR (wavelength 5.6 cm) that can operate in either imaging or wave mode. In imaging mode the SAR collects images with 30 m resolution and 80 km swath width. The incidence angle is 20.1 - 25.9 degrees. The polarisation is VV (vertical polarisation in both transmitter and receiver). URL http://earth.esa.int/ers/

The data policy described with MERIS applies also to ERS-1/2 data.

2.5.2 JERS-1

The Japanese JERS-1 satellite is carrying, in addition to the high-resolution optical imager, a SAR. The satellite was operational between 1992 and 1998. The SAR operated in the L-band (wavelength 23 cm). The pixel size was 18 by 18 meters and swath width 75 km. The incidence angle was 36 - 41 degrees. The polarisation is HH. URL http://www.nasda.go.jp/sat/jers1/index_e.html

JERS-1 data is available on commercial terms.

2.5.3 Radarsat

The Canadian Radarsat-1 was the first satellite in the commercial series. It was launched in 1995 and is still operational. The SAR on-board this satellite operates in the C-band (wavelength 5.6 cm) with HH polarisation. It is a very versatile imager that can operate in several different imaging modes with different pixel size, swath width and incidence angle. In standard mode, one of seven incidence angle ranges (20-27.4, 24.2-31.2, 30.4-36.9, 33.6-39.6, 36.5-42.2, 41.4-46.5, 44.9-49.4 degrees) can be chosen. The pixel size is about 25 meters and swath width about 100 km. In wide swath mode, two incidence angle ranges are available. The pixel size is about 30 m and swath width 150 km. In fine-beam mode, five incidence angle ranges between (36.9-40.1, 39.3-42.3, 41.6-44.2, 43.6-46.0, 45.3-47.8 degrees). The pixel size 9 meters and swath width 50 km. The ScanSAR modes scan over two to four wide swath beams. The resolution is 50 or 100 meters and the swath width 300-500 km. In the extended high mode the incidence angle is between 50 and 60 degrees, pixel size 25 m and swath width 50 km. In the extended low mode the incidence angle is between 10 and 20 degrees, pixel size about 25 m and swath width 75 km. URL http://www.rsi.ca/

The cost of Radarsat data for single images is about 3000 USD per image. Significant volume discounts are available.

Launch of the Radarsat-2 satellite is planned for 2003. It will have imaging modes similar to Radarsat-2. Additional ultra-fine modes with pixel size 3 by 3 m and swath width 10-20 km have been added. The third satellite, Radarsat-3 is also being planned.

2.5.4 Envisat ASAR

One of the instrument on the ESA Envisat satellite is ASAR. It is a flexible SAR operating in the C-band. In addition to the different pixel sizes and incidence angles, ASAR can also use different polarisations. ASAR can be in one of the following operating modes:

§ Global monitoring mode: ScanSAR mode with 1 km pixel and 405 km swath, HH or VV polarisation.

§ Wave mode: 5 x 5 km vignettes each 100 km along track, HH or VV polarisation.

§ Image mode: 30 m pixel, HH or VV polarisation, other parameters in Table 35

§ Alternating polarisation mode: Image mode but polarisation alternating from subframe to subframe. Results in two images (HH/VV or HH/HV or VV/VH polarisation pairs)

§ Wide swath mode: ScanSAR, 405 km swath with 150 km pixel, HH or VV polarisation

URL http://envisat.esa.int/

Table 35 ASAR image mode products.

ASAR Swathes	Swath Width [km]	Near Range Incidence Angle	Far Range Incidence Angle
IS1	108.4 - 109.0	14.1 - 14.4	22.2 - 22.3
IS2	107.1 - 107.7	18.4 - 18.7	26.1 - 26.2
IS3	83.9 - 84.3	25.6 - 25.9	31.1 - 31.3
IS4	90.1 - 90.6	30.6 - 30.9	36.1 - 36.2
IS5	65.7 - 66.0	35.5 - 35.8	39.2 - 39.4
IS6	72.3 - 72.7	38.8 - 39.1	42.6 - 42.8
IS7	57.8 - 58.0	42.2 - 42.6	45.1 - 45.3

The Envisat data policy described previously with MERIS applies also to ASAR.

2.5.5 Future SAR satellites

The Japanese ALOS satellite (launch planned for 2004) will carry the PALSAR imager. It is operating in the L-band (wavelength 23 cm) and can be in one of three modes. In the high-resolution mode the pixel size is 10 m and swath width is 70 km. In high resolution mode it can collect data with 10-50 m pixel size and 40-70 km swath width. It can operate in either HH or VV polarisation mode or in combined HH+HV or VV+VH modes. The incidence angle can be selected between 8 and 60 degrees. In ScanSAR mode it can image 250-350 km swath with 100 m pixel size. The polarisation is either HH or VV and the incidence angle 8-43 degrees. Additionally this SAR has an experimental polarimetric mode with 24-89 m pixel size and 20-65 km swath width.

There are several other SAR satellites or satellite constellations under different stages of planning, including TerraSAR and Cosmo Skymed.

2.5.6 Airborne SAR

There exist several airborne SAR systems. Most of these are research systems, like the JPL AIRSAR, the German ESAR, the Danish EMISAR, and the Canadian CCRS C/X-SAR. There also exist at least commercial airborne SAR systems. One of these is Intermap Star-3i SAR (URL http://www.globalterrain.com/Technology.html). However, this is an X-band SAR and not very useful in forest applications.

2.6 Laser Scanners and LIDARs

Laser scanners and LIDARs are active instruments operating in the visible or infrared range. Narrow pulses are produced by a laser is sent towards a target and the received light energy and arrival time are measured. The time gives the distance between the instrument and the target whereas the intensity of the received light reflects the properties of the target (and the atmosphere). In many cases the instruments measure only the time of return of the pulses. In this case the instrument measures only the distance. There may be several returned pulses corresponding to one sent pulse in case the pulse energy is reflected by several targets. In forest applications this can mean reflections from the target and the ground, in which case the instrument can be used to measure the height of the trees. In addition to these reflections, there are reflections from other targets between the top of canopy and the ground.

Image data can be formed by scanning the laser and sensor over the target area while the platform is moving forward.

A laser scanner has been used in space on two Space Shuttle flights. The first mission, SLA-1 (Shuttle Laser Altimeter) was on STS-72 (74) 1996 and the second mission was SLA-2 on STS-85 in 1997. Some forest data from low latitudes was collected on these flights.

The first spaceborne LIDAR applicable to forest analysis will be the US Vegetation Canopy LIDAR (VCL). The launch of this satellite is planned for 2003. The inclination of the orbit is 67 degrees and this limits the data to latitudes below 67 degrees. The instrument uses three fixed lasers operating at 1064 nm. The ground tracks of the lasers are spaced by 4 km apart on the ground and the pulse spacing along track is nearly contiguous. The footprint of each laser is 25 m and the expected canopy top height and ground elevation accuracy is 1 m. URL http://www.geog.umd.edu/vcl/

There are a few airborne laser scanners operating in the world. These can be used to obtain accurate tree height data from limited areas. The pixel size of these instruments is about 1 m and the height accuracy is about 20 cm. Considering the inaccuracies in locating the tree tops and combining the errors, the accuracy of tree height measurements is about 0.5 m - 1m.

2.7 Imaging Spectrometer Data

An imaging spectrometer is an optical imager that can collect data simultaneously using a large number of narrow spectral channels. The spectral channels are contiguous or a subset of contiguous channels. The advantage of imaging spectrometers is that they can very accurately analyse the spectra of different targets. Whether this is useful or not, depends on the application. In forest applications the success has so far been moderate.

Most of the current imaging spectrometers are airborne instruments (e.g., AVIRIS, CASI, AISA, DAIS, ROSIS, MIVIS). The best known airborne imaging spectrometer is AVIRIS. It can collect 224 channels in the wavelength range 380-2500 nm (10 nm channels). It is flying at 20 km altitude with pixel size 20 m and 11 km swath or at 4 km altitude with pixel size 4 m and 2 km swath. URL http://makalu.jpl.nasa.gov/aviris.html

The CASI and AISA instruments are operating in the visible and infrared range (450-900 nm). Both are programmable and can be flown on small aircrafts with varying pixel size. URLs http://www.itres.com, http://www.specim.fi/

The first spaceborne, earth-viewing imaging spectrometer is Hyperion on the NASA EO-1 experimental satellite. It was launched in year 2000 and is still operating. The spectrometer can collect data with 220 spectral channels in the wavelength range 400-2500 nm. The pixel size is 30 m and swath width is 7.5 km. URL http://eo1.gsfc.nasa.gov/

The US NAVY NEMO (Naval Earth Map Observer) satellite is planned to be launched in 2002. One of its instruments is the COIS (Coastal Ocean Imaging Spectrometer) spectrometer. It collects data using 10 nm spectral channels in the wavelength range 400-2500 nm (210 bands). The pixel size is 30 or 60 m and the swath width is 30 km. The PIC (Panchromatic Imaging Camera) camera can be used to simultaneously image the same area with 5 m pixel size. NEMO will be downlinking spectral signature recognition data instead of raw data. URL http://nemo.nrl.navy.mil/

The plans for the Australian ARIES-1 (Australian Resource Information and Environment Satellite) satellite are not clear at the moment. It should be carrying an imaging spectrometer operating in the 400-2500 nm range and a panchromatic camera.

2.8 Aerial photography

The most important airborne data source is aerial photography in various forms. It is being commercially used all over the world and provides spatially accurate images over limited areas. In some countries, even countrywide aerial photograph archives exist or are being planned.

Aerial photography can be divided into classes based on the film type, flight altitude, resolution, and image format.

The film type is usually either panchromatic, colour, or false colour. In the false colour film, one of the layers is sensitive to red light, one to green light, and one to near infrared light. The pigments have been arranged so that the red light is shown as green, green light as blue, and infrared as red. The vegetation reflects infrared light and the red areas in the false-colour images correspond to vegetation.

The aerial photographs on film are usable for visual analysis but not for any automatic analysis. The photographs can be transformed into digital form by scanning. Currently there exist high-quality scanners that can be used operationally. The tendency is to move to digitised aerial photographs also for visual use because the geometric processing and archiving are easier.

There are some completely digital aerial photography cameras nowadays. However, the pressure to develop digital cameras is not high because the digitised analog photographs work well enough. The completely digital systems provide better geometric stability and they can be calibrated radiometrically (an advantage in numerical analysis of the images).

The area covered by the photographs and the smallest usable details depend on the size of the film, the optical system of the camera, and the flight altitude. Usually the aerial photographs are characterised by the scale. When the digitisation pixel size is known, this enables computation of the pixel size on ground. For instance, assuming scale 1: 30000 and pixel size 20 µm (typical values), we get the pixel size 0.6 m.

With a fixed scale, the film size, the area being digitised, and the desired overlap (30-60 %) determine the number of images needed. The typical film width of the metric cameras used in aerial photography is 240 mm (230 by 230 mm image area). With 1: 30000 scale this gives the image size 6.9 by 6.9 km on ground. Narrower film (70 mm or 35 mm) or video cameras are used in detailed photography but these formats are less suitable for covering larger areas.

The price of digital photography varies according to the imaging parameters and supplier. A typical price level for a set of high-altitude aerial photographs mosaiced to a geocoded orthophoto is about USD 0.15 per hectare. However, the price varies widely according to the scale and whether existing imagery can be used or new flights are required.

3. Covering the Whole Earth with Images

The land area of the world (excluding Antarctica) is about 135 000 000 km². If we assume that there is 10 % overlap between the images, the images should cover an area of 148 500 000 km². Rough estimates of the numbers of images needed to cover the land areas of the Earth are given in Table 37. The image height is assumed to be equal to the image width although in practice this does not always apply. The height of the images from lower resolution satellites often is limited only by the receiving station visibility. The prices are computed according to the current price lists for single images. AVHRR and MODIS data can be obtained by paying the handling costs and the total cost is not computed for these data sources.

Table 37 The number of images needed to cover the land area of the Earth for different satelliteinstruments. The amount of data per channel is computed assuming rectification to the nadir pixel size. The number of channels is the maximum number of channels usable in forest applications.

Satellite	Nbr. images	Gigabytes per channel	Channels	Price
AVHRR	26	0.25	5	N.A.
MODIS	30	0.30 - 1.2	7 (36)	N.A.
Landsat TM/ETM+	4 339	165 - 660	7 (8)	2.2 - 8.7 Million USD
Spot HRV	41 250	370 - 1 480	4/5	100 - 200 Million USD
IRS LISS-III	7 365	280	4	20 Million USD
QuickBird	545 455	25 000 - 100 000	4	3 800 Million USD

Table 37 shows that it is feasible to cover the whole earth with low-resolution data (AVHRR, MODIS) but using medium-resolution data is questionable. The amount of data is within limits of the processing capabilities of reasonable computers. The price is not reasonable but a mosaic made for research purposes by some organisation may be available for a fraction of this price. The high-resolution data is clearly usable only for local sampling.

FRA Working Papers

0.	How to write a FRA Working Paper (10 pp. – E)
1.	FRA 2000 Terms and Definitions (18 pp. - E/F/S/P)
2.	FRA 2000 Guidelines for assessments in tropical and sub-tropical countries (43 pp. - E/F/S/P)
3.	The status of the forest resources assessment in the South-Asian sub-region and the country capacity building needs. Proceedings of the GCP/RAS/162/JPN regional workshop held in Dehradun, India, 8-12 June 1998 (186 pp. - E)
4.	Volume/Biomass Special Study: georeferenced forest volume data for Latin America (93 pp. - E)
5.	Volume/Biomass Special Study: georeferenced forest volume data for Asia and Tropical Oceania (102 pp. - E)
6.	Country Maps for the Forestry Department website (21 pp. - E)
7.	Forest Resources Information System (FORIS) – Concepts and Status Report (20 pp. E)
8.	Remote Sensing and Forest Monitoring in FRA 2000 and beyond (22 pp. - E)
9.	Volume/Biomass special Study: Georeferenced Forest Volume Data for Tropical Africa (97 pp. – E)
10.	Memorias del Taller sobre el Programa de Evaluación de los Recursos Forestales en once Países Latinoamericanos (194 pp. - S)
11.	Non-wood forest Products study for Mexico, Cuba and South America (draft for comments) (82 pp. – E)
12.	Annotated bibliography on Forest cover change – Nepal (59 pp. – E)
13.	Annotated bibliography on Forest cover change – Guatemala (66 pp. – E)
14.	Forest Resources of Bhutan - Country Report (80 pp. – E)
15.	Forest Resources of Bangladesh – Country Report (93 pp. – E)
16.	Forest Resources of Nepal – Country Report (78 pp. - E)
17.	Forest Resources of Sri Lanka – Country Report (77 pp. – E)
18.	Forest plantation resource in developing countries (75 pp. – E)
19.	Global forest cover map (14 pp. – E)
20.	A concept and strategy for ecological zoning for the global FRA 2000 (23 pp. – E)
21.	Planning and information needs assessment for forest fires component (32 pp. – E)
22.	Evaluación de los productos forestales no madereros en América Central (102 pp. – S)
23.	Forest resources documentation, archiving and research for the Global FRA 2000 (77 pp. – E)
24.	Maintenance of Country Texts on the FAO Forestry Department Website (25 pp. – E)
25.	Field documentation of forest cover changes for the Global FRA 2000 (40 pp. – E)
26.	FRA 2000 Global Ecological Zones Mapping Workshop Report Cambridge, 28-30 July 1999 (53 pp. –E)
27.	Tropical Deforestation Literature:Geographical and Historical Patterns in the Availability of Information and the Analysis of Causes (17 pp. – E)
28.	Global Forest Survey – Concept Paper (30 pp. - E)
29.	Forest cover mapping and monitoring with NOAA-AVHRR and other coarse spatial resolution sensors (42 pp. - E)
30.	Web Page Editorial Guidelines (22 pp. – E)
31.	Assessing state & change in Global Forest Cover: 2000 and beyond (15 pp. – E)
32.	Rationale & methodology for Global Forest Survey (60 pp. – E)
33.	On definitions of forest and forest change (13 pp.- E)
34.	Bibliografía comentada. Cambios en la cobertura forestal: Nicaragua (51 pp. – S)
35.	Bibliografía comentada. Cambios en la cobertura forestal: México (35 pp. – S)
36.	Bibliografía comentada. Cambios en la cobertura forestal: Costa Rica (55 pp. – S)
37.	Bibliografía comentada. Cambios en la cobertura forestal: El Salvador (35 pp. – S)
38.	Bibliografia comentada. Cambios en la cobertura forestal: Ecuador (47 pp. – S)
39.	Bibliografia comentada. Cambios en la cobertura forestal: Venezuela (32 pp. – S)
40.	Annotated bibliography. Forest Cover Change: Belize (36 pp. – E)
41.	Bibliografia comentada. Cambios en la cobertura forestal: Panamà (32 pp. – S)
42.	Proceedings of the FAO Expert consultation to review the FRA 2000 methodology for Regional and Global Forest Change Assessment (54 pp. – E)
43	Bibliografia comentada. Cambios en la cobertura forestal: Colombia (32 pp. – E)
44	Bibliografia comentada. Cambios en la cobertura forestal: Honduras (42 pp. – E)
45	Proceedings of the South-Asian Regional Workshop on Planning, Database & Networking for Sustainable Forest Management (254 pp. – E)
46	Missing
47	Proceedings of the Regional workshop on forestry information services, Stellenbosch, South Africa, 12-17 feb. 2001
48 Forest cover assessment in the Argentinean Regions of Monte and Espinal (25 pp. – E) 49 Under preparation
50	Global Forest Cover Mapping – Final Report (29 pp. – E)
51 FAO Workshop Data Collection for the Pacific Region, 4-8 September 2000, Apia, Samoa (370 pp. - E)
52 53 54 55	Under preparation Forest occurring species of conservation concern: Review of status of information for FRA 2000 (E) Assessing Forest Integrity and Naturalness in Relation to Biodiversity(64 pp.) Global Forest Fire Assessment 1990-2000 (495 pp. – E)

56 Global Ecological Zoning for the Global FRA 2000 – Final Report (210 pp. – E)

57 Missing

58 Missing

59 Comparison of forest area and forest area change estimates derived from FRA 1990 and FRA 2000
(70 pp. - E)

60 On sampling for estimating global tropical deforestation (12 pp. - E)

61 The role of remote sensing in global forest assessment (73 pp. - E)

Please send a message to [email protected]  for electronic copies or download from www.fao.org/forestry/fo/fra/index.jsp  (under Publications)