GeoComp-n

1
GeoComp-n

• GeoComp-n

Natural Resources Canada
Gunar Fedosejevs
Canada Centre for Remote Sensing
Natural Resources Canada
Ressources naturelles Canada
2
GeoComp - n, an advanced system for generating
products from coarse and medium resolution
optical satellite data. Part 1 System
characterisation
3
GeoComp nPart 1 System characterisation

• Canada Centre for Remote Sensing (CCRS) Geocoding
  and Compositing system (GeoComp) has been
  processing the Advanced Very High Resolution
  Radiometer (AVHRR) data from the United States
  National Oceanic and Atmospheric Administration
  (NOAA) series of satellites since 1992.

4
GeoComp nPart 1 System characterisation

• Original GeoComp system built on Digital VAX
  platform by MacDonald Detwiller and Associates.
• Reference
• Robertson, B., A. Erickson, J. Friedel, B.
  Guindon, T. Fisher, R. Brown, P. Teillet, M.
  D'Iorio, J. Cihlar, and A. Sancz. 1992. GeoComp,
  a NOAA AVHRR geocoding and compositing system,
  Proceedings of the ISPRS Conference, Commission
  2, pp. 223-228, August 1992, Washington, D.C..

5
GeoComp nPart 1 System characterisation

• GeoComp-n (next generation) includes a modular
  system architecture, a fully functional operator
  GUI and a revamped data product format.
• GeoComp-n was built by PCI Geomatics of Richmond
  Hill, Ontario and was delivered to the Manitoba
  Centre for Remote Sensing in 1999.

6
GeoComp nPart 1 System characterisation

• GeoComp-n supports four primary functions
• AVHRR data input
• pre-processing
• geocoding and resampling
• composite product generation

7
GeoComp nPart 1 System characterisation

• GeoComp-n employs plain text auxiliary databases
  for system operation and product generation.
• The radiometric calibration coefficient auxiliary
  file must be updated annually as it contains both
  time- and satellite-dependant parameters.  
• The product coefficient auxiliary file contains
  product processing parameters and scaling
  coefficients. 
• Other databases include the image chip database,
  digital elevation model, land cover map of
  Canada, and seasonal NDVI database for cloud
  removal.

8
GeoComp nPart 1 System characterisation

• Composite products are generated as a flat raster
  image file with an associated product metadata
  file describing in plain-text the algorithm and
  processing parameters used.
• GeoComp-n can generate browse imagery in JPEF
  format and catalogue update files (CUFs) of the
  composite products for the CCRS Earth Observation
  Catalogue (CEOCat) database.
• The browse images consist of three
  operator-specified channels of the 1 km
  resolution data at an operator-specified reduced
  spatial resolution.

9
GeoComp nPart 1 System characterisation

• Input data consists of High Resolution Picture
  Transmission (HRPT) AVHRR data with a nominal
  ground resolution of 1.1 km at nadir in CEOS
  format as produced by the CCRS NOAA AVHRR
  Transcription and Archive System (NATAS) at the
  Prince Albert Satellite Station (PASS) or Level
  1B (L1B) data produced by any number of
  commercial AVHRR receiving systems.
• The data can be read from Exabyte tape, CD-ROM or
  downloaded over the network to a local disk
  system.

10
GeoComp nPart 1 System characterisation

• Data Pre-processing
• Replace the missing lines for L1B data from the
  NOAA Selective Active Archive (SAA).
•  Noisy lines can be detected using header
  information supplied with the raw image files
  and/or automatically using simple heuristics.
• Line replacement algorithm can replace a noisy
  scan line with the line below or above, or by the
  average of the two if the scan line is not at the
  beginning or end of the file.
• Noisy lines are recorded in the Quality Control
  (QC) bit mask as noisy pixels/lines (0bad or
  missing pixel, 1clear or good pixel).

11
GeoComp nPart 1 System characterisation

• Data Pre-processing (continued)
• A cloud detection algorithm using raw count
  thresholds identifies cloudy pixels for the QC
  mask.  
• The raw data are calibrated to top-of-atmosphere
  (TOA) radiance using onboard calibration data for
  the thermal channels or calibration parameters
  provided in the radiometric calibration
  coefficient file for the visible and
  near-infrared channels.
• The output of the pre-processing is a PCIDSK
  format file (16-bit unsigned integer) containing
  the scaled raw or calibrated radiance data for
  AVHRR channels 1 to 5.  
• The pre-processed file contains a fully
  determined orbit model.

12
GeoComp nPart 1 System characterisation

• GeoCoding and Resampling
• An orbit model refinement process is employed in
  geocoding where the satellite position and
  attitude are modelled using TBUS ephemeris
  information and refined using GCPs that are
  automatically matched against an image chip
  database using image correlation.
• The absolute 2-dimensional residual error for 1
  km resolution geocoded products shall be 0.55 /-
  0.07 RMSE of the IFOV, or better that is, 95
  of geocoded AVHRR products shall have a measured
  absolute 2-dimensional error of 750 m or less.

13
GeoComp nPart 1 System characterisation

• GeoCoding and Resampling (continued)
• While radiometric resampling may employ a variety
  of algorithms the damped Kaiser method is
  typically applied for maximum radiometric
  fidelity and geometric accuracy.  
• While GeoComp-n supports a wide selection of map
  projections the default projection used for CCRS
  products is the Lambert Conic Conformal (LCC)
  projection.
• The output consists of a precision geocoded
  product in a PCIDSK format file containing the
  calibrated radiance data for AVHRR channels 1 to
  5, the solar zenith and azimuth angles, and the
  satellite zenith and azimuth angles.
• All data layers are scaled according to scaling
  coefficients provided in the radiometric
  coefficient auxiliary file.

14
GeoComp nPart 1 System characterisation

• Composite Product Generation
• Pixels are selected from overlapping geocoded
  full swath images based on maximum NDVI and/or
  minimum satellite zenith angle.
• With maximum NDVI, the least contaminated pixels
  are chosen from a series of images collected over
  several days.
• With minimum satellite zenith angle, near-nadir
  pixels have reduced atmospheric effects and
  footprint size.
• Pixels may also be masked out from the final
  composite product by applying a water mask.
• The product layers are scaled according to
  scaling coefficients provided in the product
  coefficient auxiliary file.

15
GeoComp nPart 1 System characterisation

• Basic Composite Product Layers
• TOA radiance and brightness temperature
• TOA reflectance, surface reflectance,
  BRDF-corrected surface reflectance and associated
  NDVI
• Satellite and sun zenith/azimuth angles
• Quality control (QC) mask
• Input scene mask
• Pixel count mask
• Relative date
• Residual geometric error mask

16
GeoComp nPart 1 System characterisation

• Advanced Composite Product Layers
• Pixel contamination (CECANT) mask
• Land surface temperature
• Leaf area index (LAI)
• Instantaneous FPAR
• Daily Mean FPAR
• Instantaneous APAR Daily Total APAR
• Composite Mean APAR
• Fire (hot spot) mask
• PAR albedo

17
GeoComp nPart 1 System characterisation

• Manitoba Remote Sensing Centre Products
• MRSC in Winnipeg generates composite products on
  a routine basis for CCRS and assorted composites
  for MRSC clients by subscription.
• Composite products are shipped on Exabyte tapes,
  CD-ROMs and DVDs or delivered electronically.
• Daily and multi-day composite products are
  produced from April 1 to October 31.
• For the 10-day products, the composite periods
  for any given month are from the 1st to 10th,
  11th to 20th and 21st to end of the month.

18
GeoComp nPart 1 System characterisation

• Future Improvements
• GeoComp-n processing time of one hour to geocode
  an AVHRR orbit containing 5700 lines using a
  16-point damped sin x/x resampling function on a
  450MHz NT PC will decrease with faster CPUs.
• With the advent of large capacity RAID systems,
  disk space is becoming less of an issue. Typical
  disk requirements are such that one months worth
  of data consumes approximately 110 gigabytes of
  disk space if all the intermediate products (raw,
  pre-processed and geocoded) remain on-line.
• A series of Practical Extraction and Report
  Language (Perl) scripts, which are layered on top
  of the GeoComp-n software, can automate the
  import of raw data, geocoding and compositing
  processes.
• The GCP chip database will be augmented to
  include more GCPs to cover most of North America.

19
GeoComp nPart 1 System characterisation

• Conclusions
• GeoComp-n can deliver geometric and radiometric
  accuracies in composite products unprecedented in
  other operational AVHRR data processing systems.
• GeoComp-n supports a broad spectrum of
  operational and experimental products from AVHRR
  data, but expansion to other data types and new
  products is possible once their characteristics
  are established and the required algorithms
  defined.
• The higher-level products are being developed and
  validated for use in boreal and temperate
  ecosystems in Canada for applications such as
  forest fire management, crop forecasting and
  environmental monitoring.

20
GeoComp nSolar zenith angle product for the
period August 11-20, 2000
21
GeoComp nSatellite zenith angle product for
the period August 11-20, 2000
22
GeoComp nGeometric error magnitude product for
August 8, 2000
23
GeoComp nGeometric error direction product for
August 8, 2000
24
GeoComp n, an advanced system for the
processing of coarse and medium resolution
satellite data. Part 2 Biophysical products for
northern ecosystems
25
GeoComp nPart 2 Biophysical products

• Much of the research work at CCRS has focused on
  vegetation dynamics, in view of the role of
  boreal ecosystems in the global carbon cycle.
• The GeoComp-n products are used to drive or
  validate vegetation process or global climate
  models.

26
GeoComp nPart 2 Biophysical products

• Sensor Calibration
• Because of post-launch sensor degradation and the
  absence of onboard calibration for AVHRR channels
  1 and 2, time-dependent calibration coefficients
  have been derived from vicarious calibration data
  provided by NOAA and other investigators.
  Radiometric calibration of channels 1 and 2 raw
  data counts into radiance uses the piece-wise
  linear calibration coefficients as recommended by
  CCRS (Cihlar and Teillet, 1995).
• The TOA radiance can be converted to TOA
  reflectance using the method of Teillet and
  Holben (1994).
• The thermal data in AVHRR channels 3, 4 and 5 are
  converted to TOA radiance and/or brightness
  temperature using onboard calibration data with
  the Kidwell (1998) method.

27
GeoComp nPart 2 Biophysical products

• Atmospheric Correction
• TOA reflectance is converted to surface
  reflectance by applying the Simplified Method for
  Atmospheric Correction (SMAC) radiative transfer
  code (Rahman and Dedieu, 1994).
• This simplified version of the code to Simulate
  the Satellite Signal in the Solar Spectrum (5S)
  (Tanre et al., 1990) is much faster because it
  uses semi-empirical formulations and
  coefficients, which depend on the sensor spectral
  band of interest.
• Based on the analysis of AEROCAN data (Bokoye et
  al., 2002), Fedosejevs et al. (2000) found that
  aerosol optical depth of 0.06 at 550 nm is an
  acceptable value for clear-sky conditions across
  Canada.
• Nominal values of 2.3 gm cm-2 for column water
  vapour content and 319 Dobson Units for ozone are
  used in SMAC.
• The accuracy of SMAC decreases if solar zenith
  and viewing (satellite) zenith angles are above
  60o and 50o, respectively. Such cases can occur
  over the Canadian landmass.

28
GeoComp nPart 2 Biophysical products

• BRDF Correction
• AVHRR observations at northern latitudes are not
  sufficient to reconstruct a bi-directional
  reflectance distribution function (BRDF) on a
  per-pixel basis but are best used by grouping a
  complete season of surface reflectance
  observations according to the land cover type.
• AVHRR data are corrected to a standard viewing
  geometry with sun zenith angle of 45? and nadir
  viewing angle.
• A delta term was introduced to the modified
  Roujean BRDF model as a proxy for the effect of
  the changing leaf area during the growing season
  (Latifovic, Cihlar and Chen, 2002).

29
GeoComp nPart 2 Biophysical products

• Identification of Contaminated Pixels
• Despite the selection of a 10-day composite
  period for boreal ecosystems (consistent with the
  IGBP specification, the resulting composite
  products still contain some contaminated pixels.
• A procedure for Cloud Elimination from Composites
  using Albedo and NDVI Trend (CECANT) was
  developed that takes advantage of the effect of
  the atmospheric noise on NDVI over land.
• Threshold coefficients were derived from a
  reference seasonal NDVI data set.
• An adjustment was applied for near real time
  applications because of possible shifts in NDVI
  distribution among years.
• CECANT was further refined by decreasing the
  number of thresholds from three to two (deviation
  measure and surface reflectance ) per composite
  period and by making the threshold coefficients
  dependent on land cover type (Cihlar et al.,
  2002).

30
GeoComp nPart 2 Biophysical products

• Leaf Area Index
• Leaf area index (LAI) is a vegetation structural
  parameter of fundamental importance for
  quantitative assessment of physical and
  biological processes in vegetation canopies.
• LAI provides the key input for process-based
  terrestrial carbon cycle modelling (Liu et al.,
  1999).
• The LAI algorithm, which is land cover type
  dependent, was based on the simple ratio (SR) of
  Landsat 5 TM NIR to red bands after atmospheric
  and BRDF corrections, and was validated against
  ground LAI measurements acquired in eight Landsat
  TM scenes selected from across Canada (Chen et
  al., 2002).
• A spectral adjustment factor of 1.27 was applied
  to the algorithm for NOAA-11 AVHRR data.

31
GeoComp nPart 2 Biophysical products

• Fraction of Photosythetically Active Radiation
  (FPAR)
• FPAR determines the proportion of available PAR
  that a green canopy absorbs.
• In terrestrial carbon cycle estimation, FPAR is
  used to drive some empirical photosynthetic
  models or simple process models.
• The acuracy for canopy-level photosynthesis
  estimation can be improved through the use of LAI
  and a vegetation clumping index (Chen et al.,
  1999a), where the clumping index can be derived
  from multi-angle remote sensing (Chen et al.,
  1999b Lacaze et al., 2002).
• Instantaneous and daily mean (computed for the
  solar zenith angle at noon) FPAR are produced by
  GeoComp-n for use in computing APAR absorbed by
  the green canopy and for the consistency with
  similar products in other parts of the world.

32
GeoComp nPart 2 Biophysical products

• Absorbed Photosynthetically Active Radiation
  (APAR)
• APAR denotes the total PAR (incident solar energy
  between 400 and 700 nm) absorbed by the surface
  canopy/soil layers. APAR can be converted to PAR
  reaching the top of a canopy with knowledge of
  the surface PAR albedo, or to PAR absorbed by
  canopy only with knowledge of the FPAR.
• APAR is one of the most important variables
  affecting the net primary productivity of
  vegetation.
• Cloud is the main modulator of APAR, followed by
  Raleigh scattering and absorption due to
  aerosols, ozone and other gases. 
• GeoComp-n can generate instantaneous, daily
  total, and composite period mean APAR products.
• Determination of Instantaneous APAR consists of
  three steps angular correction of channel 1 TOA
  reflectance to TOA albedo using the ERBE angular
  model, spectral adjustment of channel 1 TOA
  albedo to PAR TOA albedo (Li and Moreau, 1996),
  and conversion of PAR TOA albedo to Instantaneous
  APAR.

33
GeoComp nPart 2 Biophysical products

• PAR surface albedo
• PAR surface albedo is derived from surface
  reflectance in channel 1, NDVI, solar zenith
  angle and surface cover type, following an
  integration based on the surface BRDF.  
• GeoComp-n generates a LUT of integrated BRDF
  values according to the double integral in zenith
  and azimuth direction accomplished by a summation
  of BRDF values over an angle range of 0 to 90º at
  increments of 1o. This integration is repeated
  for NDVI values (computed from surface
  reflectance) from 0 to 1.0 at increments of 0.05,
  for each sun zenith angle from 0 to 90o at
  increments of 1o, and for each of 14 land cover
  types.
• Nearest neighbor sampling of the LUT is employed
  for the actual land cover type, NDVI value and
  sun zenith angle for a given pixel.
• The current implementation assumes an inherent
  surface albedo corresponding to black-sky
  conditions when all solar radiation comes from
  one particular direction as a collimated beam
  without diffuse component. The presence of
  diffuse component may affect the magnitude of the
  albedo, especially in cloudy conditions.

34
GeoComp nPart 2 Biophysical products

• Fire Processing
• Wildfire represents a dominant disturbance to
  Canadian boreal forests, burning an annual
  average 1 of the national forested area.
• About 97 of this burning is caused by crown
  fires consuming gt 1000 ha each. Thus, fire exerts
  a major control on landscape successional
  patterns, stand age distribution, and carbon
  storage within the boreal forest.
• A satellite-based algorithm to detect actively
  burning boreal fires (as small as 0.1 of a
  pixel) has been developed at CCRS (Li et al.,
  2000). The hotspot algorithm for NOAA 14
  identifies active fires using brightness
  temperature in the mid-infrared channel (3B) and
  uses a series of threshold tests to eliminate
  false hotspots caused by highly reflective clouds
  or warm surfaces such as cropland or recently
  burnt areas. Single fire pixels that are not
  surrounded by neighbouring fire pixels are
  assumed to be caused by sun-glint and are
  eliminated.

35
GeoComp nPart 2 Biophysical products

• Fire Processing
• Raw data from the NATAS data file server are
  imported into GeoComp-n, geocoded using a nearest
  neighbour resampling algorithm and composited
  into daily products.
• A daily colour Tagged Image File Format (TIFF)
  image of the composite is also produced with the
  fire hotspots overlaid on the image.
• Both the TIFF image and the fire map are sent
  electronically to the Canadian Forestry Service
  (CFS) in Edmonton where they are imported into a
  geographic information system (GIS), overlain
  with map information and made available to forest
  fire managers via an Internet based system.
• The system is also used as a research tool for
  developing burnt area and smoke detection
  algorithms.

36
GeoComp nPart 2 Biophysical products

• Future products
• Fire Smoke
• In 2000, the threshold algorithm of Li et al.
  (2001) was applied to create daily Canada-wide
  smoke masks for the boreal forests in near real
  time as part of the Fire M3 Project.
• Cumulative Burnt Areas
• An algorithm is under development, which is
  designed to compute a daily cumulative burnt area
  mask during the forest fire season based on the
  observation that burnt boreal forest exhibits a
  strong increase in the mid-infrared spectral band
  (AVHRR channel 3A).
• A pixel is labelled as burnt if it satisfies four
  conditions (1) it is cloud-free, (2) it is
  classified as forest, (3) the pixel is spatially
  connected to an active fire or previously
  identified burnt pixel, and (4) the pixel has an
  elevated channel 3 response similar to that of
  the adjoining cluster of active hotspot/burnt
  pixels.

37
GeoComp nPart 2 Biophysical products

• Future products (continued)
• Net Primary Productivity
• The Boreal Ecosystem Productivity Simulator
  (BEPS) (Liu et al., 1999) can be run within
  GeoComp-n to produce daily net primary
  productivity (NPP) values per pixel and
  accumulate them throughout the year to obtain the
  annual NPP distribution. The prerequisites are
  (i) to compute daily NPP values before and after
  the growing season based on LAI and (ii) to
  provide GeoComp-n with daily gridded
  meteorological data.

38
GeoComp nPart 2 Biophysical products

• Future products (continued)
• Evapotranspiration
• BEPS can also be run to produce daily
  evapotranspiration (ET) distributions. As ET is
  calculated based on the Penman-Monteith method,
  daily meteorological data, LAI and clumping index
  are needed for ET calculations. ET and NPP
  products can therefore be generated
  simultaneously.
• Runoff
• Runoff maps can be produced in two stages (i)
  pixel-level excess water estimation based on the
  water balance for each pixel, and (ii) routing of
  the excess water for large watersheds. However, a
  detailed hydrological model for simulating
  discharge rates of the major rivers in Canada is
  not feasible within GeoComp-n in the near future.

39
GeoComp nPart 2 Biophysical products

• Quality Assurance
• A step towards overcoming the spatial limitation
  has been taken in LAI validation (Chen et al.,
  2002) by sampling the Canada-wide product. Such
  extensive validation studies may be required for
  other products.
• However, another strategy is needed to ensure
  that the algorithm performance does not change
  over time (Cihlar, Chen and Li, 1997).
• In addition, steps need to be taken to maintain
  consistency in the input data characteristics.
  The diverging effects may be of different origin,
  including spectral sensor characteristics even
  for nominally the same sensor type (Trishchenko,
  Cihlar and Li, 2002), calibration drift, or
  random effects in the down-linked signal and its
  initial processing.
• To compensate for sensor degradation, the most
  likely calibration coefficients are predicted for
  the current season based on the time history
  (Cihlar and Teillet, 1995).

40
GeoComp nPart 2 Biophysical products

• Quality Assurance (continued)
• As techniques for generation of automatic quality
  reports are under development and are not
  implemented as part of GeoComp-n the product
  quality assessment thus remains the
  responsibility of the analyst.
•  While near real time higher-level products have
  much greater potential value they are not likely
  to be as accurate as products generated in
  delayed mode such as at the end of the year when
  all measurements can be analyzed together.
• In case of AVHRR products from GeoComp-n, the two
  main potential causes of errors are sensor
  calibration and the identification/replacement of
  contaminated pixels.
• It should be noted that better near real time
  screening of contaminated pixels will be possible
  for other sensors like MODIS with its larger
  number of spectral bands at key wavelengths.

41
GeoComp nPart 2 Biophysical products

• Summary and Conclusions
• GeoComp-n can create a suite of higher-level
  products for the monitoring and assessment of the
  terrestrial biosphere.
• As CCRS has developed/validated products for
  application in the boreal and temperate forest
  environments the higher-level products are
  more-or-less biome-specific and should not be
  generated for areas outside of Canada without a
  thorough validation procedure.
•  Research in product error (e.g. LAI) assessment,
  including the development of automated product
  quality assessment methods, is presently
  underway.

42
GeoComp nPart 2 Biophysical products

• References
• Adair, M., J. Cihlar, B. Park, G. Fedosejevs, A.
  Erickson, R. Keeping, D. Stanley, and P.
  Hurlburt. 2001. GeoComp - n, an advanced system
  for generating products from coarse and medium
  resolution optical satellite data. Part 1 System
  characterization, Canadian Journal of Remote
  Sensing, vol. 28, no. 1, pp. 1-20.
• Cihlar, J., J. Chen, Z. Li, G. Fedosejevs, M.
  Adair, W. Park, R. Fraser, A. Trishchenko, B.
  Guindon, and D. Stanley. 2001. GeoComp-n, an
  advanced system for the processing of coarse and
  medium resolution satellite data. Part 2
  biophysical products for the northern ecosystem,
  Canadian Journal of Remote Sensing, vol. 28, no.
  1, pp. 21-44.

43
GeoComp nPart 2 Biophysical products

• References (continued)
• Bokoye, A.I., A. Royer, N.T. ONeill, G.
  Fedosejevs, P.M. Teillet and B. McArthur. 2002.
  Characterization of atmospheric aerosols across
  Canada from a ground-based sunphotometer network
  AEROCAN, Atmosphere-Ocean, Vol. 39, pp.
  429-456).
• Chen, J.M., J. Liu, J. Cihlar and M.L. Goulden.
  1999a. Daily canopy photosynthesis model through
  temporal and spatial scaling for remote sensing
  applications, Ecological Modelling, Vol. 124,
  pp. 99-119.
•  Chen, J.M., R. Lacaze, S.G. Leblanc, J.-L.,
  Roujean, and J. Liu. 1999b. POLDER BRDF and
  photosynthesis an angular signature useful for
  ecological applications, Abstract to 2nd
  international workshop on multi-angular
  measurements and models, Ispra, Italy.
• Chen, J., G. Pavlic, L. Brown, J. Cihlar, S.G.
  Leblanc, P. White, R.J. Hall, D. Peddle, D.J.
  King, J.A. Trofymow, E. Swift, J. van der Sanden,
  and P. Pellikka. 2002. Derivation and
  validation of Canada-wide coarse resolution leaf
  area index maps using high resolution satellite
  imagery and ground measurements. Remote Sensing
  of Environment, Vol. 80, pp. 165-184.

44
GeoComp nPart 2 Biophysical products

• References (continued)
• Cihlar, J., and P.M. Teillet. 1995. Forward
  piecewise linear model for quasi-real time
  processing of AVHRR data, Canadian Journal of
  Remote Sensing, Vol. 21, pp. 22-27.
• Cihlar, J., J. Chen, and Z. Li. 1997b. On the
  validation of satellite-derived products for land
  applications, Canadian Journal of Remote
  Sensing, Vol. 23, pp. 381-389.
• Cihlar, J., Latifovic, R., Chen, J., Trishchenko,
  A., Du, Y., Fedosejevs, G., and Guindon, B. 2002.
  Systematic corrections of AVHRR image composites
  for temporal studies, Remote Sensing of
  Environment (in press).
• Fedosejevs, G., N.T. ONeill, A. Royer, P.M.
  Teillet, A.I. Bokoye and B. McArthur. 2000.
  Aerosol optical depth for atmospheric correction
  of AVHRR composite data, Canadian Journal of
  Remote Sensing, Vol. 26, pp. 273-284.
• Kidwell, K. B. (Ed). 1998. NOAA Polar Orbiter
  Data Users Guide, NOAA-NESDIS, Washington, D.C.
  lthttp//www2.ncdc.noaa.gov/docs/podug/gt.

45
GeoComp nPart 2 Biophysical products

• References (continued)
• Lacaze, R., J.M. Chen, J-L. Roujean, and S.G.
  Leblanc. 2002. Retrieval of vegetation clumping
  index using hotspot signatures measured by
  multi-angular POLDER instrument Remote Sensing
  of Environment, Vol. 79, pp. 84-95.
• Latifovic, R., J. Cihlar, and J. Chen. 2002. A
  comparison of BRDF models for the normalisation
  of satellite optical data to a standard
  sun-target-sensor geometry, IEEE Transaction for
  Geoscience and Remote Sensing (in press).
• Li, Z., and L. Moreau. 1996b. A new approach for
  remote sensing of canopy-absorbed
  photosynthetically active radiation. I Total
  surface absorption, Remote Sensing of
  Environment, Vol. 55, pp. 175-191.
• Li, Z., S. Nadon, and J. Cihlar. 2000. Satellite
  detection of Canadian boreal forest fires
  Development and application of an algorithm,
  International Journal of Remote Sensing, Vol. 21,
  pp. 3057-3069.
• Liu, J., J.M. Chen, J. Cihlar, and W. Chen. 1999.
  Net primary productivity distribution in the
  BOREAS study region from a process model driven
  by satellite and surface data, Journal of
  Geophysical Research, Vol. 104, No. D22, pp.
  27,735-27,754.

46
GeoComp nPart 2 Biophysical products

• References (continued)
• Rahman, H. and G. Dedieu. 1994. SMAC A
  Simplified Method for the Atmospheric Correction
  of Satellite Measurements in the Solar Spectrum,
  International Journal of Remote Sensing, Vol. 15,
  pp. 123-143.
• Tanre, D., C. Deroo, P. Duhaut, M. Herman, J.J.
  Morcrette, J. Perbos, and P.Y. Deschamps. 1990.
   Description of a computer code to simulate a
  satellite signal in the solar spectrum the 5S
  code, International Journal for Remote Sensing,
  Vol. 14, pp. 659-668.
• Teillet, P.M. and B.N. Holben. 1994. Towards
  operational radiometric calibration of NOAA AVHRR
  imagery in the visible and infrared channels,
  Canadian Journal of Remote Sensing, Vol. 20, pp.
  1-10.
• Trishchenko, A.P., J. Cihlar, and Z. Li. 2002.
  Effects of spectral response function on the
  surface reflectance and NDVI measured with
  moderate resolution sensors, Remote Sensing of
  Environment, Vol. 80, pp. 1-18.

47
GeoComp nBRDF-corrected surface reflectance
for AVHRR channel 1 August 11-20, 2000
48
GeoComp nAverage APAR product for the period
August 11-20, 2000
49
GeoComp nPAR albedo product for August 8, 2000
50
Useful Web Sites

• CCRS GeoComp-n http//www.ccrs.nrcan.gc.ca/ccrs/rd
  /ana/geocomp/geocomp_e.html
• NATAS http//ceocat.ccrs.nrcan.gc.ca/client_acc/gu
  ides/avhrr/ch4.html
• NOAA Reception at CCRS http//www.ccrs.nrcan.gc.ca
  /ccrs/data/satsens/sats/noaa_e.html
• CCRS CalVal http//www.ccrs.nrcan.gc.ca/ccrs/rd/an
  a/calval/calhome_e.html
• CEOCAT
• http//ceocat.ccrs.nrcan.gc.ca/cgi-bin/client_acc
  /ceocate/holdings.phtml
• MRSC http//www.gov.mb.ca/conservation/geomatics/r
  emote_sensing/index.html
• Fire M3
• http//fms.nofc.cfs.nrcan.gc.ca/FireM3

51
Contact

• Gunar Fedosejevs
• Data Acquisition Division
• Canada Centre for Remote Sensing
• 588 Booth Street
• Ottawa, Ontario, Canada K1A 0Y7
• E-mail Gunar.fedosejevs_at_ccrs.nrcan.gc.ca
• URL http//www.ccrs.nrcan.gc.ca/ccrs/