BSRN Validation for GEWEX/ISCCP

1
BSRN Validation for GEWEX/ISCCP

• R. T. Pinker
• Department of Meteorology
• University of Maryland
• College Park, MD
• CEOS/WGCV Land Product Validation
• Workshop on Albedo
• April 27-28, 2005
• EGU 2005 Vienna

2

• Highlights
• What is available on global scale from GEWEX
• Scale issues-same satellites, different spatial
• gridding
• Scale issues-different satellites
• Evaluation of products
• Discussion in context of
• CEOS/WGCV Land Product Workshop
• Boston, 23-24 October, 2002

3

• Error budgets
• Scaling may be much larger source of error than
• calibration or narrow to broadband
  transformation.
• What to measure-albedo reflectance spectral
  intervals
• The scientific community has primary interest in
  the total albedo. Therefore, it is recommended
  that priority should be given to measurements
  of this parameter.
• There is also interest in spectral albedo, both
  broadband and narrow band.
• Broadband intervals of interest are the visible
  and NIR for which instrumentation of acceptable
  quality is available.
• High spectral resolution albedo or BRDF are also
  of interest in building spectral libraries and
  for evaluating narrow band albedo from
  satellites.

4
What we do
Global distribution of surface SW radiation at
0.5 degree, Jan 1992 Both downward and upward
fluxes are computed, allowing derivation of albedo
5

• Approach to Derive SW
• Start with spectral reference albedo models as
• boundary conditions
• Compute surface upwelling and down- welling
  fluxes
• Clear conditions
• All conditions
• Compute spectrally for
• diffuse
• total
• Albedo ratio of upwelling to down -welling flux

6
Narrow to broad band transformations-updates
Objective
To derive angularly and seasonally dependent
relationships between narrowband reflectance and
broadband albedo, under clear sky conditions, as
observed from the several different satellites.
The simulations will utilize surface types based
on the University of Maryland land cover
classifications.
Narrow to broadband transformations are based on

• Extensive model simulations with MODTRAN 3.7
• 10 solar zenith angle bins
• 16 gaussian points in zenith and 8 in zenith
  direction to obtain the spectral
• flux
• 20 climatological profiles for temperature,
  water vapor and ozone in 4
• seasons
• Newly developed spectral surface albedo models

7
Updated Surface Spectral Reflectance
ModelsCompatible with Current Global Land Cover
Classifications Scaling Factors for the surface
types according to UMD land cover classification
Ch1 Ch2 Ch2 Scaling Scaling Scaling
Source Surface type factor factor
factor (0.3-0.5) (1.3-2) (2-4)
Evergreen Needle leaf Forest 0.74 0.39
0.14 ASTER Evergreen Broadleaf Forest
0.83 0.46 0.18 ASTER Deciduous
Needle leaf Forest 0.74 0.39 0.14
ASTER Deciduous Broadleaf Forest 0.83 0.46
0.18 ASTER Mixed Forest 0.79 0.42
0.16 ASTER Woodlands 0.35 0.25
0.09 ASTER/Bowker Wooded
Grasslands 0.36 0.32 0.12
ASTER/Bowker Closed Bushland/Shrubland 0.37
0.36 0.13 ASTER/Bowker Open
Shrubland 0.19 0.22 0.07
ASTER/Bowker Grassland 0.07 0.19 0.03
ASTER Cropland 0.54 0.51 0.20
ASTER/Bowker Barren 0.55 1.35
1.13 ASTER/Bowker Urban
ASTER
8

• What albedo products are available
• At global scale
• ISCCP D1 UMD GEWEX/SRB, 2.5 deg, 1983-2001
• (There are other estimates based on ISCCP
  D1, such as LaRC
• GEWEX/SRB ISCCP D1-FD)
• MODIS 1 deg, UMD SRB model, about 3 years
• At continental scale
• ISCCP DX based estimates, 0.5 deg 1990-2001
• At regional scale
• GOES based estimates at 0.5 degree for the US
  1996-current
• Improved above product for 1996-2000 at 0.5 and
  1/8 deg resolution
• GOES based estimates at 0.5 deg, Amazon basin
  for 1998-2000.
• This is not a comprehensive review of
  everything available

9
Examples of low resolution albedo products
10
Based on GOES pixel level data available for
1998-2000
11
Annual mean Variability among sites Seasonal variability
Forest Obs 0.133 0.122-0.147 Weak, higher in dry, lower in wet season
Forest GOES 0.088 0.079-0.104 Very weak
Forest DX 0.113 0.103-0.129 Lower in Jun/Jul/Aug
Pasture Obs 0.176 0.169-0.193 Stronger, Lower in dry season
Pasture GOES 0.090 0.075-0.100 Very weak
Pasture DX 0.116 0.102-0.127 Lower in Jun/Jul/Aug
Analysis based on Berbet, M. et al. (2003)
observations made in 1990-1993 (Culf et al.,
1996). The observation uncertainty of the order
of 0.006 (Wright et al. 1996). GOES and DX
estimates are available from 1998/03 2001/02,
and from 1998/07 2001/02, respectively.

• Both observed and satellite derived albedo are
  generally lower in dry and higher in wet
  months.
• Satellite driven albedo show smaller difference
  between the vegetation cover, and weak seasonal
  variability.

12
Based on GOES pixel level
1/2 deg
1/8th deg
13
GOES 1/8th
ISCCP DX, 0.5 deg
14
Albedo based on MODIS using ratio of surface
fluxes The surface parameters used to calculate
spectral surface albedo needed in the inference
scheme for SRB were taken from the MODIS
Bidirectional Reflectance Distribution Function
(BRDF) and Albedo Product (MOD43B) at the 0.25
resolution (Lucht et al., 2000 Schaaf et al.,
2002). The three weighting parameters associated
with the RossThickLiSparseReciprocal BRDF model
that best describes the anisotropy of each pixel
are provided for each of the MODIS spectral bands
as well as for the three broad bands (0.3-0.7 µm,
0.7-5.0 µm, and 0.3-5.0 µm). For two broad bands
(0.3-0.7 µm, 0.7-5.0 µm), these parameters were
used with simple polynomials to estimate the
white sky albedo and the black sky albedo for the
monthly mean solar zenith angle.  
15
Auxiliary information for driving SRB model
Level-3 MODIS Atmosphere Monthly Global Product
at 1 x 1 resolution, processed with the latest
collection 4 algorithms for  Cloud_Fraction_Total
, Cloud_Optical_Thickness_Combined,Optical_Depth_L
and_And_Ocean aerosol, Total_Ozone, and
Atmospheric_Water_Vapor  Missing aerosol optical
depths over arid areas were filled from the
MODIS-GOCART integrated monthly aerosol optical
depth data, School of Earth and Atmosphere
Sciences, Georgia Institute of Technology.
Missing cloud optical thickness values were
replaced by interpolated values.
16
One degree spatial resolution
17
From MODIS, V004 product
18
MODIS swath data at highest resolution
19
Evaluation of GOES surface albedos against
SURFRAD stations during summer and winter, Fort
Pack, MT (0.5 degree)
20
Evaluation os GOES surface albedo against SURFRAD
stations during summer and winter Goodwin Creek,
MS (0.5 degree)
21
Evaluation

• Comparison between derived broadband albedo for
  different satellites
• Comparison between the derived broadband surface
  albedo and broadband surface observations - The
  methodology will be tested with ground
  observations at the semi-arid USDA-ARS Walnut
  Gulch Experimental Watershed in Arizona. The
  surface observations started in July 1999 and
  continue up to now. Total short-wave, infrared
  and PAR upward and downward fluxes are measured
  at 5 minute intervals.
• Comparison between the derived broadband albedo
  from GOES and broadband albedo observed from a
  satellite with broad band sensor

22
Semi-Arid region Walnut Gulch, AZ
23
Ongoing observations in sub-Sahel
Need for ground truth on surface albedo
24
No information on savannah type vegetation
25
(No Transcript)
26

• Summary
• Global products of spectral and total surface
  albedos are available
• As yet, not fully evaluated
• Usefulness of ground observations for
  evaluation of low resolution
• products not obvious-preferably, consistency
  among satellites
• important
• Ground observations of albedo important, but
  should not be
• expected to match the satellite based
  albedos
• Documentation of the annual cycle for different
  surface types
• important for benchmarking