707 
nay remain in our 
INFLUENCE OF THE SKY RADIANCE DISTRIBUTION ON VARIOUS 
FORMULATIONS OF THE EARTH SURFACE ALBEDO 
in visioned for the 
soundness of the 
ve accuracy. The 
onal signatures at 
ance corrected for 
P. Lewis and M.J. Barnsley 
ratoire d’Optique 
Spatiales (CNES). 
die “La Crau-91" 
Herman and C. 
1. The POLDER 
ponsorship of the 
Remote Sensing Unit, Department of Geography, 
University College London, 
26 Bedford Way, LONDON, WC1H OAP. England. 
ABSTRACT: 
ocq J.F., UstinS., 
>espontin C.. and 
l. 0. Green (Ed.), 
c JPL. 
ER measurements 
This paper presents an investigation of the derivation of albedo from remotely-sensed data, and its subsequent 
use in modelling shortwave energy interactions at the Earth surface. Analytical BRDF and sky radiance models 
are used to simulate the effects of the irradiance field on albedo for a variety of cover types in order to assess 
various formulations of albedo products to be derived from remotely-sensed data from current and future 
satellites. 
KEYWORDS: Albedo; Bidirectional Reflectance Distribution Function; Reflectance Modelling; Irradiance. 
Met., 10. 1285. 
.. Deuzd. G. Sdze. 
• ans. Geosc. Rem. 
1 - INTRODUCTION 
s of the POLDER 
Shortwave energy interactions at the earth surface are often characterized by the surface albedo, which is the 
[RR data for crop 
proportion of incident solar radiation that is reflected from the earth surface (Dorman and Sellers 1989). Global 
Climate Models (GCMs) and Soil-Vegetation-Atmosphere Transfer Schemes (SVATS), in particular, are 
nation through its 
dependent on the accuracy with which surface albedo can be specified. For example. Sellers (1993) gives an 
accuracy requirement of ±0.02 for current GCMs. Notwithstanding this, the albedo values currently used in many 
ns., 8, 1235-1243. 
HRR data'. Int. J. 
models of the radiation regime at the earth surface are often derived from a disparate set of studies reported in 
the literature. These values — frequently of unknown or variable accuracy — are then assigned to broad 
o., 38. 193-210. 
anopies’, Applied 
ecological categories in composite, often generalized, land cover maps (Matthews 1983, Wilson and Henderson- 
Sellers 1985). 
, and G.F. Epema, 
Sens. Enviro., 18, 
The lack of spatially and temporally disaggregated information on albedo at continental to global scales 
necessarily limits the accuracy of GCM and SVATS simulations, and therefore acts as a barrier to an improved 
understanding of the earth surface processes that they describe. Satellite remote sensing offers the potential to 
distributions of a 
overcome this barrier, by re-specifying albedo as a spatially variable and temporally dynamic value. 
Unfortunately, satellite sensors do not measure surface albedo directly. Rather, they record the radiance at the 
surfaces. Applied 
top of the atmosphere (TOA), integrated directionally over the Instantaneous Field-Of-View (IFOV) of the sensing 
itrusions. Applied 
element and spectrally over the sensor’s wavebands (i.e. hemispherical-conical spectral radiance). Thus, remote 
sensing measurements not only need to be converted into ’at ground’ reflectance data, but also integrated with 
tional reflectances 
respect to viewing angle and wavelength to yield surface albedo. The former requires the effect of the 
atmosphere to be taken into account; the latter requires a model of the directional scattering properties of earth 
ance bidirectional 
surface materials — hereafter referred to as a surface scattering model (SSM). 
rth surface for the 
Even with an appropriate SSM, retrieval of the surface albedo is dependent on having a sufficient number of well 
distributed hemispherical-conical radiance measurements against which to invert the SSM (The term well 
iro, 17, 165-178. 
the bidirectional 
distributed’ refers to the angular distribution of the radiance samples within the viewing and illumination 
hemispheres). In practice, the range of angles over which such data can be collected using spacebome 
instruments is conditioned by the geometry of the sensor and the orbital characteristics of the satellite on which 
it is mounted (Barnsley et al. 1994). Further limitations will be imposed by the length of time that it takes to 
acquire the sample set of multi-angle measurements {ibid). Provided that this requirement is met, the SSM can 
be used to provide either an analytical integration with which to infer albedo, or a more comprehensive set of 
reflectance samples for numerical integration.