168
M. Menenti
posed a different type of linear relationship by relating (Rn - LE)day to the rate of
increase in surface temperature during the morning. This approach avoids the need
for ground observations of air temperature. As with the original linear relationship,
the parameters depend on meteorological and surface variables and a planetary
boundary layer model was used to study this dependence. Carlson and Buffum
(1989) observed that the parameters were highly sensitive to wind speed and surface roughness.
8.3.4 Relationships between evaporation, surface temperature and spectral
indices [3]
Price (1990) explicitly related observed variability of surface temperature and a
spectral index to the range of actual evaporation. The spectral index provides a
measure of the green vegetation cover. Potential evaporation may be estimated
independently with e.g. the methods mentioned in the preceding pages. Maximum
evaporation for a given amount of green vegetation may be estimated as described
by Inoue and Moran (1997). Minimum and maximum surface temperatures change
as a function of the amount of green vegetation: the range is largest for bare surfaces and decreases with increasing green vegetation.
Carlson et al. (1995) proposed to modify the simplified linear relationship to account for the spatial variability of vegetation cover through variable B (the constant in the linear relationship) and n (an exponent accounting for deviations from
linearity). To determine the required parameters a scatter plot of NDVI vs. radiometric surface temperature was used. To explain this pattern and to estimate the B
and n parameters a SV AT model was used. The results indicated that deviations
from linearity were significant, although a family of linear relationships (i.e. for
different combinations of the two parameters B and n) could still be a practical and
useful approximation. To estimate the near surface air temperature, the surface
radiometric temperature corresponding with the apex of the scatter plot, i.e. at the
highest observed NDVI was taken. The air temperature at 50 m elevation was used
as a reference and estimated from the near surface air temperature as (Ta - 1).
Carlson et al. (1995) noted that the estimation and mapping of evaporation requires accurate values of net radiation.
The slope of the NDVI vs. Trod relationship has been interpreted as a proxy of
canopy resistance by Nemani and Running (1989). Smith and Choudhury (1991)
analyzed in detail the relation of surface radiant temperature and NDVI for agricultural land and native evergreen forests in New South Wales. Surface radiant
temperature increased with decreasing NDVI for agricultural land, but not for forests. They also noted that changes in soil water availability had opposite effects on
the slope of the NDVI vs. Trad relationship according to whether the reduction in
evaporation was from soil or from vegetation. Another evaluation of the NDVI vs.
Trod relationship was presented by Friedl and Davis (1994) who used observations
of a tall-grass prairie in NE Kansas. While the negative correlation was observed
consistently through time, the slope was highly date- and time-specific. The
M. Menenti
posed a different type of linear relationship by relating (Rn - LE)day to the rate of
increase in surface temperature during the morning. This approach avoids the need
for ground observations of air temperature. As with the original linear relationship,
the parameters depend on meteorological and surface variables and a planetary
boundary layer model was used to study this dependence. Carlson and Buffum
(1989) observed that the parameters were highly sensitive to wind speed and surface roughness.
8.3.4 Relationships between evaporation, surface temperature and spectral
indices [3]
Price (1990) explicitly related observed variability of surface temperature and a
spectral index to the range of actual evaporation. The spectral index provides a
measure of the green vegetation cover. Potential evaporation may be estimated
independently with e.g. the methods mentioned in the preceding pages. Maximum
evaporation for a given amount of green vegetation may be estimated as described
by Inoue and Moran (1997). Minimum and maximum surface temperatures change
as a function of the amount of green vegetation: the range is largest for bare surfaces and decreases with increasing green vegetation.
Carlson et al. (1995) proposed to modify the simplified linear relationship to account for the spatial variability of vegetation cover through variable B (the constant in the linear relationship) and n (an exponent accounting for deviations from
linearity). To determine the required parameters a scatter plot of NDVI vs. radiometric surface temperature was used. To explain this pattern and to estimate the B
and n parameters a SV AT model was used. The results indicated that deviations
from linearity were significant, although a family of linear relationships (i.e. for
different combinations of the two parameters B and n) could still be a practical and
useful approximation. To estimate the near surface air temperature, the surface
radiometric temperature corresponding with the apex of the scatter plot, i.e. at the
highest observed NDVI was taken. The air temperature at 50 m elevation was used
as a reference and estimated from the near surface air temperature as (Ta - 1).
Carlson et al. (1995) noted that the estimation and mapping of evaporation requires accurate values of net radiation.
The slope of the NDVI vs. Trod relationship has been interpreted as a proxy of
canopy resistance by Nemani and Running (1989). Smith and Choudhury (1991)
analyzed in detail the relation of surface radiant temperature and NDVI for agricultural land and native evergreen forests in New South Wales. Surface radiant
temperature increased with decreasing NDVI for agricultural land, but not for forests. They also noted that changes in soil water availability had opposite effects on
the slope of the NDVI vs. Trad relationship according to whether the reduction in
evaporation was from soil or from vegetation. Another evaluation of the NDVI vs.
Trod relationship was presented by Friedl and Davis (1994) who used observations
of a tall-grass prairie in NE Kansas. While the negative correlation was observed
consistently through time, the slope was highly date- and time-specific. The