414
W.G.M. Bastiaanssen
China
Data from the HEIFE experiment in Central China (Wang et al., 1994) has been considered to
analyse the aT.(To) relationship for landscapes where extremes in physiography and hydrology
exists. The study area encompasses the Gobi desert, irrigated oases near the towns of Zhan Ye
and Linzhe and mountainous hillslopes. The distance between the most far away measurement
points was approximately 50 km. Fig. 15.6 shows the observed trends at 11:00 hours local time
for To, T. and aT •. At this time, the lowest air temperature was T. = 26°C while the hot spots
gave already T. = 31°C at the same elevation above the land surface. The depicted T.(To)
relationship in Fig. 15.6A expresses much scatter with significant standard deviations from the
regression line. On the contrary, the aTa(To) trend of Fig. 15.6B has a strongly reduced scatter.
Hence, the expected linear trend for the aTa(To) relationship is also confirmed for the arid
conditions of the HEIHE basin for several days with C3 = 0.97°C °C-l and C4 = 27.8°C on
average.
The experimental evidence of the aTa(To) relationship can be summarized by the following
conclusions:
- The linear aTa(To) hypothesis for heterogeneous terrain conditions can be experimentally
proven and the slope lies close to one (Niger 0.98, Egypt 1.09, China 0.97) which implies
that C4 is a good first order estimation of T:JJ;
- aTa can be spatially interpolated using a To-image if C3 and C4 for that particular image
and moment of acquisition are determined;
15.7 Implications of the Current kB- 1 and 8T a Findings
for Meso-scale Sensible Heat Flux Estimations
Because T;o~x and T;:ix or Ta- i need to be sampled at exactly the same moment and place
(remote radiometers: T;o~x; in-situ observations: Ta- i, radiosoundings: T;:ix) and with a very
high precision, it is challenging to obtain T;o~x and T;:ix for Eq, (15.21) with the required accuracy. Furthermore, it was concluded previously that kB- 1 needs to be calibrated against local
T and H-fluxes, which outside HAPEX experiments at meso-scale is hardly feasible. However,
kB-l and aT. are related because To increases with kB- 1 (Fig. 15.1) which directly affects aT.
(aT. = To - T.). In other words, an increased resistance r;:';x by an increase of kB- 1 should
according to the concept of logarithmic profiles in the constant flux layer result in an increased
aT;:ix value. On similar grounds, Kustas et al. (1989) found that kB- 1 can be related to aT.
and u. They calibrated kB- 1 and extrapolated kB-l(t) in the time domain using To(t) (not
in the space domain). Hence, either kB- 1 or aT;:ix can be calibrated in one-layer resistance
schemes (Fig. 15.1) to obtain appropriate H-fluxes. Considering on one hand the results of the
previous section showing the potential to interpolate aT. regionally for a wide range of surface
types and climate conditions and on the other hand the difficulties to predict the spatial distribution of kB- 1 in complex terrain (Dolman and Blyth, 1996), it is preferred for heterogenoues
meso-scale land surface to calibrate aT;:ix rather than kB- 1 • Then kB- 1 should be areally
fixed a priori and aT;:ix will be adjusted to that kB- 1 value in order to maintain adequate
H-fluxes. Since H-fluxes are usually not available, it is further proposed to optimize aT;:ix
from sub-areas (groups of pixels) where H reaches its limits. For those sub-areas, an inversion
W.G.M. Bastiaanssen
China
Data from the HEIFE experiment in Central China (Wang et al., 1994) has been considered to
analyse the aT.(To) relationship for landscapes where extremes in physiography and hydrology
exists. The study area encompasses the Gobi desert, irrigated oases near the towns of Zhan Ye
and Linzhe and mountainous hillslopes. The distance between the most far away measurement
points was approximately 50 km. Fig. 15.6 shows the observed trends at 11:00 hours local time
for To, T. and aT •. At this time, the lowest air temperature was T. = 26°C while the hot spots
gave already T. = 31°C at the same elevation above the land surface. The depicted T.(To)
relationship in Fig. 15.6A expresses much scatter with significant standard deviations from the
regression line. On the contrary, the aTa(To) trend of Fig. 15.6B has a strongly reduced scatter.
Hence, the expected linear trend for the aTa(To) relationship is also confirmed for the arid
conditions of the HEIHE basin for several days with C3 = 0.97°C °C-l and C4 = 27.8°C on
average.
The experimental evidence of the aTa(To) relationship can be summarized by the following
conclusions:
- The linear aTa(To) hypothesis for heterogeneous terrain conditions can be experimentally
proven and the slope lies close to one (Niger 0.98, Egypt 1.09, China 0.97) which implies
that C4 is a good first order estimation of T:JJ;
- aTa can be spatially interpolated using a To-image if C3 and C4 for that particular image
and moment of acquisition are determined;
15.7 Implications of the Current kB- 1 and 8T a Findings
for Meso-scale Sensible Heat Flux Estimations
Because T;o~x and T;:ix or Ta- i need to be sampled at exactly the same moment and place
(remote radiometers: T;o~x; in-situ observations: Ta- i, radiosoundings: T;:ix) and with a very
high precision, it is challenging to obtain T;o~x and T;:ix for Eq, (15.21) with the required accuracy. Furthermore, it was concluded previously that kB- 1 needs to be calibrated against local
T and H-fluxes, which outside HAPEX experiments at meso-scale is hardly feasible. However,
kB-l and aT. are related because To increases with kB- 1 (Fig. 15.1) which directly affects aT.
(aT. = To - T.). In other words, an increased resistance r;:';x by an increase of kB- 1 should
according to the concept of logarithmic profiles in the constant flux layer result in an increased
aT;:ix value. On similar grounds, Kustas et al. (1989) found that kB- 1 can be related to aT.
and u. They calibrated kB- 1 and extrapolated kB-l(t) in the time domain using To(t) (not
in the space domain). Hence, either kB- 1 or aT;:ix can be calibrated in one-layer resistance
schemes (Fig. 15.1) to obtain appropriate H-fluxes. Considering on one hand the results of the
previous section showing the potential to interpolate aT. regionally for a wide range of surface
types and climate conditions and on the other hand the difficulties to predict the spatial distribution of kB- 1 in complex terrain (Dolman and Blyth, 1996), it is preferred for heterogenoues
meso-scale land surface to calibrate aT;:ix rather than kB- 1 • Then kB- 1 should be areally
fixed a priori and aT;:ix will be adjusted to that kB- 1 value in order to maintain adequate
H-fluxes. Since H-fluxes are usually not available, it is further proposed to optimize aT;:ix
from sub-areas (groups of pixels) where H reaches its limits. For those sub-areas, an inversion