422
W.G.M. Bastiaanssen
15.10 Concluding Remarks
- Land surface fluxes for climate studies are simulated by means of land surface parameterization schemes coupled to Atmospheric Models. Although the validation of these
schemes at micro- scale is useful (e.g. van den Hurk et al., 1995), studies on their peformance at meso-scale should be pursued using meso-scale flux estimates from remote
sensing algorithms such as illustrated in Fig. 15.4.
- The key-issue in evaporation estimations based on remote sensing data is an adequate
description of sensible heat flux: Latent heat flux will be obtained from energy budget
residuals and any error in H will be directly conveyed into error in >.E;
- Patch scale fluxes can be best estimated by a near-surface level for the upper air layer
to which Eq. (15.21) applies because (i) the height of the internal adapted layer over
homogeneous patches is generally small and effects from (ii) horizontal advection are
reduced. The latter implies that the areal variation of near-surface T.-values need great
attention and that the surface energy balance can be regarded as a descriptor of vertical
exchange processes;
- Regional scale fluxes can be well estimated by considering T. of the mixed layer (remote
sensing flux algorithm classes 3 and 4) and it is not strictly necessary to apply class 5 of
algorithms with spatially variable near-surface air temperature for this purpose;
- The use of aerodynamic temperatures (kB-l = 0) to determine sensible heat fluxes overestimates of the fluxes (Hall et al., 1992): kB- 1 needs therefore to be calibrated locally
to obtain realistic Hmix_ fluxes from one-layer resistance schemes. At the same time it is
concluded that kB-l is almost impossible to extrapolate spatially;
- The linearity of .sT. (To) relationships provides a solution to spatially interpolate .sT. in
a simple manner for a wide variety of surface types and climate conditions. If .sT. is
coupled to an a priori chosen and areal fixed kB- 1 value (physics shows that there is a
relationship among them), it forms a promising technique to calculate Hmix at composite
terrain with a one-layer parameterization scheme;
- Inversion of the transport equation for sensible heat flow for land surface elements where
>'Emix ~ 0 and Hmix ~ 0 provides an opportunity to derive .sT. in an inverse mode which
excludes the need to involve collateral T.-i data. This 'internal calibration' procedure at
meso-scale, may substitute kB-l calibrations valid at micro scale only (e.g. Lhomme et
al., 1994).
- The SEBAL parameterization is less sensitive to Tr;'ix as compared to conventional schematizations of H because Tr;'ix is used as a predictive variable to estimate .sT::ix rather
as a dependent parameter to compute OT::ix. Since in general, the error in deriving OTr;'ix
from narrow band radiometers is ± 2 to 3°G, this may be conceived as an essential step
forward.
- Validation of fluxes calculated with low resolution data against multiple tower-based fluxes can only be realized qualitatively. A better option may be to validate the regional
evaporation against regional water balances, or against high resolution images being validated with in-situ fluxes, i.e. a nested calibration procedure.
W.G.M. Bastiaanssen
15.10 Concluding Remarks
- Land surface fluxes for climate studies are simulated by means of land surface parameterization schemes coupled to Atmospheric Models. Although the validation of these
schemes at micro- scale is useful (e.g. van den Hurk et al., 1995), studies on their peformance at meso-scale should be pursued using meso-scale flux estimates from remote
sensing algorithms such as illustrated in Fig. 15.4.
- The key-issue in evaporation estimations based on remote sensing data is an adequate
description of sensible heat flux: Latent heat flux will be obtained from energy budget
residuals and any error in H will be directly conveyed into error in >.E;
- Patch scale fluxes can be best estimated by a near-surface level for the upper air layer
to which Eq. (15.21) applies because (i) the height of the internal adapted layer over
homogeneous patches is generally small and effects from (ii) horizontal advection are
reduced. The latter implies that the areal variation of near-surface T.-values need great
attention and that the surface energy balance can be regarded as a descriptor of vertical
exchange processes;
- Regional scale fluxes can be well estimated by considering T. of the mixed layer (remote
sensing flux algorithm classes 3 and 4) and it is not strictly necessary to apply class 5 of
algorithms with spatially variable near-surface air temperature for this purpose;
- The use of aerodynamic temperatures (kB-l = 0) to determine sensible heat fluxes overestimates of the fluxes (Hall et al., 1992): kB- 1 needs therefore to be calibrated locally
to obtain realistic Hmix_ fluxes from one-layer resistance schemes. At the same time it is
concluded that kB-l is almost impossible to extrapolate spatially;
- The linearity of .sT. (To) relationships provides a solution to spatially interpolate .sT. in
a simple manner for a wide variety of surface types and climate conditions. If .sT. is
coupled to an a priori chosen and areal fixed kB- 1 value (physics shows that there is a
relationship among them), it forms a promising technique to calculate Hmix at composite
terrain with a one-layer parameterization scheme;
- Inversion of the transport equation for sensible heat flow for land surface elements where
>'Emix ~ 0 and Hmix ~ 0 provides an opportunity to derive .sT. in an inverse mode which
excludes the need to involve collateral T.-i data. This 'internal calibration' procedure at
meso-scale, may substitute kB-l calibrations valid at micro scale only (e.g. Lhomme et
al., 1994).
- The SEBAL parameterization is less sensitive to Tr;'ix as compared to conventional schematizations of H because Tr;'ix is used as a predictive variable to estimate .sT::ix rather
as a dependent parameter to compute OT::ix. Since in general, the error in deriving OTr;'ix
from narrow band radiometers is ± 2 to 3°G, this may be conceived as an essential step
forward.
- Validation of fluxes calculated with low resolution data against multiple tower-based fluxes can only be realized qualitatively. A better option may be to validate the regional
evaporation against regional water balances, or against high resolution images being validated with in-situ fluxes, i.e. a nested calibration procedure.