114
varying partitioning of energy and water fluxes by different vegetation types (e.g. Seth et
a!. 1994), these approximations can lead to significant errors in the calculation of grid-box
averaged surface energy and water budgets. Having recognized this potential source of
uncertainty, many investigators have recently attempted to address the issue of describing
the effects of surface heterogeneity in ESEMs.
In this section, the basic problem associated with heterogeneous surface representation
in ESEMs is first defined. Attempts to quantify the uncertainties related to surface heterogeneities and modeling approaches developed to include heterogeneous surfaces within
AMs and ESEMs are then reviewed.
4.1. The problem of surface heterogeneity representation
Land surface heterogeneities occur for three basic reasons: i) variations in vegetation
cover or, more generally, surface type (e.g. different types of vegetation, bare soil, snow,
inland water, urban areas); ii) variations in terrain morphology (e.g. slope and elevation);
and iii) variations in soil characteristics (e.g. color and texture). A further degree of
heterogeneity is added by the climate forcing, which can be highly variable in space and
time. A typical example of this is given by summertime convective precipitation, which
can strongly vary on scales of a few km or even less. All these effects can strongly alter
the local surface energy and water budgets, and although for some aspects they can be
partially correlated, they generally vary quite independently from each other.
This suggests that surface heterogeneity likely spans a wide range of spatial scales.
Therefore, although based on one of the criteria above it might appear that the surface is
a mosaic of well defined patches, some characteristics based on other criteria may produce
within each patch a high degree of variability. An enlighting example is that of A vissar
et a!. (1991), who measured the plant stomatal resistance in a homogeneous potato field.
They found that the stomatal resistance followed a quasi log-normal distribution, mostly
in response to variations of the leaf micro-environment (inclination, orientation, shading,
level within the canopy, wind). For larger patches, e.g. a forest, where different species
share the same environment and where terrain and soil texture can significantly vary, this
effect could be far more pronounced.
These considerations allow us to define two types of heterogeneity, what we can call
inter-patch and intra-patch heterogeneity. The separation between inter- and intra-patch
heterogeneit.y cannot be strictly based on spatial scale, since we have commented that
heterogeneity likely varies on a continuum of scales, but rather on the mathematical
approach most suitable to treat them. Inter-patch heterogeneity can be treated with
"discrete" methods, while "intra-patch" heterogeneity requires continuous approaches.
The effects that sub-grid scale surface heterogeneity has on an AM can also be roughly
divided in two categories, what we here refer to as "direct" and "indirect" effects. Direct
varying partitioning of energy and water fluxes by different vegetation types (e.g. Seth et
a!. 1994), these approximations can lead to significant errors in the calculation of grid-box
averaged surface energy and water budgets. Having recognized this potential source of
uncertainty, many investigators have recently attempted to address the issue of describing
the effects of surface heterogeneity in ESEMs.
In this section, the basic problem associated with heterogeneous surface representation
in ESEMs is first defined. Attempts to quantify the uncertainties related to surface heterogeneities and modeling approaches developed to include heterogeneous surfaces within
AMs and ESEMs are then reviewed.
4.1. The problem of surface heterogeneity representation
Land surface heterogeneities occur for three basic reasons: i) variations in vegetation
cover or, more generally, surface type (e.g. different types of vegetation, bare soil, snow,
inland water, urban areas); ii) variations in terrain morphology (e.g. slope and elevation);
and iii) variations in soil characteristics (e.g. color and texture). A further degree of
heterogeneity is added by the climate forcing, which can be highly variable in space and
time. A typical example of this is given by summertime convective precipitation, which
can strongly vary on scales of a few km or even less. All these effects can strongly alter
the local surface energy and water budgets, and although for some aspects they can be
partially correlated, they generally vary quite independently from each other.
This suggests that surface heterogeneity likely spans a wide range of spatial scales.
Therefore, although based on one of the criteria above it might appear that the surface is
a mosaic of well defined patches, some characteristics based on other criteria may produce
within each patch a high degree of variability. An enlighting example is that of A vissar
et a!. (1991), who measured the plant stomatal resistance in a homogeneous potato field.
They found that the stomatal resistance followed a quasi log-normal distribution, mostly
in response to variations of the leaf micro-environment (inclination, orientation, shading,
level within the canopy, wind). For larger patches, e.g. a forest, where different species
share the same environment and where terrain and soil texture can significantly vary, this
effect could be far more pronounced.
These considerations allow us to define two types of heterogeneity, what we can call
inter-patch and intra-patch heterogeneity. The separation between inter- and intra-patch
heterogeneit.y cannot be strictly based on spatial scale, since we have commented that
heterogeneity likely varies on a continuum of scales, but rather on the mathematical
approach most suitable to treat them. Inter-patch heterogeneity can be treated with
"discrete" methods, while "intra-patch" heterogeneity requires continuous approaches.
The effects that sub-grid scale surface heterogeneity has on an AM can also be roughly
divided in two categories, what we here refer to as "direct" and "indirect" effects. Direct