1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>Spatial</strong> <strong>Modelling</strong> <strong>of</strong> <strong>the</strong> <strong>Terrestrial</strong> <strong>Environment</strong>: Outlook 265<br />
erosion and Burke et al. show how vegetation cover not only affects soil moisture status<br />
but also must be accounted for when soil moisture retrievals are conducted. For estimates<br />
<strong>of</strong> snow pack properties (such as snow depth), Kelly et al. note that forest cover influences<br />
<strong>the</strong> accuracy <strong>of</strong> retrievals from passive microwave instruments. In all <strong>of</strong> <strong>the</strong>se examples,<br />
small errors in vegetation parameter estimation lead to significant errors in <strong>the</strong> retrieved or<br />
<strong>the</strong> modelled variable.<br />
Vegetation characterization can be achieved in many ways. The most usual remote sensing<br />
tool is <strong>the</strong> multispectral scanning instrument such as Landsat or <strong>the</strong> Advanced Very High<br />
Resolution Radiometer (AVHRR). Enhanced instruments have more recently been used for<br />
vegetation mapping and potentially enable <strong>the</strong> improved characterization <strong>of</strong> different vegetation<br />
parameters. For example, NASA’s Moderate Resolution Imaging Spectroradiometer<br />
on board <strong>the</strong> Terra and Aqua platforms has at least twice <strong>the</strong> fineness <strong>of</strong> spatial resolution<br />
<strong>of</strong> <strong>the</strong> AVHRR with a near daily repeat and 32 (mostly narrow) spectral bands in <strong>the</strong> visible<br />
and infra-red part <strong>of</strong> <strong>the</strong> electromagnetic spectrum. This is ideal for classical vegetation<br />
classification and leaf area index mapping. With better specified instruments for particular<br />
science applications, and with <strong>the</strong> increased understanding <strong>of</strong> instrument synergisms<br />
for particular applications (for example, combining optical reflectance signals with radar<br />
canopy geometry models to produce vegetation canopy type condition and physical geometry),<br />
more diverse and accurate vegetation parameters will undoubtedly be available for<br />
modellers.<br />
13.2.3 <strong>Spatial</strong> Resolution: Scales <strong>of</strong> Variation and Size <strong>of</strong> Support<br />
<strong>Spatial</strong> resolution is a cross-cutting issue running throughout many <strong>of</strong> <strong>the</strong> book contributions<br />
(especially in <strong>the</strong> chapters by Okin and Gillette, Kelly et al., Drake et al., Wooster<br />
et al., Barr and Barnsley). It is a multi-dimensional issue that is important for several<br />
different reasons. <strong>Spatial</strong> resolution refers to <strong>the</strong> spatial resolving power <strong>of</strong> <strong>the</strong> desired<br />
activity. In <strong>the</strong> case <strong>of</strong> remote sensing it usually refers to <strong>the</strong> size <strong>of</strong> instantaneous field <strong>of</strong><br />
view <strong>of</strong> <strong>the</strong> sensor (which generally translates to a pixel) but in spatial modelling it could<br />
easily refer to <strong>the</strong> cell size used for <strong>the</strong> modelling framework. Both definitions should be<br />
determined for <strong>the</strong> modeller by <strong>the</strong> scale(s) <strong>of</strong> variation <strong>of</strong> <strong>the</strong> environmental variable under<br />
consideration. However, spatial resolution <strong>of</strong> remote sensing instruments is technologically<br />
determined and until recently, spatial resolution <strong>of</strong> <strong>the</strong> cell grid in numerical modelling has<br />
been determined by <strong>the</strong> computational hardware available, i.e. processor speed, memory<br />
and storage. With advancements in both computer and remote sensing technologies, <strong>the</strong><br />
spatial resolution issue is becoming controlled more by limitations in our understanding<br />
<strong>of</strong> <strong>the</strong> scale <strong>of</strong> variation <strong>of</strong> an environmental variable than by technological limitations.<br />
This is an important development because <strong>the</strong> scale <strong>of</strong> variation <strong>of</strong> an environmental variable<br />
should control <strong>the</strong> way that modellers discretize space. In <strong>the</strong> past, <strong>the</strong> technological<br />
constraints have determined <strong>the</strong> size <strong>of</strong> support (or spatial framework) used by <strong>the</strong> model<br />
and have resulted in sometimes very inappropriate spatial models. We are now at a stage<br />
<strong>of</strong> technological development when scientists can <strong>of</strong>ten define (at least <strong>the</strong>oretically) levels<br />
<strong>of</strong> spatial resolution that can reflect <strong>the</strong> natural scale <strong>of</strong> variation <strong>of</strong> an environmental<br />
variable. However, this statement assumes that scientists can accurately define <strong>the</strong> spatial<br />
scale <strong>of</strong> variation <strong>of</strong> a desired variable but in many cases this is only possible with a high<br />
degree <strong>of</strong> uncertainty. For example, in <strong>the</strong> chapter by Kelly et al., <strong>the</strong> scales <strong>of</strong> variation