22.03.2015 Views

1 Spatial Modelling of the Terrestrial Environment - Georeferencial

1 Spatial Modelling of the Terrestrial Environment - Georeferencial

1 Spatial Modelling of the Terrestrial Environment - Georeferencial

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

70 <strong>Spatial</strong> <strong>Modelling</strong> <strong>of</strong> <strong>the</strong> <strong>Terrestrial</strong> <strong>Environment</strong><br />

soil moisture (0–5 cm) and deeper soil moisture (5–120 cm) predicted using MICRO-<br />

SWEAT derived over one year for a typical crop growing in sandy soil using hourly driving<br />

data from <strong>the</strong> SGP97 study area. There is a significant, but noisy, relationship that is<br />

structured around several separate linear correlations, which are <strong>the</strong>mselves related to <strong>the</strong><br />

water content in <strong>the</strong> soil pr<strong>of</strong>ile immediately after <strong>the</strong> most recent precipitation event. (Note:<br />

at <strong>the</strong> time <strong>of</strong> precipitation <strong>the</strong> modelled relationship between surface and deep soil moisture<br />

depends strongly on <strong>the</strong> amount <strong>of</strong> precipitation and modelled run<strong>of</strong>f, but once precipitation<br />

has ceased, <strong>the</strong> modelled relationship between <strong>the</strong> surface and deep soil moisture changes<br />

mainly as a result <strong>of</strong> <strong>the</strong> surface soil drying.) Figure 4.4b shows <strong>the</strong> good agreement between<br />

<strong>the</strong> MICRO-SWEAT modelled values <strong>of</strong> <strong>the</strong> ratio <strong>of</strong> near-surface to deep-soil moisture and<br />

<strong>the</strong> values calculated using a simple derived relationship that acknowledges <strong>the</strong> effect <strong>of</strong><br />

recent precipitation.<br />

O<strong>the</strong>r methods <strong>of</strong> extending estimates <strong>of</strong> surface soil moisture deep in <strong>the</strong> soil pr<strong>of</strong>ile<br />

generally involve assimilation techniques, such as direct insertion <strong>of</strong> <strong>the</strong> surface soil moisture<br />

(Walker et al., 2001) or use <strong>of</strong> <strong>the</strong> extended Kalman filter (Walker et al., 2001; Walker<br />

and Houser, 2001; Hoeben and Troch, 2000) or variational assimilation (Reichle et al.,<br />

2001). Walker et al. (2001) compared <strong>the</strong> direct insertion and Kalman filter assimilation<br />

methods using syn<strong>the</strong>tic data and demonstrated that using <strong>the</strong> Kalman filtering is superior,<br />

with correction <strong>of</strong> <strong>the</strong> soil moisture pr<strong>of</strong>ile being achieved in 12 hours as compared to<br />

8 days or more with direct insertion. Variational data assimilation is computationally more<br />

effective than using <strong>the</strong> Kalman filter, but variational assimilation requires an adjoint model<br />

that is numerically well behaved and <strong>the</strong>re are currently no adjoint models available for <strong>the</strong><br />

commonly used land surface models.<br />

Assimilation procedures ei<strong>the</strong>r introduce retrieved soil moisture into <strong>the</strong> land surface<br />

model (Walker et al., 2001; Montaldo et al., 2001; Hoeben and Troch, 2000; Wigneron<br />

et al., 1999; Li and Islam; 2002, Calvet and Noilhan, 2000), or <strong>the</strong> measured microwave<br />

brightness temperatures is itself assimilated through <strong>the</strong> use <strong>of</strong> a coupled land surface<br />

and microwave emission model (Crosson et al., 2002; Galantowicz et al., 1999). They<br />

generally build on information already present in a land surface model and, in general,<br />

result in improved pr<strong>of</strong>ile soil moisture estimates by <strong>the</strong> land surface model, regardless <strong>of</strong><br />

<strong>the</strong> assimilation methods used. It has been demonstrated (Walker and Houser, 2001) that<br />

through <strong>the</strong> assimilation <strong>of</strong> near-surface soil moisture observations, errors in forecast soil<br />

moisture pr<strong>of</strong>iles that result from poor initialization may be removed, and <strong>the</strong> resulting<br />

predictions <strong>of</strong> run<strong>of</strong>f and evapotranspiration by a hydrological model improved.<br />

4.3.3 Subpixel Heterogeneity<br />

In <strong>the</strong> near future, any passive microwave satellite mission is likely to measure with resolution<br />

between approximately 30 and 60 km. At this resolution, <strong>the</strong> land surface is strongly<br />

heterogeneous and <strong>the</strong> impact <strong>of</strong> this heterogeneity on <strong>the</strong> accuracy <strong>of</strong> <strong>the</strong> retrieved soil<br />

moisture is a significant issue.<br />

In <strong>the</strong> case <strong>of</strong> a heterogeneous bare soil, both Burke and Simmonds (2002) and<br />

Galantowicz et al. (2000) demonstrated that errors in <strong>the</strong> retrieved soil moisture are negligible.<br />

For a pixel with heterogeneous vegetation, Drusch et al. (1999), Liou et al. (1998),<br />

Crow et al. (2001), Crow and Wood (2002), and Burke and Simmonds (2002) suggested<br />

that errors in <strong>the</strong> retrieved soil moisture are generally less than 0.03 cm 3 cm −3 . This is a

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!