| Home > Newsevents > Training > Rcourse_notes > PARAMETRIZATION > SURFACE_ASSIMILATION > |
Land surface assimilation
March 2001
By Jean-François Mahfouf and Pedro Viterbo
European Centre for Medium-Range Weather
Forecasts
Table of contents
Next Section
Previous Section
6 Other techniques to initialise soil moisture
6.1 Methods based on precipitation data
Two methods have been introduced to initialise soil moisture from precipitation data, both of them requiring the availability of measurements over large areas and an algorithm to perform a precipitation analysis beforehand. They have only been applied over areas with a good observational coverage like the US or the UK.
The first technique is an uncoupled initialisation where a land surface scheme is forced with conventional meteorological observations (temperature, humidity, wind, radiation and precipitation) to provide an estimate of soil moisture. Feasibility studies have been undertaken in various limited area models by Smith et al. (1994), Macpherson (1996) and Mitchell (1994). The improvement of the forecasts of screen level humidity when soil water is initialized with such a method, using the UK Met Office water budget scheme, is shown on Fig. 7 (from MacPherson 1996) for all UK stations.
Another technique makes use of both observed precipitation rates and model first-guess. Assuming a rainfall rate increment over a 6-hour period:
|
(11) |
This quantity is converted in soil moisture increment by using the tangent linear model of a soil moisture budget scheme:
|
(12) |
where E is the mean evaporation rate and
the mean soil moisture content during the 6-hour assimilation period (trajectory).
Then, the soil moisture increment 
is added to the superficial reservoir. Studies have been
undertaken at ECMWF by Vasiljevic (1989, personal communication) and at
UKMO by Jones and Macpherson (1995). Such a method
is sensitive to the specification of surface run-off and can converge slowly
when biases in the root zone are large.6.2 On-site observations and methods based on satellite imagery
Existing techniques for ground based observations of soil moisture (see recent reviews in Schulin et al. 1992; Wei 1984) are time consuming and normally require human intervention. The representativeness error of the on-site estimates is best avoided by deploying several instruments within a relatively small area ( 100 m2), increasing the cost of the measurements. In spite of its problems, on-site soil moisture data are very useful for regional esimates for climatic studies, essential to close the water budget in large-scale hydrologic experiments (Cuenca and Noilhan 1991; Goutorbe et al. 1989; Mahfouf 1990) and to calibrate remote sense retrieval techniques (Georgakakos and Baumer 1996).
There is no prospect of obtaining real-time global estimates of soil moisture based on existing technology of ground based instruments. For this reason, several algorithms have been developed to infer soil moisture from satellite observations, although none of them is currently used in an operational data assimilation system. Three types of techniques have been proposed (see reviews in Paloscia 1996; Schulin et al. 1992; Wei 1984) based on infrared measurements, passive microwave and, more recently, active microwave (radar) instruments.
In the infrared channels, the sensitivity of the diurnal cycle of surface temperature to soil moisture has been used to define methods based on the observed changes on the infrared skin temperature (which avoid the problem of absolute calibration of the satellite sensor). For reviews of applications see Carlson (1991), Schmugge and Becker (1991), and Schulin et al. (1992). Geostationary satellites allow for a better temporal sampling (Wetzel et al. 1984 ; McNider et al. 1995). These methods can only be applied in clear sky conditions but provide an information about soil moisture in the root zone over vegetated areas. Bastianssen (1995) developed recently a technique to estimate regional evaporation over heterogeneous terrain, based on a separate estimate of the evaporation of unstressed pixels, based on potential evaporation, and fully stressed pixels, based on the infrared diurnal cycle technique. The evaporation of the remaining cloud-free pixels can be obtained by interpolation between the wet pixels and the dry pixels. van den Hurk et al. (1997) has applied this technique to initialize the soil water of a limited area model over the Iberian peninsula. The model soil moisture is the linearized solution of a variational problem that minimizes the difference between model and satellite estimates of evaporative fraction
).Microwave channels can be used to infer soil moisture due to the important variations of the dielectric constant of a soil with volumetric water content for frequencies between 1 and 5 GHz (Schmugge and Jackson 1994). Passive microwave techniques use the fact that soil emissivity changes with its water content. In active microwave sensors (radar) the signal is emitted by an artificial source and the intensity of the backscattered radiation, after reflection by the surface, is measured. The reflectivity of the soil changes with its water contents, hence the intensity of the reflected signal can be related to the soil moisture. Active microwave systems allow, for the same wavelength (same maximum penetration depth), a finer horizontal resolution, because the ground can be scanned with an angularly confined beam. One of the drawbacks of microwave retrievals is that the surface emissivity/reflectivity is also sensitive to the surface roughness and the water contents of the vegetation canopy. Nevertheless, it appears that simple estimates of surface roughness of broad vegetation classes are sufficient to correct the soil moisture estimate (Njoku and Entekhabi 1996). On the other hand, the oppacity of the vegetation layer increases with its water content, making the corrections due to vegetation increasingly unreliable for moist soils. Perhaps the major drawback of microwave estimates is the depth of penetration of the signal, limited to the top layer of the soil (2 to 10 cm, depending on the wavelength). However, for specific soil hydrological and atmospheric conditions, the soil water contents of the root layer is correlated with the top soil water. Recent studies show that it is possible to infer, in a physically consistent way, the whole profile of soil water from its values at the top layer (e.g. Njoku and Entekhabi 1996; Calvet et al. 1998).
We have shown in this section that ground-based estimates of soil moisture, although very important for calibration purposes and in intensive field efforts, cannot give a near real-time global stimate of soil moisture. Satellite estimates can achieve global coverage, but are limited to clear-sky conditions (infrared channels) or sense only the top few centimetres of soil (microwave channels). The future relies on physically based estimates of soil moisture from a combination of satellite measurements and model short-term forecasts, using a variational technique in order to find the soil water contents that fits best the satellite signal.
Training Course Notes Front Page
Table of contents
Next Section
Previous Section
 |
12.06.2002 |
 |
© ECMWF |
