| Home > Newsevents > Training > Rcourse_notes > PARAMETRIZATION > SURFACE_ASSIMILATION > |
The role of the land surface in
the climate system
April 2002
European Centre for Medium-Range Weather
Forecasts
Table of contents >>
Next Section >>
Previous Section >>
3 . Time scales and the role of soil moisture
3.1 Global time scales
Fig. 2 represents the size of the moisture reservoirs of the terrestrial atmosphere and the marine atmosphere (rectangles in the figure), the exchanges of moisture between them, and between the atmosphere and the surface below (arrows in the picture). The sea surface evaporates at the potential rate, while over land there are additional mechanisms that reduce the evapotranspiration rate: dryness of the soil or, over vegetated areas, physiological mechanisms that can reduce or shut transpiration from the plant leaves and trunks making the water from the root zone effectively unavailable for the atmosphere above. Precipitation over land is about a quarter of that over sea. Note that precipitation exceeds evaporation over land, while over sea the reverse is true. In order to have a closed budget for the terrestrial atmosphere, advection of moisture across a vertical wall projecting over the continent boundaries has to match the difference precipitation minus evaporation. Advection is roughly half of the water evaporated over land (see Peixoto and Oort 1992 for estimates based on radiosonde observations), suggesting an annual recirculation ratio (ratio of the rainfall coming from local evaporation over total rainfall rate) of 67% (71/107). To close the hydrological cycle, the advection has to be matched by the river runoff: the global averaged influx of fresh water into the ocean is estimated in this way as 36 x 1015 kg yr-1. For continental areas, annual runoff, evaporation and precipitation are approximately in the ratio 1:2:3.
Rigorous formulations of the atmospheric branch of the hydrological cycle can be found in Peixoto 1973 and Peixoto and Oort 1983, 1992. So-called "aerological estimates of runoff", based on measurements of atmospheric water vapour transport and a closure at the surface, have been applied successfully to basins with areas larger than 106 km2 (Rasmusson 1967, 1968, 1971; Peixoto 1973).
The rainfall rate and the size of the reservoir can be combined to give a time scale 4.5/107 = 0.042 years = 15 days: the terrestrial atmosphere would be emptied by rainfall in 15 days. In a similar way the reservoir would be replenished by surface evapotranspiration in 23 days (4.5/71 years)1. The time scales associated to marine rainfall are only 7.5 days, and the corresponding value for evaporation is 6.8 days. This suggests (a) a more vigorous hydrological cycle over the ocean; (b) a land surface control over large time scales (weeks to months), through the evapotranspiration flux of water at the surface. The implication of the above on the extended predictability of the atmosphere due to exchanges of water with the land surface has been discussed by many authors (see, e.g., Namias 1958, Mintz 1984, Dümenil and Bengtsson 1993, Dirmeyer and Shukla 1993).
The accuracy of the numbers shown in Fig. 2 varies widely: see Chahine 1992, for the sources used to produce these particular estimates. Any literature review shows a very large dispersion in those numbers (see, e.g., Viterbo 1996): global estimates of the total column water vapour can vary by as much as 34%, while runoff estimates differ by 45%. Independent estimates of precipitation have smaller ranges of uncertainty (notwithstanding the extensive areas of the planet where observations are very scarce), but direct or indirect estimates of evaporation are subject to very large uncertainty.
3.2 The role of soil moisture
In order to illustrate the role of soil moisture in shaping the interaction surface-atmosphere, we will use model data over the Red-Arkansas River basin, a sub-basin of the Mississippi River basin. Data is based on the ECMWF reanalysis ERA15 (Gibson et al. 1997) used also in Section 2. Betts et al. (1998, 1999) have studied the ECMWF model energy and water budget over the Mississippi River basin, as described by short-term forecasts. Basin-average data was used for nine years, 1985-93 and analysed at different temporal scales; we will concentrate in this sectio on summertime 5-day average ECMWF data.
We will illustrate here that the ECMWF data shows a similar coupling in the midsummer on the 5-day timescale between soil water, evaporation and low-level thermodynamics as the First International Satellite Land Surface Climatology Project Field Experiment (FIFE) data2 (Betts and Ball 1998) show on the diurnal time scale. We define a 5-day mean evaporative fraction as
 .
|
where H and LE are the 5-day averages of the surface sensible and latent heat fluxes. Fig. 3 (a) shows the strong coupling between EF and the top layer model soil water, SW1. The data plotted are all the 5-day values (1985-1993) for which the 0-7 cm mean soil temperatures are > 296 K, representative of the warm months.
Fig. 3 (b) shows a similar coupling of PLCL (the pressure height of the lifting condensation level, determining cloud base) to SW1 for the same data. The model resistance to evaporation between the saturated interior of a "leaf" and the surrounding air is dependent on soil water, and this vegetation resistance is therefore one key factor in determining the equilibrium saturation level difference, PLCL, in the saturation pressure budget of the BL (Betts and Ball 1998).
We have shown 5-day averages, but the patterns and slopes in Figs. 3 (a) and (b) are similar (but shifted slightly to higher PLCL and lower EF) if the 12-h daytime average are used instead. Note that the lower limit in Fig. 3 (b) (corresponding to very wet soils) is near the oceanic equilibrium of
hPa (e.g., Betts and Ridgway 1989). The oceanic surface
boundary is saturated, and has no additional resistance of evaporation corresponding
to the vegetative resistance over land.
Figs. 3 (a) and (b) are particularly significant because neither of
these relationships of EF and PLCL on soil water
depend strongly on soil temperature at this warm temperatures.
3.3 The interplay between the diurnal and the seasonal time scales
Fig. 4 (a) (from Betts et al. 1996) shows the daytime diurnal cycle of the FIFE 2-m thermodynamic data for the predominantly sunny and dry days from May to October 1987. The axes are potential temperature (
) and mixing ratio (q). This
(, q) plot can be regarded as the heat and moisture
budget on orthogonal axes (Betts 1992). There are 19, 21, 25, 22,
23, and 22 dates in each average from May to October. The selection criteria
were near-noon surface net radiation above a threshold (which was 450 W
m-2 in midsummer, falling to 300 W m-2 in October)
and no significant daytime rainfall. Here we can see the diurnal and seasonal
cycle together. The points are plotted hourly, starting at 1145 UTC, shortly
after sunrise in midsummer. The seasonal rise and fall of mean temperature
and mixing ratio can be seen: July is the warmest month. October is noticeably
drier, after the vegetation has diedand evaporation is low. Saturation pressure
lines of 970 and 800 hPa are shown dashed. The surface pressure is near
970 hPa. It can be seen that at the morning minimum temperature, the 2 m
air is about 30 hPa from saturation, except in October, when it is more
unsaturated. The diurnal range of mixing ratio q is relatively small
in all months. There is generally a rise of q in the morning, when
the BL is shallow and capped by relatively moist air from the BL of the
preceding day, and a fall in the afternoon, as the growing BL entrains drier
air from higher levels. May shows no afternoon fall of q, probably
because of the higher soil moisture and evaporation. May and June do not
reach as low afternoon saturation pressures as the later months of July,
August, September, and October. This means a lower mean lifting condensation
level (see also previous section) or cloud base in spring. Probably this
reflects a seasonal drying of the surface, although changes in upper air
thermodynamic structure may be involved. It is clear that the afternoon
maximum of equivalent potential temperature e is controlled mostly by the seasonal
shift. The isopleths of e
=310, 330, 350 K are shown dotted. The rise of e from morning minimum to afternoon
maximum is around 14 K in all months. The sum of surface sensible and latent heat fluxes is a surface source for increasing
e (e.g., Betts
and Ball 1998). This surface e is proportional to the sum of H+LE,
and it is not affected by the Bowen ratio. It is entrainment of low e air from above the BL, together with
the deepening of the BL, that reduce the BL e rise,
and so feed back on both the shallow and even more importantly on precipitating
convection. Thus one of the important aspects of the BL evolution over land
is how large entrainment at BL top is. The daytime BL over land is primarily
thermally generated (in strong winds, shear plays a role), and thus linked
to the surface virtual heat flux (which over land is usually dominated by
the sensible heat flux). Hence if the surface Bowen ratio is large, although
the surface e
flux may be unchanged, the large H flux drives more entrainment,
produces a deeper BL, and the diurnal rise of e is reduced. Fig.
4 (b) shows how this diurnal cycle over land depends on soil moisture
and, as a consequence, the surface evaporation. A total of 28 days from July and August 1987 during FIFE, which were affected little by precipitation or cold air advection, were composited by soil moisture (SM: measured gravimetrically in the top 10 cm). The points are hourly values from 1115 UTC (near sunrise) to 2315 UTC. The dry soil composite (SM = 13%, for which the measured mean surface Bowen ratio at noon was 0.8) reaches a higher afternoon maximum
K (the e
isopleths are shown dotted). In contrast, the wet soil composite (SM = 23.4%),
for which Bowen ratio at noon was 0.4), reaches a much cooler afternoon
maximum, but a much wetter q value, so that
the afternoon 
K. Some of this shift of
e is associated with the shift of the entire diurnal path
of higher q with higher soil moisture, but about half is the result
of reduced entrainment of dry low e into
the BL. Over wet soils, H is much reduced and the BL deepens less
rapidly. For all the three composites, the surface available fluxes (net
radiation minus ground heat flux) were nearly identical, so that the surface
e fluxes
were similar. This local feedback between soil moisture, evaporation and
afternoon e equilibrium probably
produces on large spatial scales a positive feedback between soil moisture
and precipitation, which has been the subject of much research (see, for
example, Brubaker et al. 1993, Trenberth
1999). The analogy over the tropical oceans is the link between BL and sea
surface temperature (SST), which influences the prevalence of deep convection
over warm water. Over land variations in soil moisture can lead to as large
variations in BL e as larger as those
produced by several degrees of change in SST. On continental scales, higher
soil moisture and higher evaporation over land would lead to a higher afternoon
e maximum relative to
the surrounding ocean and shift more of the global precipitation over the
continents (Betts et al. 1996). This feedback
has been seen in global models (Mintz 1984). On the regional scale, the
Mississippi flood case study, presented in Section 5.1 below, suggested that
the multiple BLs over the Midwestern United States controlled the location
of precipitation, rather than this mechanism. 3.4 A schematic view of the role of land surface
To conclude this section, a schematic description of the interactions between the surface and the atmosphere will be presented. Inspired by an early, much more complex diagram by Horton (1931), Dooge (1992) (see also Kuhnel et al. (1991)) summarised the interaction between the land surface and the atmosphere in the picture reproduced (with adaptations) in Fig. 5 . The diagram illustrates the behaviour of the soil and the atmosphere within a complete cycle composed of a wet period followed by a dry period. Let us start just after a long episode of rainfall, point A in the figure. The soil water is available in abundance in the root layer3 and its evolution is going to be determined by evaporation. While the soil has plenty of water, the rate of evaporation is controlled by the near-surface atmospheric moisture: the regime is controlled by the atmosphere and the evaporation is at the potential rate. Below a certain level of soil moisture (point B in the picture), physiological mechanisms will limit the supply of water from the root layer into the atmosphere, and the evaporation will drop below its maximum value (potential evaporation, Epot). The regime is under a vegetation (soil) control. When precipitation starts (points C) it will meet a soil dry enough during the initial stages, so that infiltration (If, that part of water that falls as precipitation and is effectively collected by the soil for future use) will equal precipitation. The evolution of water in the soil is once more atmospheric controlled, via the rate of precipitation. Beyond a certain value (point D), the soil does not have the ability to soak all precipitation, some of it goes into runoff. This last phase is again soil controlled; the state of the soil determines the rate of infiltration.
Land surface parametrizations have to represent correctly the surface fluxes and the evolution of soil moisture in all four phases of the cycle, and to switch from the atmospheric control into the soil control regime. The evolution of soil moisture will determine when point D will occur, and the evaporation formulation will determine point B. The crucial areas, from the point of view of the atmosphere, are the quadrants BC and CD. During spring and summer (where the atmospheric demand can be very large), the system remains much longer in the half-circumference BCD than in the opposite part of the cycle.
Training Course Notes Front Page >>
Table of contents >>
Next Section >>
Previous Section >>
1 The global view on the atmospheric water budget presented in the text can be regionalised, allowing, under certain restrictive assumptions, an estimate of how much moisture that precipitates comes out from local evaporation versus horizontal transport. For a discussion on the "intensity of the hydrological cycle", i.e., the time scales associated to the emptying and repleneshing of the atmospheric water reservoir at different locations on the globe, see Trenberth (1999).
2 FIFE was an experiment measuring the different components of the surface energy and water budget of a 15 x 15 km2 of natural grassland located in Kansas, US, and heavily instrumented over the period 1987-1989.
3 For the purpose of this discussion, field capacity may be considered as the threshold beyond which there is a minimum canopy resistance to evaporation.
Copyright © 2002, ECMWF. All rights reserved.
 |
12.06.2002 |
 |
© ECMWF |
