46
Chapter 3: Climate Spectra and Stochastic Climate Models
significantly contribute to ENSO, even though the triggering effects of equatorial Kelvin wave generated by westerly wind bursts have been documented
in the observations.
On longer time scales, the SST anomaly spectra become red again, suggesting a significant interdecadal variability (Figure 3.11). This redness is linked
to that of atmospheric variables (Figure 3.5), and it may be in part governed by basin-scale dynamical interactions between the large scale oceanic
circulation and the atmosphere (Deser and Blackmon, 1993; Kushnir, 1994).
3.5 Variability of Other Surface Variables
Delworth and Manabe (1988) have shown that soH moisture anomalies in an
atmospheric GCM having a simple soil representation had the red spectrum
(3.14) and were well-represented by the stochastic climate model. The soil
acted as an integrator of the white noise forcing by rainfall and snowmelt,
and the feedback time increased with latitude, except in regions of frequent
runoff. This was explained by considering the dissipation processes that limit
soil wetness: as the energy available for evaporation decreases with increasing
latitude, potential evaporation, hence soil moisture damping decreases. However, if precipitation exceeds potential evaporation, the excess precipitation
is removed by runoff when the soil is saturated, and >. increases. Although
other factors (vegetation, varying soil characteristics, subsurface water flow)
need to be considered, Delworth and Manabe's (1988) interpretation seems
sturdy: soil moisture anomaly observations in the Soviet Union are indeed
well-modeled by a first-order Markov process, with a damping time approximately equal to the ratio of field capacity to potential evaporation (Vinnikov
and Yeserkepova, 1991).
On monthly to yearly time scales, sea ice anomalies in the Arctic and
Antarctic have been shown by Lemke et al. (1980) to be reasonably wellrepresented by an advected linearly damped model forced by stochastic
weather fluctuations. On longer time scales, larger sea ice fluctuations occur
in the Arctic (e.g., Stocker and Mysak, 1992). Decadal sea ice fluctuations in
the Greenland and Labrador seas are related to high latitude surface salinity
variations (cf. the "Great Salinity Anomaly" in the northern North Atlantic),
and appear to be lagging long term changes in the Northern Hemisphere atmospheric circulation (Walsh and Chapman, 1990). Mysak et al. (1990) have
shown that the sea ice anomalies mainly lag the sea surface salinity anomalies and suggested that the latter had been caused by earlier run offs from
northern Canada into the western Arctic Ocean. They then postulated the
existence of a complex climate cycle in the Arctic involving advection along
the subpolar gy re and changes in the convective overturning in the Greenland
sea, but the role of the stochastic forcing by the atmosphere was not assessed.
Chapter 3: Climate Spectra and Stochastic Climate Models
significantly contribute to ENSO, even though the triggering effects of equatorial Kelvin wave generated by westerly wind bursts have been documented
in the observations.
On longer time scales, the SST anomaly spectra become red again, suggesting a significant interdecadal variability (Figure 3.11). This redness is linked
to that of atmospheric variables (Figure 3.5), and it may be in part governed by basin-scale dynamical interactions between the large scale oceanic
circulation and the atmosphere (Deser and Blackmon, 1993; Kushnir, 1994).
3.5 Variability of Other Surface Variables
Delworth and Manabe (1988) have shown that soH moisture anomalies in an
atmospheric GCM having a simple soil representation had the red spectrum
(3.14) and were well-represented by the stochastic climate model. The soil
acted as an integrator of the white noise forcing by rainfall and snowmelt,
and the feedback time increased with latitude, except in regions of frequent
runoff. This was explained by considering the dissipation processes that limit
soil wetness: as the energy available for evaporation decreases with increasing
latitude, potential evaporation, hence soil moisture damping decreases. However, if precipitation exceeds potential evaporation, the excess precipitation
is removed by runoff when the soil is saturated, and >. increases. Although
other factors (vegetation, varying soil characteristics, subsurface water flow)
need to be considered, Delworth and Manabe's (1988) interpretation seems
sturdy: soil moisture anomaly observations in the Soviet Union are indeed
well-modeled by a first-order Markov process, with a damping time approximately equal to the ratio of field capacity to potential evaporation (Vinnikov
and Yeserkepova, 1991).
On monthly to yearly time scales, sea ice anomalies in the Arctic and
Antarctic have been shown by Lemke et al. (1980) to be reasonably wellrepresented by an advected linearly damped model forced by stochastic
weather fluctuations. On longer time scales, larger sea ice fluctuations occur
in the Arctic (e.g., Stocker and Mysak, 1992). Decadal sea ice fluctuations in
the Greenland and Labrador seas are related to high latitude surface salinity
variations (cf. the "Great Salinity Anomaly" in the northern North Atlantic),
and appear to be lagging long term changes in the Northern Hemisphere atmospheric circulation (Walsh and Chapman, 1990). Mysak et al. (1990) have
shown that the sea ice anomalies mainly lag the sea surface salinity anomalies and suggested that the latter had been caused by earlier run offs from
northern Canada into the western Arctic Ocean. They then postulated the
existence of a complex climate cycle in the Arctic involving advection along
the subpolar gy re and changes in the convective overturning in the Greenland
sea, but the role of the stochastic forcing by the atmosphere was not assessed.