van den Dool [1990] correctly cautions against equating available moisture and precipitation. Using National Meteorological Center (NMC) assimilated observations on a regular grid, he shows that the climatology of vertical motion is clearly the most important factor for precipitation formation. The vertical motion and instability in the atmosphere are also influenced by the thermal regime at the land surface. There are numerous other important factors as well. For example, the dust and aerosol source found in large urban areas (as well as the heat island) contribute to the formation of precipitation. Changnon et al. [1991] demonstrate that down-wind of St. Louis, there is 17% more precipitation accumulation and 13% more precipitating events.
Zhao and Khalil [1993] explore the thermal regime-precipitation relationship by correlating monthly temperature and precipitation for the 1905-1984 climate records at 1000 surface stations in the continental United States. They find a strong negative correlation between summertime precipitation and temperature. This relationship is strongest in the Central United States and over the Great Plains. There is a mutual interaction between land and atmosphere evident in this observation. The establishment of a high pressure ridge over the continent inhibits precipitation due to the sinking motion in the air column. The temperature is elevated for the same reasons as those found to be the bases of subtropical climate regimes. In turn, the drier surface experiences a deficit in cooling by latent heat and the surface temperatures increases. The pressure ridge is intensified and dry convection extends the adiabatic layer to inhibit cloud formation. This cycle and mutual land-atmosphere reinforcement of the warm-dry summer conditions becomes evident in the correlations such as those reported by Zhao and Khalil [1993]. Huang and van den Dool [1993] extend the analysis to include lagged correlation. Since they find significant correspondence when negative precipitation anomaly precedes above-average summer temperature by one month, the hypothesis of surface soil moisture deficit forcing persistent warm-dry summer conditions may be the dominant mechanism in the mutual land-atmosphere interaction.
Modeling studies also support this finding. Atlas et al. [1993] use a GCM to model the evolution of the surface preconditions to the 1988 drought over large parts of the mainland United States territory. They use late spring-early summer sea-surface temperature and soil moisture anomalies to force the GCM. They find that the initial soil moisture anomaly are especially important to maintain the dry and warm summer of 1988. Furthermore, Oglesby [1991] finds that a March soil moisture anomaly is rapidly erased whereas an early summer, month of May anomaly persists through the summer and forces the North American drought.
Analogous modeling studies for other regions also indicate similar feedback mechanisms. Xue and Shukla [1993] use a GCM to investigate the surface forcing of the Sahel region drought. Following the desiccation of the surface through the removal of the soil moisture, the moisture convergence and precipitation over West Africa are reduced significantly. Consistent with observations during drought years, this surface forcing results in a weakening of the 200 mb tropical easterly jet, a strengthening of the mid-tropospheric (700 mb) African easterly jet and a reduction in the intensity of the easterly wave disturbances that force precipitation over the region. An anomalous descending motion over the dry area persists throughout the rainy season. Bounoua and Krishnamurti [1993a,b] additionally find that anomalous north-south propagation of the heat source following the rainfall season remarkably affects the precipitation, moisture convergence and the regional hydrologic cycle.
These phenomena over the Sahel are documented, based on observations, by Nicholson [1993]. Nicholson [1993] shows that the intense drought in West Africa, as revealed in an extensive precipitation observation data set, has extended to 1990 from the date of its establishment around 1970. Furthermore the pattern of opposition between equatorial and subtropical latitudes continues [ Nicholson, 1993]. A shift and weakening in the Hadley cell (a meridional thermally-direct circulation between the equator and subtropical latitudes) and Inter-Tropical Convergence Zone (near-continuous low pressure belt across the equator), bring negative anomalies to the mean rising motion over the region during the rainy season.
Additional evidence, albeit based on modeling results, for the positive
surface feedback that forces the persistence of precipitation anomalies may
be found in deforestation experiments. Nobre et al. [1991] show that
when the land surface in the GCM is converted from tropical forest to grass
over portions of the Amazon, the temperature is increased by 2.5
C mostly as the result of the 30% reduction in surface evaporation. The
precipitation is lower by 25% but in total amounts the reduction in
precipitation is larger than the reduction in evaporation. This suggests
that the vapor convergence and precipitation efficiency for the overlying
atmosphere have changed as well. Eltahir and Bras [1993b] show that
the response of the tropical atmosphere to deforestation is composed of two
competing factors. The increased surface heating induces more intense
thermal convection but the reduced precipitation leads to a negative anomaly
in the heating of the atmospheric column by droplet condensation. The
sensitivity of the land-atmosphere system is thus reduced as a result of the
competing factors.
The physical mechanisms governing the atmospheric response to surface anomalies are partly illustrated by Cook [1994] where he places one simple continent centered at the equator in a GCM. Longitudinal precipitation gradients develop. When the continent is relatively dry and soil water is limiting, the precipitation rates are lower in the center and west of the continent. The eastern portion of the continent is strongly influenced by the easterlies that bring oceanic moisture to the continent at the equator. Cook [1994] finds that dry land surfaces force dry convection and reduce the low-level atmospheric humidity. Eventually these two processes inhibit precipitation. In a related modeling study, Cook and Gnanadeskian [1991] show that dry conditions create more unstable conditions, but the dry convection limits condensation up to 830 millibars and hence inhibits significant precipitation.