Climate studies of the hydrological cycle are frequently based on the estimation of the terms of the moisture balance equation using observations of the atmospheric fields. In particular, a common assumption is that the vertically integrated moisture flux divergence, div(Q), equals the Evaporation minus Precipitation (E - P), which is then interpreted as the ground runoff (in minus sign). This work focuses on aspects of the moisture balance in the subtropics, using the state of the art assimilation data bases, e.g. NASA’s reanalysis GEOS-1/DAS dataset, together with other satellite based datasets (e.g. cloudiness from ISCCP, liquid water from SSMI, precipitation from GPCP, and more). A list of acronyms is given in Appendix D. In a case study for a typical eastern Mediterranean storm GEOS-1/DAS was found to be in a good agreement with a mesoscale simulation, and to show also a pronounced improvement over the operational ECMWF analysis (Alpert et al., 1996b). This work is also presented here in Appendix B.
One aspect of the moisture balance in the subtropics is related with the inclusion of liquid water in the moisture budgets. In climate studies, the horizontal transport of condensed water is usually neglected. This is often justified by the much smaller value of the horizontal transport of cloud liquid water (CLW) compared to the horizontal transport of water vapor. A possible exception to this was suggested by Alpert and Shay-El (1993) as an explanation for a paradoxical net moisture sink found over the Arabian-Iraqi desert. This scenario is re-examined in chapter 2 (Shay-El et al., 1998a) by developing a method of global assessment to the CLW content in the atmosphere and is described next.
First, GEOS-1 multi-year data is used to confirm the earlier finding based on ECMWF data of an apparent net sink. A net sink is reflected by a negative moisture flux divergence. In GEOS data assimilation system the vertically integrated moisture flux divergence div(Q) equals the parameterized E - P plus another term called the Incremental Analysis Updates of moisture, or IAU(q). These updates represent the correction of the model state (first guess) by the observational data, and are inserted as a forcing term in the continuous model moisture equation. Averaging over a long period, say 5-10 yr, the updates compensate for biases in the estimates of or any physical process not represented in the atmospheric general circulation model used in the assimilation system, such as the CLW. Here, it is shown that the negative div(Q) over the Arabian peninsula is balanced mainly by negative IAU(q) removing moisture from the region. The moisture fluxes reveal strong convection but without precipitation in a shallow convection cell. Vertical profiles are employed to demonstrate that the moisture removal process is associated with middle and high clouds, and probably with CLW flux divergence.
Second, SSM/I retrievals of cloud liquid water over water bodies, global ISCCP cloud estimates and winds from the GEOS-1 assimilation are used in conjunction to evaluate global CLW transport for 1992. The CLW fluxes are estimated explicitly and globally from ISCCP and SSM/I by using linear regression methods. Areas of significant CLW divergence are found over the eastern coasts of both the US and Asia, in the vicinity of the Gulfstream and Kuroshio currents, as conjectured by Peixoto (1973). In both the Arabian-Iraqi desert and over the Sahara, divergence of vertically integrated CLW flux opposes the convergence of vertically integrated horizontal moisture flux, thus explaining at least partially the paradoxical net sink and source in these regions, respectively. It is also shown, however, that the magnitude of the annual CLW flux estimates as calculated here is too small to play any significant role in the vertically integrated water budget, except perhaps along coastal regions, and over the dry subtropical deserts where precipitation minus evaporation is relatively small.
Unlike the Arabian desert, the Sahara desert was quite surprisingly classified by Starr and Peixoto (1958) and many others as a moisture source region. The question is how can a desert be a source of moisture? In chapter 3 (Shay-El et al., 1998b) the components of the moisture balance equation are therefore calculated for the North African Sahara Desert, based on GEOS-1 data. These include the Evaporation (E), Precipitation (P), moisture flux divergence (div(Q)), and errors associated with the moisture updates IAU(q). The Annual mean div(Q) corresponds well with the results of Vitart et al. (1996), based on NCEP data. IAU(q) reveals a strong moisture source over the Eastern Mediterranean. Over the Sahara Desert the moisture flux was shown to converge through the northern and southern boundaries mainly at low levels (~900 hPa) and to diverge through the eastern and western boundaries at higher levels (~700 hPa).
Area averaging of div(Q) over a box with varying dimensions reveals that it can be classified not as a net source (as calculated by all previous studies), but just the opposite. That means, the Sahara becomes a net sink if the box is small enough (less than 24 deg lon, 9 deg lat, equal to about 2400 km, 1000 km, respectively) and located over the center of the desert. Only if the box is so large to include also the boundaries of the continent then it can be classified as net source or divergence zone. Inspection of the inter-monthly and diurnal variability, as well as the model biases, weakens also the net source argument. It is therefore suggested that the earlier finding of a net source is largely associated with the smoothing of the Water/Land boundary, as well as due to various atmospheric diffusion processes such as the sea-breeze cycle and clouds intrusion and evaporation.
In chapter 4 (Shay-El et al., 1998c) we use the temperature increments of the assimilation data, IAU(T), to study another subtropical puzzle, i.e. the climatic response to dust. Unlike moisture and clouds that are generally parameterized in climate models the radiative effects of dust are ignored. Alpert et al. (1998) suggested that the IAU(T) -- roughly speaking model errors -- might have a component directly related with the missing description for dust effects (this paper is enclosed in App. C). Hence, the tropospheric temperature response to dust is calculated from correlations between dust frequencies over the eastern part of the Atlantic Ocean and the IAU(T). A strong similarity was found between the monthly distributions of the dust frequencies and those of the IAU, with a correlation of r = 0.67 between the latitudinal location of maximum positive IAU over ocean and that of maximum number of dusty days.
The correlations between the IAU inferred heating rates and the number of dusty days for the inter-monthly variation, spatial variation, and total variation, all indicate that for an average dust event, the dust heats the lower troposphere at 850-650 hPa by ~0.2 K/d corresponding to an annual rate of about 6 K/y over the Eastern Atlantic having on the average about 30 dusty days per year. This value is in agreement with calculated heating rates from models. The major advantage of this method, however, is that it suggests, for the first time, a method to deduce the total response of the atmosphere to dust. That means, including both direct and indirect dust effects as well as the potential feedbacks that finally result with the atmospheric response. A comparison with ISCCP low cloudiness shows that the model errors in this region are also correlated with the presence of clouds. The vertical distribution of IAU and the diurnal cycle show that there is also a residual heating associated with the Saharan Air Layer (SAL). Unlike dust, clouds and SAL are present in the assimilation model scheme, and therefore their role in the model errors cannot be readily assessed. On the other hand the dust effect stands out very clearly.
Alpert, P., Y. J. Kaufman, Y. Shay-El, D. Tanre, A. da Silva, S. Schubert, and J. H. Joseph, 1998: Quantification of dust-forced heating of the lower troposphere. Nature, 395, 367-370.
Alpert, P., Y. Shay-El, and A. da Silva, 1996a: Moisture sinks/sources over the Mediterranean and Arabia/Iraqi desert. Preprint, 2nd International Scientific Conference on the Global Energy and Water Cycle, WCRP, June 17-21, Washington D.C., USA, 105-106.
Alpert, P., Y Shay-El, and A. da Silva, 1996b: Evaluation of the GEOS-1 DAS reanalysis data over a mesoscale Mediterranean domain. Proceedings of the 11th Conference on NWP, AMS, Aug. 19-23, Norfolk, VA, 183-185.
Peixoto, J. P., 1973: Atmospheric vapor flux computations for hydrological purposes. WMO Publi. 357, Geneva, Switzerland, 83 pp.
Shay-El, Y., and P. Alpert, 1991: A diagnostic study of winter diabatic heating in the Mediterranean in relation to cyclones. Q. J. R. Met. Soc., 117, 715-747.
Shay-El, Y., P. Alpert, and A. da Silva, 1998a: Preliminary evaluation of cloud liquid water in relation to moisture budget studies employing ISCCP, SSMI and GEOS-1 datasets. Submitted to Annales Geophysicae.
Shay-El, Y., P. Alpert and A. da Silva, 1998b: Reassessment of the moisture source over the Sahara Desert based on NASA reanalysis. J. Geoph. Res. (in press).
Shay-El, Y., P. Alpert, Y. J. Kaufman, D. Tanre, A. da Silva, S. Schubert, and J. H. Joseph, 1998c: Lower-tropospheric response to dust as inferred from correlations between dust frequencies and analysis increments from GEOS-1 multiyear assimilation. Submitted to Tellus (to the special Rossby-100 Symposium issue).
Starr, V. P., and J. P. Peixoto, 1958: On the global balance of water vapor and the hydrology of deserts. Tellus, 10, 189-194.
Vitart, F., A. H. Oort, and K. Mo, 1996: New results on the hydrology of the North African desert, based on the NMC reanalysis. In Proceedings of the 20th Climate Diagnostics Workshop, U.S. Dept. of Commerce/NOAA/NWS, 191-194
Back to List of Publications