Hydrology of Central and Southwest Asia: Connections Between Regional Atmospheric Circulation and Large-scale Climate Variability

Ingrid-based Data Catalog access to selected Central and Southwest Asia hydroclimate data and analyses


The access to the Data Catalog of selected Central and Southwest Asia riverflow, temperature, and precipitation data sets and their reduced space optimal interpolation analyses is here.

Technical details

The data set of 112 monthly river streamflow records for Central and Southwest (CSW) Asia extending from 1894 to 1985, was analyzed (smoothed and interpolated) using the Reduced Space Optimal Interpolation (RSOI) method. The majority of the observational records were available only for the period after 1938. Monthly climatology for all stations were estimated for station-depended subperiods of 1936-1984. Anomalies with regards to this climatology were normalized by the annual climatological streamflow for each station. These normalized anomalies were then subjected to the RSOI analysis procedure. Month-dependent location-by-location covariance matrices and empirical orthogonal functions were computed in order to apply the RSOI to a subset of 84 stations for which these covariance matrices could be computed for each calendar month. A technical novelty of the RSOI procedure was developed and applied: the generalized cross-validation approach to the simultaneous optimal selection of the reduced space dimension and the effective relative observational error. Results of the analysis for 84 stations were presented as normalized anomalies and converted back to the dimensional values with the climatological cycle included.

Locations of 112 river flow measurement stations. Sizes of circles are proportional to drainage areas




Climatological total discharge rates for different stations vary by 4 order of magnitudes in a rough correspondence with the discharge area; these climatological totals are used for normalizing climatological and anomaly records of riverflows



Standard deviations of binned normalized anomalies of 112 riverflow records: 1894-1985



Standard deviations of binned normalized anomalies of 84 riverflow records subjected to the RSOI analysis: 1894-1985

The 28 stations that could not be analyzed by the standard RSOI procedure included all 16 Afghan stations, nine Iranian stations, and one station from Pakistan, Tajikistan, and Turkmenistan each. An additional interpolation procedure, based on the nearest neighbor (NN) approach, was developed to match each station left out by the RSOI method with one of its 10 NN whose RSOI estimates positively correlated with the available target station observations at 95% significance level. These matched stations were used as linear predictors for target stations. This approach resulted in the extension of the objective analysis to 111 out of the original 112 stations. Error estimates were produced for both techniques: RSOI and the NN-based regression. To facilitate the use of the streamflow data set in large-scale climate analyses, in addition to the station-based representation of the data, binned representations of them were produced for a 2 degree spatial grid. Results of the analysis were checked for variance consistency of the observational records with the resolved and error variance. Analyzed records were compared with historical precipitation and temperature records from the RSOI analysis of binned Global Historical Climatology Network (GHCN) stations and Climate Research Unit (CRU) analyses.

NN match of 28 stations (colored circles) omitted from the RSOI analysis to the stations analyzed by the RSOI; color of circles indicates correlation coefficients of target and matched stations


Standard deviations of binned normalized anomalies of 111 riverflow records from the RSOI analysis with NN regression augmentation: 1894-1985



Standard deviations for months with available observations of binned normalized anomalies of 111 riverflow records from the RSOI analysis with NN regression augmentation

Climatological analysis of records from 112 river flow stations in CSW Asia have shown a general progression of climatological cycle from the west to the east of the area, with westernmost stations reaching their maximum flow in April-May and easternmost stations achieving it in July-August. The data set was found to exhibit strong geographical dependence in the type of climate variability and the phase and shape of seasonal cycle. The East-West progression of seasonal cycle in streamflows was characterized and connected to the similar progression in precipitation.

Meridional mean of normalized climatology for riverflows in CSW Asia

Comparison of analyzed gridded riverflow anomalies for April-September period with a gridded precipitation analysis of GHCN stations have shown their significant positive correlations with lags from 2 to 7 months (precipitation precedes riverflows) for the entire 1894-1985 period virtually everywhere in the CSW Asia domain, with an exception of a few southeast gridpoints. Maximum correlation were achieved at 5 months lag. A similar comparison of riverflows with air temperatures has found significant anticorrelations in the western part of the domain for lags 0 to 4 months (air temperature precedes riverflows), with strongest anticorrelations achieved at 2-3 month lags. Comparisons using CRU precipitation and air temperature analyses for the period 1902-1985 gave similar results.

Acknowledgement: Supported by the NSF grant ATM02-33651 to A.Kaplan and M.K.Tippett (LDEO of Columbia University); in collaboration with M.Barlow (University of Massachusetts at Lowell), H.M.Cullen (Georgia Institute of Technology), D.Salstein (Atmospheric and Environmental Research, Inc.)