Wintertime precipitation climate of Central Southwest Asia: Diagnostics and forecasting
Central Southwest Asia (CSWA; 20°–47°N, 40°–85°E) is a water-scarce and a societally vulnerable region, prone to significant variations in precipitation during the winter months of November–April. Wintertime precipitation variations have a direct impact on CSWA's water resources, agricultural productivity, energy use, and human society. Because of the close relationship between climate and human well-being, an improved understanding of winter season precipitation and its variability over CSWA is of critical importance. However, due to multiple regional challenges (e.g. socio-political instability, extreme topographical heterogeneity, poor coverage of in situ stations, and others) analysis of precipitation in this region has been limited. In an attempt to bridge the existing knowledge gap, this thesis aims to advance our understanding of CSWA's wintertime precipitation climate through three separate, but inter-related studies on 1) evaluation of multi-source gridded precipitation dataset, 2) investigation of spatial and temporal patterns of precipitation and its links with large-scale modes of climate variability, 3) development of a statistical forecast model. Additionally, precipitation evaluation is also relevant to the overlapping and important region of the Indian subcontinent; a detailed seasonal analysis for which is also presented. First, the performance of several commonly used gridded precipitation products from multiple sources: gauge-based, satellite-derived, and reanalysis is analysed for all four seasons over the Indian Subcontinent. Results show that the degree of uncertainty in all precipitation estimates varies by region (e.g. topographic relief) and the type of precipitation (e.g. convective, orographic). At the seasonal scale, satellite-products perform better, while reanalyses generally overestimate precipitation. Greater discrepancies occur in areas with low gauge densities, owing to which a complete understanding of the accuracy and limitations of precipitation estimates is hampered for the northwestern region of the Indian subcontinent. In an extension study, ten multi-source precipitation products are evaluated against an ensemble of four gauge-only datasets. This analysis is carried out for CSWA, which also includes the northwestern region of the Indian subcontinent. Spatial and temporal analysis of results shows that GPCC is a suitable observational dataset for studying long-term wintertime precipitation variations over CSWA. The satellite-derived TRMM 3B42-V7 is a potentially reliable alternative to gauge measurements, while the performance of MERRA reanalysis is satisfactory. Further, the spatial-temporal patterns of wintertime precipitation variability over CSWA are explored. Three leading patterns are identified by empirical orthogonal function (EOF) analysis, and the associated time series are related to global SST and other large-scale atmospheric circulation fields. The leading patterns of winter precipitation are significantly linked with the El Niño–Southern Oscillation (ENSO); East Atlantic–Western Russia (EA-WR); Siberian High; North Pacific Oscillation (NPO); Scandinavian pattern; and the long-term warming of the Indian Ocean SST. The inter-decadal change of relationship between the first-mode of winter precipitation and ENSO is also investigated, which shows that CSWA precipitation variability was closely related to the extratropical EA-WR (tropical ENSO) teleconnection before (after) the 1980's. Finally, the level and origin of seasonal forecast skill of wintertime precipitation anomalies over CSWA are examined using the statistical method of canonical correlation analysis (CCA). The preceding months’ (September–October) SST is used as predictors, and CCA experiments are performed for two sets of time periods, 1950/51–2014/15 and 1980/81–2014/15. For both prediction periods, the potential source of predictability originates largely from SST variations related to ENSO and the Pacific Decadal Oscillation (PDO). A higher (lower) correlation skill of 0.71 (0.38) is obtained between observations and cross-validated precipitation forecasts for the period 1980/81–2014/15 (1950/51–2014/15); which shows that ENSO played a dominant role in creating skillful predictions for CSWA wintertime precipitation in recent years.