Observation Analysis And Projection Of Seasonal Precipitation Over Tanzania

Understanding patterns of changes in precipitation over Tanzania is essential for adapting and mitigating the socio-economic impacts associated with extreme events.This thesis investigates seasonal precipitation over Tanzania from observation and projection.The methods utilized in this thesis include Correlation analysis,Regression analysis,Composite analysis,Standardized anomaly,the Mann-Kendall trend test(MK trend test),Empirical orthogonal function(EOF),Mean Bias,Root mean square error(RMSE),Standard deviation(STD),Theil-Sen slope estimator and the Taylor diagram.The main findings are as follows:(1)It is revealed that the March to May(MAM)rainfall over Tanzania experienced a downward trend in the late 1990 s.The result shows that the recent decrease in MAM rainfall is connected to the reduction of water vapour transport from the Indian Ocean to Tanzania,enhanced tropospheric moisture divergence,and suppressed convection over Tanzania.Changes in sea surface temperature(SST)in the late 1990 s from a cold phase to a warm phase over the southwestern Indian Ocean(SWIO)and western tropical Pacific Ocean(WPO)and from a warm phase to a cold phase over the southern Atlantic Ocean(SAO)also have a significant contribution.Linked to the SST warming over the SWIO,an anomalous cyclonic circulation prevails over the tropical Indian Ocean in the upper level,favourable for increasing subsidence in Tanzania.Concurrently,the Mascarene High shifts eastward,which may weaken the water vapour transport from the southern Indian Ocean toward Tanzania.The warming of SST over the WPO is related to a Walker-like circulation with the descending limb over the western Indian Ocean(WIO)and Tanzania,conducive to the subsidence over Tanzania.Associated with the SST cooling over the SAO,anomalous atmospheric convergence in the upper level is dominant over Tanzania,concurrent with the easterly outflow toward the Congo basin and the westerly and northerly outflow toward the WIO from Tanzania in the low level,favouring the decrease of MAM rainfall.(2)It is found that the sea ice concentration(SIC)over the Weddell(Ross)sea is negatively correlated with the MAM rainfall,particularly over northern(southern)Tanzania.This signifies that the high(low)SIC over the Weddell and Ross sea is associated with suppressed(enhanced)rainfall in Tanzania.The atmospheric circulations underlying the linkage between SIC anomalies and MAM rainfall variability are the upper-level wave train patterns propagating from the high latitudes to low latitudes,which results in upper-level cyclonic circulation anomaly over the WIO and Tanzania favouring subsidence of airflows over Tanzania.In addition,the outflow of winds over Tanzania due to lower-level divergence,water vapour flux divergence,and suppressed convection(descending motion)are also observed over Tanzania.Hence,the high(low)SIC anomalies in the Weddell and the Ross seas are related to dry(wet)conditions during the MAM season over Tanzania.(3)It is revealed that the SST anomalies to the east of Australia and the southern Atlantic Ocean(SAO)are negatively and significantly correlated with October to December(OND)rainfall throughout Tanzania,signifying that warm(cool)SST anomalies to the east of Australia and SAO are associated with a suppressed(enhanced)OND rainfall throughout Tanzania.The impact of SST anomalies in the Southern Oceans(SO)on the OND variability in Tanzania is revealed through the modulation of moisture and dynamic conditions.The warmer than normal SST anomalies to the east of Australia and SAO are linked to the enhanced divergence in the lower troposphere and the enhanced convergence in the upper troposphere over Tanzania and the WIO,signifying the sinking of the Indian Ocean Walker-type circulation and Atlantic Ocean Walkertype circulation over the study domain,which suppresses convection and hence the decreased rainfall over the target region.Also,the reduced OND rainfall in Tanzania is supported by the westerly moisture flux over the Indian Ocean moving away from Tanzania,moisture divergence and descending motion over the target region associated with SO SST.(4)Generally,models display good performance in simulating the annual cycle(i.e.,bimodal pattern)of precipitation;however,most models and the multimodel ensemble mean(ENSEM)underestimate(overestimating)MAM(OND)rainfall.Spatially,most models seem to perform relatively better in simulating OND rainfall compared to MAM rainfall,despite wet(dry)biases over most parts of EA.The metrics have shown that 9 models(i.e.,MPI-ESM1-2-HR,AWICM-1-MR,MPI-ESM1-2-LR,IPSL-CM6A-LR,CAS-ESM2-0,BCC-CSM2-MR,Nor ESM2-LM,CESM2-WACCM,and CMCC-CM2-SR5)outperform the other individual models during both MAM and OND rainfall seasons.Compared with individual models,the ENSEM better agrees with the observation.In addition,the ENSEM and most models can reproduce the increasing trend in OND rainfall as observed over EA.However,given that half of the models show increasing trends for the secular change of MAM rainfall,the ENSEM simulation produces an inconsistent trend with the observation.The ensemble of the best models with better performance projects an increase of rainfall averaged over EA in both MAM and OND rainfall seasons by the end of the 21 st century,with larger changes under SSP5-8.5 than under SSP2-4.5.Spatially,the largest percentage increase in MAM rainfall is projected to occur in central Tanzania and northwestern Kenya,while southeastern Tanzania shows the greatest percentage decrease.The projected percentage increases in OND rainfall are larger in the northeastern highlands of Tanzania and northwest Kenya.