Southern Portugal is located in a semi-arid region sensitive to climate change, eventually prone todesertification. Changes in precipitation regimes, which are associated with corresponding changes in largescaleatmospheric circulation over the Euro-Atlantic region, have been observed in recent decades (Zhanget al., 1997; Corte-Real et al., 1998; Trigo and DaCamara, 2000). Precipitation regimes play an essentialrole in water resources management, which in turn controls agriculture, as well as other economic activities.Therefore, it is very important to assess impacts of climate change on water resources in the region.Evora is in the centre of the region, and thus a single- ´ site stochastic precipitation model was developedtowards obtaining daily precipitation scenarios for impact studies, by downscaling from large-scaleatmospheric circulation simulated with OAGCMs. Four daily circulation patterns were identified from meansea-level pressure (MSLP) fields over the Euro-Atlantic region, by the K-means clustering algorithm coupledwith principal component analysis (PCA), mainly for differentiating between precipitation states at Evora ´(Corte-Real et al., 1998). It has been verified that changes in the frequencies of occurrence of certaincirculation patterns may correspond to the observed changes in precipitation at Evora, as well as in other ´parts of the country. A stochastic weather generator, conditioned on daily circulation patterns, can reproducewell the major statistical characteristics of daily precipitation in the present climate, by coupling it with GCMoutput from the control run of the Hadley Centre’s second generation model (HadCM2) (Johns et al., 1997).The weather generator comprises a first-order Markov chain, conditioned on daily circulation patterns ofthe current day as well as of the previous day, and serially independent precipitation amounts on wet daysobeying a two-parameter gamma distribution. Details of this study can be found in Corte-Real et al. (1999a).The Guadiana River (Rio Guadiana) is an international river near the border between southern Portugaland Spain, where a big dam, Alqueva, has been built on the Portuguese side. Water resources management insuch an international river basin is definitely a very important issue to both countries, especially when climatechange is taken into account. Water management in the Guadiana Basin has been designed as a case study inthe European project SWURVE (Sustainable Water — Uncertainty, Risk and Vulnerability in Europe) to bedealt with by hydrological models. Therefore, it is necessary to have appropriate climate scenarios as an inputfor the hydrological models. Given the limitation of available climate data in the river basin, only six stationson the Portuguese side along the Guadiana River have been included in this study, although more stationscan be easily incorporated. Daily precipitation, and maximum and minimum temperature data for the stationsin the Guadiana Basin were kindly provided by the Portuguese National Meteorological Institute (Instituto deMeteorologia, IM). Figure 1 shows the stations along the river and the spatial distributions of probabilitiesof precipitation in January, conditioned on the four daily circulation patterns identified by Corte-Real et al.(1998). It is clear that higher probabilities of precipitation are present in the rainy pattern (CP4), but muchlower probabilities are observed under the winter dry pattern (CP3); in general, though, the precipitationprobabilities tend to decrease from north to south. Details about the four circulation patterns and conditionalprecipitation statistics at Evora can be found in Corte-Real ´ et al. (1998, 1999a,b).We assume that the spatial correlation in the precipitation field is related to circulation patterns, e.g. therainy pattern can induce large-scale precipitation and the dry patterns can result in wide-spread dry weatherin the river basin. Therefore, daily precipitation series at different stations, generated by a single-site weathergenerator, may hold some interstation correlations when daily circulation patterns are incorporated, even ifthe generating processes are independent at different stations. However, it is doubtful whether the interstationcorrelations can be reasonably captured, especially for daily precipitation amounts. Figure 2 exhibits coherenceof occurrence of wet and dry days in synthetic series and observations for all station pairs (15) in wintermonths (October–March) (a total of 90 points). Synthetic series were obtained from single-site precipitationgenerators, either conditioned or unconditioned on daily circulation patterns. Not very surprisingly, syntheticseries capture a reasonable fraction of coherency of occurrence of dry days, even when daily circulationpatterns are not considered, reflecting that dry days are common in the region even in the rainy season.However, the spatial coherency of rain occurrence is reproduced less well in the synthetic series. Anyway,the weather generators conditioned on daily circulation patterns can do a relatively better job for both wetand dry occurrences. More importantly, interstation correlation coefficients of daily precipitation amounts are
đang được dịch, vui lòng đợi..
