بكارگيري مدلSDSM جهت ريزمقياس نمائي داده هاي GCM بارش و دما مطالعه موردي: پيش‌بيني‌هاي اقليمي ايستگاهي در ايران

Using SDSM Model to Downscaling Precipitation and Temperature GCM Data for Study Station Climate Predictions over Iran

بكارگيري روش‌هاي جديد در حل معادلات جوي و در اختيار داشتن پيش بيني هاي اقليمي با توجه به ماهيت بازه زماني طولاني مدت آنها، نقش بسيار ارزنده‌اي در مديريت‌هاي كلان ايفا مي‌نمايد. وليكن در بازه زماني دراز مدت بدليل محدوديت جدي در قدرت تفكيك مكاني، قادر به پيش‌بيني آب و هواي واقعي در مقياس ايستگاهي و خرد مقياس نمي‌باشد. لذا جهت بكارگيري خروجي مدل‌هاي اقليمي تمام كره‌اي و دستيابي به قدرت تفكيك فضائي كم، روش ريز مقياس نمائي مورد استفاده قرار گرفته كه به دو دسته آماري و ديناميكي و گاها تلفيقي از آن دو تقسيم‌بندي مي‌گردند. در اين ميان صحت سنجي داده‌هاي ريزمقياس شده جهت تحليل پيش‌بيني‌هاي درازمدت بعنوان يكي از پارامترهاي اساسي در كسب دقت اين گونه مدل‌ها از اهميت ويژه‌اي برخوردار است. در اين تحقيق سعي گرديده تا با انتخاب  كل منطقه كشور و ابتدا براي 41 ايستگاه منتخب كشور كه داراي آمار اقليمي 41 ساله (2001-1961 ميلادي) مي‌باشند، خروجي مدل اقليميHadCM3 تحت سناريوي اقليمي A2 كه يكي از محتمل‌ترين سناريوهاي انتشار مي‌باشد، توسط مدلSDSM كه قادر است خروجي مدل‌هاي گردش عمومي جو را به مقياس ايستگاهي تبديل نمايد، ريزمقياس گردد.  سپس با استفاده از روش‌هاي آماري و بدست آوردن ضرائب وزني داده‌‌هاي ريزمقياس شده و داده‌هاي پايه را مورد تجزيه و تحليل قرار داده و واسنجي مناسبي از آنها ارائه گردد. نتايج بيانگر آنست كه بين مقادير ريزمقياس شده بارش، دماي حداقل و حداكثر و مقادير واقعي آنها تفاوت معنا داري با خطاي بحراني 0/05 وجود ندارد و بازه اطمينان داده‌ها مشتمل برمقدار صفر است. لذا بكارگيري داده‌هاي ريزمقياس‌شده مدل جهت بهينه‌سازي داده‌هاي آينده در مقياس ايستگاهي مي‌تواند بصورت قابل قبول مورد استفاده قرار گيرد.

Introduction: Iran is located in the southwest of Asia and is in the arid belt of the world and about 60% of the extent of the country is mountainous and the remaining part (1/3) is desert and arid lands. The climate of the country can be divided into three main categories: Warm temperate, rainy with dry summer in a narrow strip in the north, Dry, hot desert in the central plateau, Dry, hot steppe covering the rest of the country. So, it could be so difficult to predict climate change over whole of country. In this case, using new methods for solving weather equations and having climate prediction because of its long term temporal has so many important rolls for massive management. In climate change studies, the global circulation models (GCMs) are usually used to simulate the past and future global climate. Unfortunately, despite the advancement in GCM research and modern computing technology, the most recent generation of general circulation models still have serious problems due to their low spatial resolutions (with the field variables being represented on grid points 300 km apart). So, because of its serious limitation and resolution, using them in long term forecast couldn’t predict actual weather in station scale or small scale and it is important to assess the accuracy and uncertainty of GCMs in various climatic and geographical regions. Methodology: To employ output of Global Climate Models and accesses to good resolution, "Downscaling Methods" are used that are divided in two dynamical and statistical groups and some when syncretistic of them. A thorough evaluation of the current generation of GCMs has only started recently and the evaluation of a rich spectrum of indices on extremes is new. In this, calibrating downscaled data is very important as a main parameter to reach best resolution and for analyzing long term forecast. Two different approaches to downscaling have been employed. It has adopted a methodology that exploits mean inter station correlations to correct the statistics of gridbox means. The method, closely related to blockkriging, is demonstrated to remove the sle size sensitivity of statistics in daily gridpoint precipitation. It has adopted a direct downscaling by distance and direction weighted average of point observations. At this filed, SDSM is a Statistical down scaling model that distributed both of these aggregation techniques to the consortium. Several data of selected stations have been started applying to a dataset and coding study area. These datasets will provide a valuable reference for model evaluation simulate predictor variables across selected region. The SDSM model run on selected period and reached amount of precipitation, minimum and Maximum of temperature and their standard deviations. By using statistical methods we could evaluate SDSM outputs to reach the best conclusion and selecting best data. With acceptable results, we could use them for climate prediction over region. Materials: In this paper, at the first we tried to select 41 synoptic stations that have 41 years climate data (19612001). These stations distributed to whole country with several climates. These data applied our observation dada. At this method we used third version of the coupled global Hadley Centre Climate Model (HadCM3) Outputs as predictor of method and A2 scenario that is one of the most probable emission scenarios. Then we down scaled them by using SDSM model version4.2 that could downscale general circulation models to station scales. Then by using statistical methods and reaching differential coefficients could analyses downscaled data by base data and present suitable correlation of them. Results and Discussion: Results was shown, there is no significant deference with 0.5 critical errors and correlation of data and accepted at 0.01 significant levels. And there is a good accepted correlation between modeled data and observing minimum and maximum temperature and precipitation data. Conclusion: So, using Downscaled data is acceptable with suitable efficiency to correct future data at station scale. This study should help to fill in the knowledge gap in GCM downscaling research of climate and add an important piece in the global climatic assessment jigsaw puzzle.