بررسي تحليلي توزيع زماني بارش توفان‌هاي جنوب غرب ايران (مطالعه موردي: توفان 12 تا 15 نوامبر 2006 – 21 تا 24 آبان 1385)

Analysis of the temporal distribution of rainfall storms in South West ofIran (Case study: storm 12-15 November 2006)

شناخت دقيق توزيع زماني بارش و نحوه وقوع توفان‌ها و تاثير آن بر چگونگي وقوع سيلاب از مهم‌ترين مقوله‌هاي مطالعاتي هواشناسي- آب‌شناسي محسوب مي‌‌شود. بررسي‌هاي انجام شده براي برآورد توزيع زماني بارش به ويژه توسط سازمان حفاطت خاك آمريكا (SCS) سبب ارايه فرم‌هاي مختلفي شده است. به دليل عدم وجود اطلاعات كافي مشاهداتي در ساير نقاط دنيا از جمله ايران، در بررسي‌هاي هيدرولوژي از توابع توزيعي استفاده مي‌شود. باتوجه به وجود بزرگ‌ترين سدهاي مخزني در حال بهره‌برداري در جنوب ‌غرب ايران (زاگرس جنوبي)، بررسي توزيع زماني بارش براساس روابط تحليلي براي افزايش دقت در بررسي‌هاي هيدرولوژي، از اصلي‌ترين اهداف اين پژوهش است. در اين پژوهش، تابع توزيع زماني لجستيك بر روي داده‌هاي بارش برازش داده شد و ضريب‌هاي لجستيك منحني توزيع زماني بارش با استفاده از نتايج مدل عددي وضع هوا (WRF)، درتوفان بزرگ 12 تا 15 نوامبر 2006 محاسبه شد. نتايج نشان ‌دادكه ضريب a ازحداقل 2/0 در بخش‌هاي جنوب ‌شرقي و مركزي تا 7/0 در شمال و جنوب ‌غرب متغيراست، به‌عبارتي در شمال و جنوب ‌غرب منطقه شدت بارش به مراتب بيشتر است. بررسي مقادير محاسبه شده براي ضريب B نيز نشان داد كه اين ضريب نيز از حداقل 4 در جنوب‌ شرق و مركز منطقه تا حدود 14 در جنوب و شمال متغير است، افزايش اين مقدار نشان از تاخير بيشتر در زمان شروع بارش درتوفان است.

Understanding the reasons behind the occurrence of significant storms and their rainfall distribution is a fundamental issue to increase the accuracy of the Meteorological and Hydrological predictions. The rainfall pattern is helpful in managing flood in urban basins and designing control structures for flood transmission. The time distribution patterns of heavy rainfall events have not been investigated much in Iran. This paper investigates the temporal rainfall distribution based on an analytical numerical method in southwest of Iran. This study selected the storm that happened on 12-15 November 2006 with precipitation of over 50 mm. First, rainfall data were calculated using the result of WRF numerical weather prediction model. Then, a temporal distribution of a logistic function has been fitted on rainfall data and the logistic coefficients of the rainfall curve were computed for the mentioned great storm event. Rainfall information in the entire basin is not available due to irregular distribution and limited number of measuring stations. Thus, to study in detail and quantify the amount of rainfall in the entire basin with a regular grid, the best method is a simulation of Numerical Weather Prediction (NWP) models. Therefore, rainfall was simulated with the Weather Research Forecasting (WRF) model in two regular nested domains with the horizontal resolution of 27 and 9 km, from 47.5E to 53E degrees longitude and from 30.5N to 34N degrees latitude. The input Final operational Global Analysis (FNL) data was provided by National Center Environmental Prediction (NCEP). After running the model, three-hour interval rainfall outputs at all grid points were collected for the storm. The time distribution of rainfall in each grid point was computed based on the rainfall outputs of WRF and a three-parameter logistic function. To select the best values of the coefficients a and b, we used minimum Sum of Square Error (SSE) in the logistic curve fitting for every grid point. Since the greatest change in the slope of the logistic function is in the 70% middle of the temporal distribution curve of rainfall, the best values of the coefficients a and b are determined without 15% beginning and 15% end of the logistic curve. Since the range of the coefficient b is very large, the transfer function (B) reduced its variance. Changing the coefficients a and B leads to appearing logistic curve in various forms, which are the main parameters of the rainfall distribution. If the coefficient a increase as well as the logistic curve slope, more precipitation occurs at a fixed time in the dimensionless temporal rainfall distribution curve. Moreover, the time delay of the onset of the precipitation increases by changing the coefficient B. Finally, maps of the coefficients a and B were drawn by using the IDW method in SURFER software. The results show that the value of the coefficient a of the logistic function varies from 0.2 to 0.4 for southeastern and central parts and to 0.7 for northern and southwestern parts. Thus, in the north and south-west of the studied area, rainfall intensity becomes greater than in southeastern and central parts. Furthermore, the increase in coefficient is seen in the latitude from the bottom to up and changes of coefficient are not related to the longitude from west to east. However, it is related to changes in the topography of the area. The values of the coefficient B of the logistic function vary from 4 in southeast and central parts to 14 in south and north parts of the region. Therefore, any increase in coefficient leads to increase the time delay of the onset of the precipitation. To reveal more differences in the temporal distribution of rainfall, the studied area was zoned out to nine regions (A1, A2, A3, B1, B2, B3, C1, C2, C3) and the temporal distribution of rainfall was plotted for each region. Results show the maximum of rainfall intensity occurs from 0.42 to 0.58 of the total rainfalls time period in types A1, A2, A3, B3 and C2. Maximum rainfall intensity of types B1, B2 and C3 falls from 0.25 to 0.42, 0.25 to 0.67 and 0.42 to 0.83 of the total rainfall time periods, conversely. Additionally, the most of the total rainfall falls before 1/3 of the total rainfall time period in type C1. Thus, the maximum rainfall intensity occurs in early start time of rainfall.