بررسی دقت مرکز اروپایی پیش‌بینی‌های میان مدت جوی (ecmwf) در پیش بینی بارش مناطق گوناگون اقلیمی ایران

به منظور بررسی میزان خطای مرکز اروپایی پیش‌بینی‌های میان مدت جوی (ecmwf) در پیش‌بینی بارش مناطق مختلف ایران، داده‌های بارش ماهانه 45 ایستگاه‌ همدیدی پراکنده در سطح کشور با داده‌های بارش ماهانه erainterim در نزدیک‌ترین نقطه به هر ایستگاه مورد مقایسه آماری قرار گرفت. برای این منظور از آماره‌هایی مانند ضریب تعین (r2)، مجذور میانگین مربع خطا (rmse)، شیب خط (b slope)، اریبی (bias) و ضریب کارایی مدل (ef) برای مقایسه آماری داده‌های erainterim در برابر داده‌های مشاهده‌ای ایستگاه‌های مورد مطالعه استفاده شده است. نتیجه این بررسی نشان داد که erainterim از دقت بسیار بالایی در پیش‌بینی بارش بسیاری از نقاط کشور برخوردار است و میزان خطای آن در بیش از 70 درصد از ایستگاه‌های مورد مطالعه اندک و قابل چشم پوشی است. همچنین مشخص شد که erainterim بارش ایستگاه‌های ناحیه ساحلی خزر و برخی ایستگاه‌های ناحیه ساحلی خلیج فارس را کم‌تر از مقدار واقعی پیش‌بینی می‌کند. کم برآورد کردن بارش در ایستگاه‌های ناحیه خزری عمدتا به علت ناتوانی erainterim در پیش‌بینی درست برخی بارش‌های فرین این ناحیه می‌باشد. اگرچه بین بارش erainterim و بارش مشاهده‌ای در منطقه شمال غرب و شمال شرق کشور همبستگی بسیار قوی دیده می‌شود ولی مقدار بارش پیش‌بینی شده erainterim برای اکثر ایستگاه‌های این مناطق نیز بیش‌تر از مقدار واقعی برآورد می‌شود.

Investigation of the accuracy of the European Center for Medium Range Weather Forecast (ECMWF) in forecasting observed precipitation in different climates of Iran

The lack of reliable and updated precipitation datasets is the most important limiting factor in studying many climatological and hydrological topics including climate change and temporal variability of precipitation in many data sparse areas around the globe. This is particularly valid for Iran that encompasses vast deserts and unsettled hyperarid climate areas (centraleastern Iran) that hinders establishing an adequate network of raingauge stations required for climatological studies. Similarly, the high elevation areas of mountainous regions of western and northern Iran suffer from limited representative stations. Using the gridded or reanalysis precipitation datasets could be one of the possible solutions to overcome this obstacle; knowing that the representativeness of these datasets has been already proved for many different parts of the world. Amongst many available gridded precipitation datasets are the Global Precipitation Climatology Center (GPCC) and the Tropical Rainfall Measuring Mission (TRMM) that have been widely used in many researches; indicating their accurate estimation of precipitation values and intraannual variation for the regions studied. The reanalysis precipitation dataset which is a product of the Numerical Weather Prediction (NWP) models is an alternative source of precipitation data that is widely used in the literature and many authors have pointed to the relatively accurate precipitation prediction of reanalysis for many parts of the word. The two widely used reanalysis datasets are NCEP/NCAR and different products of ECMWF, namely ERA15, ERA40 and ERAInterim reanalysis. ERAInterim which is used in the present study is produced at T255 spectral resolution (about 80 km) and covers the period from January 1979 to present, with product updates at approximately 1 month delay from realtime (Dee et al., 2011; Balsamo et al., 2015). The ERAInterim atmospheric reanalysis is built upon a consistent assimilation of an extensive set of observations (typically tens of millions daily) distributed worldwide (from satellite remote sensing, in situ, radio sounding, profilers, etc.). To develop the reanalysis, the analysis step combines the observations with a prior estimate of the atmospheric state (firstguess fields) produced with a global forecast model in a statistically optimal manner (Balsamo et al., 2015).The representativeness and performance of ERAInterim in forecasting precipitation amount at 45 Iranian synoptic stations distributed across the country is herein examined. Spatial resolution of ERAInterim dataset used in this study is 0.125 × 0.125 in latitude and longitude. For each station, the closest grid point of ERAInterim to the station coordinates was chosen for a statistical comparison analysis. To evaluate the performance of the considered dataset when compared to the observed precipitation records at the considered locations we have used R squared, the Nash–Sutcliffe model efficiency coefficient (EF), RMSE, Bias, B slope of the regression and the standardized RMSE indicators. The performance of the dataset was also graphically represented through scatter plots of the established regression between ERAInterim and observation at the selected stations. The results of the statistical indicators were represented through plotting the indicators over the map of Iran to ease displaying spatial tendency of the indicators and explaining the possible geographical role in controlling the spatial variation of the indicators. The results indicate that the ERAInterim performs well in majority of the studied stations with strong correlation coefficient. However, it was found that the ERAInterim underestimates precipitation in most of the stations located in the coastal areas of the Caspian Sea as well as in some stations along the Persian Gulf and the Oman Sea, suggesting that ERAInterim is somewhat inefficient in adequately forecasting precipitation in the coastal areas; very likely due to not properly taking into account the complex topography of the region in its model parameterization or not being able to adequately differentiate between land and sea characteristics for the stations very close to the sea. It should be noted that the ERAInterim is less efficient in accurately forecasting extreme precipitation in the Caspian Sea region. Nevertheless, we found very high agreement between observations and ERAInterim in this region when some extreme precipitation events were excluded from the analysis. Contrarily, the results suggest an overestimation for most of the stations located in northwestern and northeastern mountainous areas of the country; once again due to perhaps improper representation of topography of these regions in the model.