ارزيابي و مقايسه داده‌هاي بازكاوي شده بارش جهت استفاده در ايران

Evaluation and comparison of reanalysis precipitation data in Iran

در پژوهش پيش رو، داده‌هاي آفروديت، GPCC و داده‌هاي بارش دانشگاه دي‌لور (UDel) براساس داده‌هاي بارش ايستگاهي ارزيابي شده اند. در اين راستا، از تكنيك‌هاي RMSE، ضريب همبستگي و دياگرام تيلور استفاده شده است. نتايج حاصل از ارزيابي داده‌ها نشان داد كه جهت ارزيابي داده‌ها، دياگرام تيلور بخاطر ارايه‌ي تصويري جامع‌تر از رابطه‌ي هندسي بين RMSD، ضريب همبستگي و انحراف معيار سري‌هاي زماني، نسبت به ساير روش‌هاي تك متغيره نظير RMSE و ضريب تعيين، مناسب‌تر است. ترسيم ميانگين بلند مدت بارش سالانه‌ي ايران براساس داده‌هاي مزبور، دقت بيشتر داده‌هاي آفروديت و GPCC را نسبت به داده‌هاي UDel نشان مي‌دهد. نتايج ارزيابي داده‌ها مشخص ساخت، داده‌‌هاي آفروديت براي مناطق شمال ، شمال غرب ، دامنه‌هاي جنوبي البرز و نواحي داخلي كشور مناسب‌تر و داده‌هاي GPCC در مناطق غرب، جنوب، جنوب شرق و شمال شرق كشور به نتايج بهتري منتهي مي‌شوند. همچنين مشخص شد كه، داده‌هاي UDel به دليل در نظر گرفتن ارتباط فضايي داده‌ها با متغير وابسته، مقادير بارشِ سري‌هاي زماني ناقص را بهتر از دو داده ديگر برآورد مي‌كند.

Extended Abstract Introduction There are significant differences in the spatial distribution of the Iran’s annual precipitation, resulting from spatial behavior of precipitation on the one hand and variation in the sources of precipitation at the other. The lack of adequate distribution of meteorological stations and the unavailability of long-term statistics of precipitation makes analysis of precipitation more complicated. Precipitation data are constant inputs of research and models related to water resources (eg, climate, agriculture, hydrology, environment). Most of research institutions are used to record the data and present it to different users. Different ways of interpolation of the data will cause different results; so it is a critical step to select the appropriate data based on research design. This study evaluates APHRODIATE, GPCC and Delaware University precipitation data (UDel) based on precipitation data stations using RMSE, R2 and Taylor diagram techniques. Materials and methods DATA APHRODITE APHRODITE’s (Asian Precipitation - Highly-Resolved Observational Data Integration Towards Evaluation) daily gridded precipitation is the only long-term (1951 onward) continental-scale daily product that contains a dense network of daily precipitation-gauge data for Asia including the Himalayas, South and Southeast Asia and mountainous areas in the Middle East. The number of valid stations was between 5000 and 12,000, representing 2.3 to 4.5 times the data available through the Global Telecommunication System network, which were used for most daily grid precipitation products. The products are available on a regional basis. Key Strengths: High density and quality station network. Key Limitations: Station network changes with time and season. GPCC The Global Precipitation Climatology Centre (GPCC) has been established in 1989 on request of the World Meteorological Organization (WMO). It is operated by Deutscher Wetterdienst (DWD, National Meteorological Service of Germany) as a German contribution to the World Climate Research Programme (WCRP). The GPCC provides gridded gauge-analysis products derived from quality controlled station data. Two products are for climate: (a) the Full Data Reanalysis Product (1901-2010) is recommended for global and regional water balance studies, calibration/validation of remote sensing based precipitation estimations and verification of numerical models, and (b) the VASClimO 50-Year Data Set which is for climate variability and trend studies. Key Strengths: Large number of stations used; gauge network extends beyond GHCN Key Limitations: Variable number of stations per grid over time can be a major inhomogeneity source. Monitoring products are frequently updated but climate products are not. Global (land) precipitation, University of Delaware (UDel) A series of gridded temperature and precipitation data sets. Station records that served as bases for the Terrestrial Air Temperature: 1900-2010 Gridded Monthly Time Series (Version 3.01) and Terrestrial Precipitation: 1900-2010 Gridded Monthly Time Series (Version 3.02) archives are used here to help create new gridded climatologies of monthly and annual average air temperature (T) and total precipitation (P). These two sets of station time series were drawn primarily from recent versions of the Global Historical Climatology Network (GHCN version 2) and the Global Surface Summary of Day (GSOD) archive. Selected averages from Legates and Willmott’s (1990a and b) long-term station averages of monthly and annual T and P also were used to help produce this new gridded archive. Key Strengths: Provides a relatively detailed global land surface temperature climatology. Higher spatial resolution than comparable data sets Key Limitations: Infrequent updates. Methods In order to evaluate the data, the closest point from the mentioned precipitation data to meteorological stations (max 40 km) were identified for the period 1961-2007. Then we used RMSE, the coefficient of determination (R2) and Taylor diagram to evaluate precipitation data. These methods are formulated as below: Where and are the precipitation value provided by instrumental data and precipitation data respectively, and are the variance values of instrumental data and precipitation data respectively, n indicate the number of stations. Results and discussion Fig.1, shows the Taylor diagram, plotted by considering spatially averaged precipitation values, The diagram summarizes the relationship between testing and reference series standard deviations, correlation coefficient, and the RMSD (root mean square difference) computed considering series centered pattern, by means of a trigonometric similitude. Fig.1: Taylor diagram obtained from spatial averaged values plotted on the basis of standard deviation values, correlation coefficients between products and reference dataset, and root mean square differences of series-centered pattern, indicated as RMSD in the plot. Aphrodite data are more accurate at, Khoi, Babolsar, Tehran and Yazd stations. GPCC data has better performance than other data at Zahedan and Bandar Abbas stations. For Shahr e Kord, Mashhad and Zanjan stations, Aphrodite and GPCC data have similar RMSD but according to the stronger correlations between GPCC and instrumental data, the GPCC data is more appropriate than Aphrodite data. Conclusion Based on the long-term average annual precipitation, Aphrodite and GPCC data are more accurate than UDel data. Taylor diagram is based on the geometrical relationship between correlation coefficient, series standard deviation and centered mean square error. It is useful than other uni-variable methods as RMSE and R2. Aphrodite data is better to use for the North West, the southern Alborz and internal areas and GPCC data will lead to better results in the West, South, South East and North East of Iran. As UDel data considers spatial association of data with dependent variable, it estimates precipitation time series better than other data. This type of data is useful to analyse characteristics of precipitation in the areas with short-term time series.