@article { author = {Mostafazadeh, Raoof and Zabihi, Mohsen}, title = {Comparison of SPI and SPEI indices to meteorological drought assessment using R programming (Case study: Kurdistan Province)}, journal = {Journal of the Earth and Space Physics}, volume = {42}, number = {3}, pages = {633-643}, year = {2016}, publisher = {Institute of Geophysics, University of Tehran}, issn = {2538-371X}, eissn = {2538-3906}, doi = {10.22059/jesphys.2016.57881}, abstract = {Drought indices are commonly used to quantify and assess drought characteristics. The Standardized Precipitation Index (SPI), and recently introduced Standardized Precipitation-Evapotranspiration Index (SPEI) are considered as universal meteorological drought indices which allow comparisons of drought conditions across different climate regions. The SPI captures anomalies in precipitation, whereas the SPEI estimates anomalies in climatic water balance that incorporates temperature. The main aim of this study is to compare historical drought occurrence based on SPI and SPEI indices using R programming. The SPEI index is used because of multi-scalar nature of index and the advantage of identifying the multi-temporal nature of droughts. According to data availability, seven synoptic stations were selected for a drought analysis across Kurdistan Province. The two-parameter gamma distribution was used for calculating SPI across the study period (1995-2013) and stations within the study area. The potential evapotranspiration (PET) was computed using the Thornthwaite's equation, and then the SPEI is calculated at a monthly temporal resolution using SPEI package in R software. The SPI and SPEI values are calculated and then the statistical analysis along with significant level scatter plot was performed. The graphical plot of 3-month SPI and SPEI values were prepared to visualize the capabilities of used indices in determination of wet and dry spells over studied stations. The relationships between computed SPI and SPEI values were analyzed using correlation coefficient and p-value at each station. The results indicated that some differences in the pattern and sequence of wet and dry spells exist based on calculated indices. Also, the SPEI index identified the longer wet and dry spell conditions than SPI in almost all cases. The results of the comparative analysis indicated that the SPI and SPEI were varied between 0.19 (p<0.01), and 0.52 (P<0.01), which were statistically significant in all stations. A very low correlation between the SPI and SPEI was identified in Saghez station (correlation coefficient= 0.19), which seems to be due to evapotranspiration and moisture loss during spring/summer with the increasing temperatures that is accounted for by SPEI. The highest correlation coefficient was calculated between SPI and SPEI in the Sanadaj station (0.52%). Since, the SPEI accounts temperature in defining drought spells, therefore, it is advisable to use SPEI instead of SPI for drought assessment. According to the graphical interpretation of the results, there was large difference between the droughts depicted by the precipitation-based SPI and the temperature influenced SPEI. Also the SPEI captured the influence of temperature and depicted severe and longer duration droughts which provide support for better performance and reliability of the SPEI index. It should be noted that in terms of lack of data, evapotranspiration can be calculated by simple methods such as Thornthwaite, but considering detailed available data, the Hargreaves and Penman methods can be used to determine drought occurrences in the SPEI calculation. The calculation of drought indices should be simple and statistically reliable, in this regard, SPEI indicators in different climatic conditions and climate change issues has a significant advantage. Also, more climate variables are needed to calculate the SPEI index than the SPI index. Also, the calculated evapotranspiration value is sensitive to the used method and requires a longer data period with natural variabilities. Further research is recommended in other climatic regions which is needed for comparison of SPEI with other common drought indices to draw comprehensive conclusions.}, keywords = {Meteorological drought,Potential evapotranspiration,R software,Standardized Precipitation Index,Wet-Dry Spell Analysis}, title_fa = {تحلیل و مقایسه شاخص‌های ‏SPI‏ و ‏SPEI‏ در ارزیابی خشک سالی هواشناسی با استفاده از نرم‌افزار R (بررسی موردی: استان کردستان)}, abstract_fa = {شاخص SPEI خشکسالی بیلان آبی را با درنظر گرفتن دما و تبخیر و تعرق برآورد می‌نماید. هدف پژوهش حاضر، مقایسه الگوی وقوع خشکسالی براساس شاخص‌های SPI و SPEI به‌دلیل ماهیت چند مقیاسی بودن و قابلیت تحلیل خشکسالی در مقیاس‌های زمانی می‌باشد. در این راستا هفت ایستگاه سینوپتیک استان کردستان انتخاب و شاخص SPI بر اساس توزیع گامای دومتغیره محاسبه گردید. تبخیر و تعرق پتانسیل جهت استفاده در شاخص SPEI توسط معادله تورنوایت محاسبه و سپس مقادیر شاخص‌های SPI و SPEI در مقیاس زمانی 3 ماهه با استفاده از برنامه نویسی R به‌دست آمد. مقادیر SPI و SPEI در قالب گراف‌های توالی دوره‌ها رسم و ارتباط آن‌ها با تحلیل همبستگی مورد آزمون قرار گرفت. نتایج مقادیر شاخص‌های خشکسالی با ضریب همبستگی و سطح معنی‌داری به‌صورت ماتریس پراکندگی ابرنقاط ارائه شد. نتایج نشان داد که ضمن وجود تفاوت در الگوی وقوع دوره‌های ترسالی و خشکسالی SPI و SPEI، شاخص SPEI دوره‌های طولانی‌تر خشکی را در اکثر ایستگاه‌ها مشخص نموده است. بر اساس نتایج، ضرایب همبستگی بین 0/19 در ایستگاه سقز و 0/52 در ایستگاه سنندج و معنی‌دار از نظر آماری (در سطح 99 درصد)، متغیر هستند. به‌نظر می‌رسد همبستگی ضعیف (0/19) میان شاخص‌های SPI و SPEI در ایستگاه سقز ناشی از افزایش تبخیر و تعرق در فصول بهار و تابستان محاسبه شده توسط شاخص SPEI است. با توجه به تاثیر دما در محاسبه SPEI، تفاوت قابل توجهی بین مقادیر شاخص استاندارد مبتنی بر بارش و شاخص تبخیر و تعرق مبتنی بر دما، زمینه کارایی و صحت شاخص مذکور را فراهم می‌نماید.}, keywords_fa = {Meteorological drought,Potential evapotranspiration,R software,Standardized Precipitation Index,Wet-Dry Spell Analysis}, url = {https://jesphys.ut.ac.ir/article_57881.html}, eprint = {https://jesphys.ut.ac.ir/article_57881_2878d7e873427f99f68d02763f2f2d2a.pdf} }