Abstract
Drought is one of the most serious problems for human societies and ecosystems. Climate change has begun to affect the whole world, but regional climate change can have different effects. Precipitation is frequently used to determine drought; besides, temperature has started to be used in recent years. The effect of temperature on the occurrence of drought was investigated for the Marmara region, located in the northeast of Turkey. The spatiotemporal variability of drought in the Marmara region was analyzed by using the standardized precipitation index (SPI) and standardized precipitation evapotranspiration index (SPEI) for 3-, 6-, and 12-month series. Results show that SPEI droughts occur with longer duration and increased magnitude, which can be attributed to the rise in potential evapotranspiration rates with an increase in temperature. The correlation between SPI and SPEI is strong for the same time series, and the number of extreme drought values for SPI is higher than SPEI, while the number of severe and moderate drought values is lower compared to SPEI. The years 1989, 1990, 2001, 2007, and 2014 were characterized as the highest drought years throughout the region. The study concludes that temperature should be considered in future studies.
HIGHLIGHTS
This study evaluates drought in terms of short- and long-time periods in the Marmara region.
The drought has been analyzed temporally and spatially.
The study gives an idea about the importance of considering the temperature in the determination of drought.
The drought duration and intensity were analyzed according to the short- and long-time periods.
Standardized precipitation and evapotranspiration index droughts were found to last longer and be more severe than standardized precipitation index.
INTRODUCTION
Drought is a native disaster that causes severe hydrological instability that adversely affects the agricultural production as a result of precipitation decreasing significantly below normal (World Meteorological Organization-WMO 1997). Drought can be seen in dry climatic regions as well as in humid climatic regions. It is the type of disaster that manifests itself for many years and is the most difficult to encounter when compared to the results of other atmospheric events. Drought, which can cause many natural events such as erosion and landslide, can be meteorological, agricultural, and hydrological (Mishra & Singh 2010). If there is a significant decrease in precipitation below normal values over a long period, meteorological drought exists. The factor that determines the meteorological drought is the length and degree of moisture deficiency (Kapluhan 2013). Agricultural drought, another type of drought, is the absence of water required for the plant in the soil (Dai 2011). It occurs when moisture loss and scarcity of water resources occur. Hydrological drought is a long-term precipitation deficiency that results in reduced soil moisture, lower water source levels, and surface flow in the hydrological system (Yüceerim et al. 2019; Avsaroglu & Gumus 2022). If the meteorological drought lasts a long time, it can be described as hydrological drought.
Drought differs from other disasters as it is a natural phenomenon whose beginning and end are difficult to determine. It gradually increases in strength and can maintain its effect even after the event is over (Dinç et al. 2016). The supply–demand imbalance caused by drought and the resulting forced migration, unemployment, poverty, and famine lead to a new drought type and social-economic drought. In addition, decreasing water quality, soil erosion, and deterioration of vegetation structure are among the environmental impacts of drought (Erdem et al. 2021).
Drought has been an important research area for scientists from the past to the present to avoid its devastating effects or to take precautions. Various definitions of drought, including meteorological, hydrological, and agricultural, have been developed in the literature, and their results cause socioeconomic drought. To understand the effects of drought and to reduce the effects that may occur, the drought process needs to be defined and monitored in detail. Many drought indices have been developed to classify drought and determine the relationship between drought and climate parameters (Zeybekoğlu & Aktürk 2021). For example, the most commonly used methods for meteorological drought can be listed as standardized precipitation index (SPI) (McKee et al. 1993) and standardized precipitation and evapotranspiration index (SPEI) (Vicente-Serrano et al. 2010). Besides, the streamflow drought index (Nalbantis 2008) method, which is an index similar to SPI, has been proposed to characterize the severity of hydrological droughts. The role of potential evapotranspiration (PET) in drought indices is considered important for the accurate definition of drought (Teuling et al. 2013). For this reason, SPEI and Reconnaissance Drought Index (RDI) have been used successfully in many agricultural drought studies in recent years (Al-Faraj et al. 2016; Chen et al. 2016; Simsek et al. 2023). Similarly, the agricultural standardized precipitation index, a modified version of SPI, is one of the agricultural drought determination methods based on substituting total precipitation with effective precipitation (Tigkas et al. 2018). As can be seen, although there are many methods, SPI, which uses only precipitation data, and SPEI, which uses PET and precipitation together, stand out among the most widely used methods in the literature (Tirivarombo et al. 2018; Eris et al. 2020; Pei et al. 2020; Shamshirband et al. 2020; Danandeh Mehr et al. 2022; Moazzam et al. 2022). Although these two methods use different input parameters, their calculations are similar (Danandeh Mehr et al. 2022).
The SPI method, considered a conventional drought index method, has been used in the determination of drought for many years (Band et al. 2022; Bonaccorso et al. 2003; Livada & Assimakopoulos 2007; Rahman & Lateh 2016; Dabanlı et al. 2017; Gumus & Algin 2017). On the other hand, the SPEI method proposed in 2010 is frequently used to monitor drought in regions with different climatic characteristics worldwide. For example, Shiru et al. (2018) used the SPEI to define climate change's impacts on Nigeria's drought, Mathbout et al. (2018) examined the spatiotemporal features of the drought occurrence in Syria according to the SPI and the SPEI indices, and Qutbudin et al. (2019) evaluated the severity of drought in Afghanistan by using the SPEI. Feng et al. (2020) analyzed the meteorological drought using the SPEI in the Qinghai–Tibet Plateau.
While the SPI contains only precipitation for calculating the drought index, the SPEI considers both precipitation and PET. For this reason, many researchers have compared these two indices in determining drought. Danandeh Mehr et al. (2020) analyzed the meteorological drought in the capital city of Turkey by using the SPI and the SPEI. The SPI and SPEI methodologies were compared by Homdee et al. (2016), and they suggested that the SPEI method was more accurate. Kamaruzaman et al. (2022) investigated drought characteristics in Bangladesh and changes in time and spatial scale in different geomorphologists using SPI and SPEI drought indices. It has been stated that SPEI is more successful in detecting drought/wet cycles in regions with complex geomorphology. Isia et al. (2022) used monthly precipitation and temperature data to examine the drought periods in Sarawak with SPI and SPEI indices. They suggested that temperature is a determining factor in drought classification and that SPI should be used only when temperature data are unavailable. Although these studies state that SPEI is more successful in drought monitoring, Eris et al. (2020) stated that SPI and SPEI have the same drought understanding capacity due to their drought analysis using SPI and SPEI in the Küçük Menderes River basin. They also reported that SPI and SPEI methods could be used in drought-related research and applications in the Küçük Menderes River Basin. Besides, it is mentioned by Harisuseno (2020) that the SPI model gives more suitable results in determining drought characteristics. Therefore, there is still a need to investigate the drought monitoring capacities of these methods in regions with different climatic and geographical characteristics.
Turkey has a semi-arid climate, and drought is a type of disaster that influences people's socioeconomic well-being and harms agriculture, constituting the country's important livelihood. The SPI method has been widely used for drought monitoring in Turkey, as well as in other parts of the world (Dogan et al. 2012; Gumus & Algin 2017; Gumus et al. 2021; Danandeh Mehr et al. 2022). The most comprehensive studies among these were conducted by Umran Komuscu (1999); Sönmez et al. (2005); Türkeş & Tatlı (2009); and Dabanlı et al. (2017). However, drought studies conducted with SPEI are very limited, and only certain regions have been evaluated (Dikici 2020; Eris et al. 2020; Danandeh Mehr et al. 2022; Yeşilköy & Şaylan 2022; Zeybekoglu 2022). Therefore, increasing drought monitoring studies with the SPEI method in different regions of Turkey, which has a very high surface area of 783 562 km2, will help to understand a complex event such as drought.
The Marmara region of Turkey boasts a unique bridge-like quality, connecting Europe and Asia. With its fertile lands and ideal climate, the region holds a significant place in the country's agricultural production. The people living in this region mostly make their living from agriculture-based activities, so drought will affect the livelihoods of the people in the region. However, studies associated with drought are insufficient in this region which is very important in terms of agriculture and industry for Turkey (Yeşilköy & Şaylan 2022). Furthermore, most previous studies have primarily utilized the SPI, without considering the influence of temperature. To address this research gap, the present study offers a comprehensive assessment of drought conditions in the Marmara region, applying both the SPI and the SPEI. Not often adopted in existing research, this approach considers temperature data, potentially providing a more thorough understanding of the region's drought conditions. In addition, this study seeks to present a detailed temporal and spatial evaluation of drought patterns by utilizing recent data and incorporating data from more meteorological stations.
Drought indices have been calculated at 3-, 6-, and 12-month timescales between 1970 and 2021 using both SPI and SPEI methods. This methodology aims better to understand historical drought patterns in the Marmara region. Through a comprehensive and detailed analysis of drought conditions in a critical agricultural region of Turkey, the current research endeavours to make a meaningful contribution to the existing body of knowledge.
MATERIALS AND METHODS
Study area
In this study, the temporal and spatial drought analyses of the Marmara region are discussed as a study area. Twenty-four meteorological gauging stations (SGS) situated in the Marmara region are used to determine drought features by using SPI and SPEI methods (Figure 1). The stations are classified according to their coordinates, and the region is divided into three different zones: northern west (NW), northern east (NE), and south (S). Spatial distribution maps are presented using inverse distance weighting, a robust deterministic technique for spatial interpolation of results (Shepard 1968; Lu & Wong 2008). Commercially available software (ArcGIS 10.1) was used to obtain spatial distribution maps for the studied area (Gumus & Algın 2017).
Climate data
The region's temperature and precipitation values of 24 meteorological measurement stations were used to determine the historical droughts in the Marmara region. The monthly meteorological values between 1970 and 2021 were gained from the Turkish State Meteorological Service (MGM, a Turkish acronym), in which precipitation (P) and maximum (Tmax) and minimum (Tmin) air temperatures (Table 1) were used to calculate the SPI and SPEI values. These data are quality controlled by MGM before they are published. In addition, very limited (less than 1%) missing data were observed in the meteorological data. These missing data were completed using the linear regression method from neighbouring stations and used in the drought analysis.
Station No . | Station name . | Sub-region . | Long-term seasonal total precipitation . | Long-term annual mean . | |||||
---|---|---|---|---|---|---|---|---|---|
Summer . | Autumn . | Winter . | Spring . | Precipitation (mm) . | Tmax (°C) . | Tmin (°C) . | |||
17050 | Edirne | NW | 101.76 | 161.53 | 182.22 | 156.33 | 50.15 | 27.19 | 2.13 |
17052 | Kırıkkale | 102.72 | 158.51 | 175.53 | 141.77 | 48.21 | 25.83 | 2.57 | |
17608 | Uzunköprü | 89.72 | 186.26 | 224.94 | 159.66 | 55.41 | 26.68 | 1.63 | |
17056 | Tekirdağ | 77.85 | 170.84 | 191.21 | 132.3 | 47.81 | 24.75 | 4.87 | |
17054 | Çorlu | 77.85 | 170.84 | 191.21 | 132.3 | 48.31 | 25.3 | 2.52 | |
17061 | Sarıyer | 122.11 | 267.88 | 310.87 | 148.4 | 71.09 | 25.49 | 6.11 | |
17631 | Lüleburgaz | 87.22 | 159.25 | 185.06 | 175.5 | 50.72 | 27.51 | 0.37 | |
17636 | Florya | 74.73 | 183.17 | 242.94 | 135.37 | 53.3 | 25 | 5.98 | |
17610 | Şile | NE | 132.64 | 276.9 | 288.72 | 146.28 | 70.65 | 25.98 | 4.34 |
17119 | Yalova | 97.08 | 220.75 | 264.3 | 154.77 | 61.74 | 26.8 | 5.13 | |
17066 | Kocaeli | 181.01 | 214.48 | 275.25 | 183.11 | 68.59 | 27.89 | 5.95 | |
17069 | Sakarya | 181.01 | 214.48 | 275.25 | 183.11 | 71.42 | 28.69 | 4.49 | |
17120 | Bilecik | 75.87 | 107.33 | 146.44 | 135.59 | 38.9 | 26.3 | 2.24 | |
17062 | Kadıköy | 82.01 | 192.12 | 247.3 | 134.23 | 54.98 | 25.81 | 6.38 | |
17662 | Geyve | 97.62 | 148.16 | 230.91 | 153.49 | 52.72 | 28.32 | 2.55 | |
17702 | Bozüyük | 72.5 | 106.83 | 154.18 | 147.16 | 40.17 | 26 | −2.82 | |
17116 | Bursa | S | 70.73 | 190.57 | 256.51 | 177.7 | 58.36 | 28.43 | 3.01 |
17674 | Balıkesir | 54.95 | 191.64 | 267.87 | 159.91 | 56.51 | 28.04 | 2.26 | |
17700 | Dursunbey | 59.16 | 131.95 | 214.7 | 154.21 | 46.83 | 26.24 | 1.05 | |
17112 | Çanakkale | 42.9 | 163.15 | 254.31 | 143.55 | 50.54 | 25.54 | 5.3 | |
17145 | Edremit | 32.49 | 191.15 | 306.89 | 161.66 | 58.03 | 28.24 | 5.8 | |
17114 | Bandırma | 54.04 | 210.61 | 288.32 | 156.07 | 59.43 | 26.93 | 4.06 | |
17175 | Ayvalık | 17.61 | 180.33 | 315.58 | 136.97 | 54.52 | 27.13 | 7.17 | |
17695 | Keles | 74.84 | 162.45 | 278.21 | 216.28 | 61.31 | 23.53 | −1.89 |
Station No . | Station name . | Sub-region . | Long-term seasonal total precipitation . | Long-term annual mean . | |||||
---|---|---|---|---|---|---|---|---|---|
Summer . | Autumn . | Winter . | Spring . | Precipitation (mm) . | Tmax (°C) . | Tmin (°C) . | |||
17050 | Edirne | NW | 101.76 | 161.53 | 182.22 | 156.33 | 50.15 | 27.19 | 2.13 |
17052 | Kırıkkale | 102.72 | 158.51 | 175.53 | 141.77 | 48.21 | 25.83 | 2.57 | |
17608 | Uzunköprü | 89.72 | 186.26 | 224.94 | 159.66 | 55.41 | 26.68 | 1.63 | |
17056 | Tekirdağ | 77.85 | 170.84 | 191.21 | 132.3 | 47.81 | 24.75 | 4.87 | |
17054 | Çorlu | 77.85 | 170.84 | 191.21 | 132.3 | 48.31 | 25.3 | 2.52 | |
17061 | Sarıyer | 122.11 | 267.88 | 310.87 | 148.4 | 71.09 | 25.49 | 6.11 | |
17631 | Lüleburgaz | 87.22 | 159.25 | 185.06 | 175.5 | 50.72 | 27.51 | 0.37 | |
17636 | Florya | 74.73 | 183.17 | 242.94 | 135.37 | 53.3 | 25 | 5.98 | |
17610 | Şile | NE | 132.64 | 276.9 | 288.72 | 146.28 | 70.65 | 25.98 | 4.34 |
17119 | Yalova | 97.08 | 220.75 | 264.3 | 154.77 | 61.74 | 26.8 | 5.13 | |
17066 | Kocaeli | 181.01 | 214.48 | 275.25 | 183.11 | 68.59 | 27.89 | 5.95 | |
17069 | Sakarya | 181.01 | 214.48 | 275.25 | 183.11 | 71.42 | 28.69 | 4.49 | |
17120 | Bilecik | 75.87 | 107.33 | 146.44 | 135.59 | 38.9 | 26.3 | 2.24 | |
17062 | Kadıköy | 82.01 | 192.12 | 247.3 | 134.23 | 54.98 | 25.81 | 6.38 | |
17662 | Geyve | 97.62 | 148.16 | 230.91 | 153.49 | 52.72 | 28.32 | 2.55 | |
17702 | Bozüyük | 72.5 | 106.83 | 154.18 | 147.16 | 40.17 | 26 | −2.82 | |
17116 | Bursa | S | 70.73 | 190.57 | 256.51 | 177.7 | 58.36 | 28.43 | 3.01 |
17674 | Balıkesir | 54.95 | 191.64 | 267.87 | 159.91 | 56.51 | 28.04 | 2.26 | |
17700 | Dursunbey | 59.16 | 131.95 | 214.7 | 154.21 | 46.83 | 26.24 | 1.05 | |
17112 | Çanakkale | 42.9 | 163.15 | 254.31 | 143.55 | 50.54 | 25.54 | 5.3 | |
17145 | Edremit | 32.49 | 191.15 | 306.89 | 161.66 | 58.03 | 28.24 | 5.8 | |
17114 | Bandırma | 54.04 | 210.61 | 288.32 | 156.07 | 59.43 | 26.93 | 4.06 | |
17175 | Ayvalık | 17.61 | 180.33 | 315.58 | 136.97 | 54.52 | 27.13 | 7.17 | |
17695 | Keles | 74.84 | 162.45 | 278.21 | 216.28 | 61.31 | 23.53 | −1.89 |
The region's long-term average maximum and minimum temperature is approximately 26.5 and 3.4 °C, respectively, and the average monthly total precipitation is 55.4 mm. In the study region, the highest precipitation was recorded in Sakarya (71.42 mm) and the lowest in Bilecik (38.9 mm). Therefore, although the long-term average monthly total precipitation of the whole of Turkey is 48 mm, it can be said that it receives a higher precipitation value than the country in general. In addition, Tmax takes values between 23 and 28 °C, and Tmin is between −3 and 8 °C.
The standard precipitation index
xij is precipitation (in mm) in month j of year, and xim and σi are mean and standard deviation, respectively. The intensity values and abbreviations given in Table 2 are used in the classification of drought.
Class . | SPI/SPEI . |
---|---|
Extremely wet | ≥2 |
Severity wet | 1.5 to 1.99 |
Moderately wet | 1.0 to 1.49 |
Near normal | −0.99 to 0.99 |
Moderately drought | −1 to −1.49 |
Severity drought | −1.5 to −1.99 |
Extremely drought | ≤− 2 |
Class . | SPI/SPEI . |
---|---|
Extremely wet | ≥2 |
Severity wet | 1.5 to 1.99 |
Moderately wet | 1.0 to 1.49 |
Near normal | −0.99 to 0.99 |
Moderately drought | −1 to −1.49 |
Severity drought | −1.5 to −1.99 |
Extremely drought | ≤− 2 |
The standard precipitation evapotranspiration index
RESULTS
Temporal evaluation of drought
Figure 4 illustrates that the maximum drought occurred in April and June of 2007 for both indices in the NW region. The highest values for SPI-6 and SPEI-6 in the S region were −2.70 in June 1989 and −1.99 in September 2007, respectively. In addition, the maximum mean value for SPI-6 in the NE region was obtained in 2020, while that for SPEI-6 was observed in 1989. The maximum average index for SPI-12 and SPEI-12 was observed in 2001 for NW, 2007 and 2014 for NE, and September 2007 for S. In summary, the drought years for the Marmara region in the 1970–2021 time series were determined as 1989, 1990, 2001, 2007, and 2014.
Drought durations and severities in NW, NE, and S of the Marmara region using SPI and SPEI at different time series for 0, −0.5, and −1 thresholds are analyzed and given in Table 3. In addition, the temporal distribution of drought was analyzed using the maximum index value and the change in drought duration. For this purpose, the consecutive occurrence of drought for different threshold (SPI/SPEI < 0, SPI/SPEI < −0.5, and SPI/SPEI < −1.0) values was also analyzed. The following conclusions can be drawn from the table. (1) As the timescale increases, the drought takes longer, and the magnitude increases for both SPI and SPEI in all regions. (2) Compared to the SPI, SPEI generally calculates higher values in terms of duration and severity, thereby indicating longer-term and more severe drought conditions. Conversely, under the threshold categories, SPI tends to identify less mean duration and severity than SPEI. (3) The ratio between the maximum and mean drought durations was calculated to be 3.8 on average for all time periods and all threshold values. (4) The ratio between the maximum and mean drought severities decreases as the threshold value decreases. The ratio between the maximum and mean drought severities was calculated to be 4.6 on average in all time periods and all threshold values.
. | . | Duration (month) . | Severity . | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
SPI3 . | SPEI3 . | SPI6 . | SPEI6 . | SPI12 . | SPEI12 . | SPI3 . | SPEI3 . | SPI6 . | SPEI6 . | SPI12 . | SPEI12 . | ||
NW | Mean (<0) | 3.76 | 4.00 | 5.41 | 5.47 | 9.73 | 8.14 | 2.33 | 2.63 | 3.40 | 3.61 | 5.72 | 5.41 |
Max. (<0) | 14.00 | 27.00 | 24.00 | 28.00 | 39.00 | 33.00 | 10.38 | 21.40 | 20.29 | 27.44 | 31.55 | 34.77 | |
Mean (<− 0.5) | 2.43 | 2.60 | 3.70 | 3.93 | 5.34 | 6.50 | 2.42 | 2.58 | 3.64 | 3.92 | 5.16 | 6.84 | |
Max. (<− 0.5) | 7.00 | 12.00 | 14.00 | 18.00 | 19.00 | 27.00 | 7.80 | 12.77 | 14.17 | 23.22 | 22.65 | 33.34 | |
Mean (<− 1) | 1.64 | 1.83 | 2.34 | 3.48 | 3.88 | 5.47 | 2.26 | 2.38 | 3.07 | 4.66 | 4.96 | 7.36 | |
Max. (<− 1) | 4.00 | 4.00 | 6.00 | 14.00 | 12.00 | 19.00 | 5.72 | 5.64 | 9.14 | 19.47 | 16.32 | 26.95 | |
NE | Mean (<0) | 3.68 | 3.75 | 4.64 | 5.74 | 6.98 | 7.92 | 2.59 | 2.57 | 3.10 | 4.06 | 4.60 | 5.50 |
Max. (<0) | 14.00 | 20.00 | 22.00 | 34.00 | 35.00 | 34.00 | 14.12 | 17.08 | 18.79 | 31.96 | 32.16 | 39.10 | |
Mean (<− 0.5) | 2.23 | 2.65 | 3.81 | 4.38 | 6.16 | 5.89 | 2.41 | 2.71 | 3.97 | 4.52 | 6.39 | 6.26 | |
Max. (<− 0.5) | 6.00 | 10.00 | 16.00 | 17.00 | 22.00 | 28.00 | 10.87 | 11.54 | 16.60 | 20.77 | 25.99 | 37.04 | |
Mean (<− 1) | 1.89 | 1.97 | 2.61 | 3.12 | 4.93 | 6.08 | 2.94 | 2.74 | 3.65 | 4.21 | 6.81 | 8.50 | |
Max. (<− 1) | 4.00 | 5.00 | 7.00 | 9.00 | 15.00 | 21.00 | 9.89 | 8.75 | 11.28 | 12.60 | 20.62 | 31.31 | |
S | Mean (<0) | 3.35 | 3.90 | 4.81 | 6.10 | 9.67 | 9.47 | 2.22 | 2.66 | 3.09 | 4.24 | 6.07 | 6.50 |
Max. (<0) | 21.00 | 19.00 | 22.00 | 32.00 | 37.00 | 39.00 | 10.32 | 13.49 | 15.50 | 28.32 | 30.45 | 35.03 | |
Mean (<− 0.5) | 2.26 | 2.65 | 3.14 | 4.07 | 4.42 | 7.12 | 2.35 | 2.62 | 3.16 | 4.07 | 4.17 | 7.06 | |
Max. (<− 0.5) | 7.00 | 10.00 | 11.00 | 14.00 | 16.00 | 27.00 | 8.93 | 13.12 | 13.93 | 15.56 | 17.70 | 35.03 | |
Mean (<− 1) | 1.72 | 1.88 | 2.58 | 2.74 | 3.65 | 4.32 | 2.50 | 2.52 | 3.54 | 3.53 | 4.69 | 5.61 | |
Max. (<− 1) | 4.00 | 4.00 | 7.00 | 10.00 | 11.00 | 20.00 | 7.99 | 6.76 | 10.70 | 14.16 | 16.74 | 28.49 |
. | . | Duration (month) . | Severity . | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
SPI3 . | SPEI3 . | SPI6 . | SPEI6 . | SPI12 . | SPEI12 . | SPI3 . | SPEI3 . | SPI6 . | SPEI6 . | SPI12 . | SPEI12 . | ||
NW | Mean (<0) | 3.76 | 4.00 | 5.41 | 5.47 | 9.73 | 8.14 | 2.33 | 2.63 | 3.40 | 3.61 | 5.72 | 5.41 |
Max. (<0) | 14.00 | 27.00 | 24.00 | 28.00 | 39.00 | 33.00 | 10.38 | 21.40 | 20.29 | 27.44 | 31.55 | 34.77 | |
Mean (<− 0.5) | 2.43 | 2.60 | 3.70 | 3.93 | 5.34 | 6.50 | 2.42 | 2.58 | 3.64 | 3.92 | 5.16 | 6.84 | |
Max. (<− 0.5) | 7.00 | 12.00 | 14.00 | 18.00 | 19.00 | 27.00 | 7.80 | 12.77 | 14.17 | 23.22 | 22.65 | 33.34 | |
Mean (<− 1) | 1.64 | 1.83 | 2.34 | 3.48 | 3.88 | 5.47 | 2.26 | 2.38 | 3.07 | 4.66 | 4.96 | 7.36 | |
Max. (<− 1) | 4.00 | 4.00 | 6.00 | 14.00 | 12.00 | 19.00 | 5.72 | 5.64 | 9.14 | 19.47 | 16.32 | 26.95 | |
NE | Mean (<0) | 3.68 | 3.75 | 4.64 | 5.74 | 6.98 | 7.92 | 2.59 | 2.57 | 3.10 | 4.06 | 4.60 | 5.50 |
Max. (<0) | 14.00 | 20.00 | 22.00 | 34.00 | 35.00 | 34.00 | 14.12 | 17.08 | 18.79 | 31.96 | 32.16 | 39.10 | |
Mean (<− 0.5) | 2.23 | 2.65 | 3.81 | 4.38 | 6.16 | 5.89 | 2.41 | 2.71 | 3.97 | 4.52 | 6.39 | 6.26 | |
Max. (<− 0.5) | 6.00 | 10.00 | 16.00 | 17.00 | 22.00 | 28.00 | 10.87 | 11.54 | 16.60 | 20.77 | 25.99 | 37.04 | |
Mean (<− 1) | 1.89 | 1.97 | 2.61 | 3.12 | 4.93 | 6.08 | 2.94 | 2.74 | 3.65 | 4.21 | 6.81 | 8.50 | |
Max. (<− 1) | 4.00 | 5.00 | 7.00 | 9.00 | 15.00 | 21.00 | 9.89 | 8.75 | 11.28 | 12.60 | 20.62 | 31.31 | |
S | Mean (<0) | 3.35 | 3.90 | 4.81 | 6.10 | 9.67 | 9.47 | 2.22 | 2.66 | 3.09 | 4.24 | 6.07 | 6.50 |
Max. (<0) | 21.00 | 19.00 | 22.00 | 32.00 | 37.00 | 39.00 | 10.32 | 13.49 | 15.50 | 28.32 | 30.45 | 35.03 | |
Mean (<− 0.5) | 2.26 | 2.65 | 3.14 | 4.07 | 4.42 | 7.12 | 2.35 | 2.62 | 3.16 | 4.07 | 4.17 | 7.06 | |
Max. (<− 0.5) | 7.00 | 10.00 | 11.00 | 14.00 | 16.00 | 27.00 | 8.93 | 13.12 | 13.93 | 15.56 | 17.70 | 35.03 | |
Mean (<− 1) | 1.72 | 1.88 | 2.58 | 2.74 | 3.65 | 4.32 | 2.50 | 2.52 | 3.54 | 3.53 | 4.69 | 5.61 | |
Max. (<− 1) | 4.00 | 4.00 | 7.00 | 10.00 | 11.00 | 20.00 | 7.99 | 6.76 | 10.70 | 14.16 | 16.74 | 28.49 |
Spatial evaluation of drought
DISCUSSION
There are numerous studies on drought monitoring using the SPI and SPEI indices in various parts of the world. For instance, Tirivarombo et al. (2018) compared the SPI and the SPEI in the Kafue Basin located in northern Zambia for the time series spanning from 1960 to 2015. Their findings revealed that the SPEI identified longer and more severe moderate and severe drought occurrences, while the SPI defined extreme droughts more frequently but with shorter durations. Similarly, the present study found that the SPEI determined longer and more severe droughts for all timescales, except for extreme droughts where the situation is reversed. Tirivarombo et al. (2018) also observed that the correlation between SPI and SPEI was lower for shorter timescales, while the present study represents that high correlation was obtained in the low time period. The mean correlation between SPI and SPEI indices at all stations was 0.906 for the 3-month time period and 0.908 for the 12-month time period. This result shows that the results obtained regionally differ, and some results cannot be evaluated in general.
Mathbout et al. (2018) reported that drought severity and duration increased with the length of the timescale and that the SPEI identified longer and more severe droughts compared to the SPI. Similarly, in this study, it can be said that the length and severity of the drought increased with the increase in the timescale, and SPEI estimated longer and more severe drought against the SPI. Eris et al. (2020) performed temporal and spatial analyses of meteorological drought and dry periods in western Turkey's Kucuk Menderes River Basin from 1960 to 2018. They utilized the SPI and SPEI indices to estimate meteorological drought at short- and long-term timescales. The results showed that short-term timescales were reliable for both indices, whereas it was better to use soil moisture drought or hydrological drought for long-term timescales. Pei et al. (2020) analyzed the SPI and SPEI drought indices at various timescales (1, 3, 6, and 12 months) in Inner Mongolia from 1981 to 2018. Their findings revealed that temporal variations in SPI and SPEI became increasingly consistent with the increasing timescale and that drought characteristics of the two indices differed in the time series. However, with the increase in time series, the spatial distributions of drought trends became more consistent for both methods. On the contrary, in this study, although there is an average of 90% agreement between SPI and SPEI indices throughout the Marmara region in this study, it was observed that the correlation between SPI and SPEI decreased as the timescale increased consistent with previous studies (Vicente-Serrano et al. 2010; Jiang et al. 2015).
Li et al. (2020) reported that SPEI detected more severe drought events and fewer mild drought events than SPI. Furthermore, within the study area, approximately 30% of the drought grids detected by SPI was not identified as drought by SPEI and 40% of the drought grids detected by SPEI were not identified as drought by SPI. As shown in Figures 7–9, the number of severe dry drought events obtained by the SPEI method is higher than the SPI method in this study. Moreover, this difference becomes even greater as the timescale increases. Therefore, it can be said that the results obtained in this study are consistent with the study by Li et al. (2020) in China.
In the analysis of the Asi Basin in southern Turkey, Dikici (2020) identified 1973–1974, 1989–1991, 1993–1994, 2000–2001, 2004–2005, 2014, and 2016 as the most drought years according to the SPEI method. In addition, Yeşilköy & Şaylan (2022) found that the occurrence of moderate and higher droughts in the northwestern part of Turkey between 1971 and 2018 was 17.2 based on the SPEI index. Zeybekoglu (2022) used the SPEI index to examine the meteorological drought in the Hirfanli Dam basin in Turkey and determined that 2001 was the most severe drought year. In the present study, 2001 was also identified as one of the drought years for the Marmara region.
CONCLUSIONS
The insights obtained from the examination of the spatiotemporal variability of drought in the Marmara region of Turkey, leveraging the comparative analysis between SPI and SPEI, have unveiled noteworthy patterns and correlations. Distinctly, the SPEI droughts were found to persist for longer durations and amplify in magnitude compared to SPI droughts, hinting at the consequential role of PET rates in influencing SPEI analyses. The correlation between SPI and SPEI for 3-, 6-, and 12-month periods manifested a robust relationship for identical time series, with an observed decrease in the CC with the increasing time series.
The study highlighted the years 1989, 1990, 2001, 2007, and 2014 as the ones witnessing the most drought conditions across the Marmara region from 1970 to 2021. The implications of these findings carry significant weight as they underline the importance of incorporating temperature and PET values in future drought projections, given their apparent impact on drought index values.
In the broader context, this investigation offers a valuable addition to understanding drought patterns and their potential impacts on a region that holds a significant position in Turkey's agriculture. Hopefully, these findings will serve as a stepping stone for future research aiming to deepen the understanding of drought phenomena and, more importantly, for establishing effective drought management strategies in regions of similar climatic and agricultural characteristics.
The main limitation of this study is that only the Hargaveres method, which is a temperature-based approach, is used to calculate the PET. However, it is known that potential evaporation is affected by temperature and different parameters such as wind and humidity. However, these parameters could not be considered since these data could not be obtained for the Marmara region. Therefore, it is recommended to investigate how PET calculation methods affect the drought structure for regions with different climate patterns in future studies.
ACKNOWLEDGEMENTS
The author is grateful to the Turkish State Meteorological Service for providing the meteorological data.
AUTHOR CONTRIBUTION
All sections in the manuscript have been prepared by Nazire Göksu Soydan Oksal.
DATA AVAILABILITY
Data are available from the Turkish State Meteorological Service.
CODE AVAILABILITY
Not applicable.
ETHICS APPROVAL
Not applicable.
CONSENT FOR PUBLICATION
The author of the article agrees to submit the article and is aware of the submission.
COMPETING INTERESTS
The author declare no competing interests.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.