Precipitation is the integral part and main driving agent of global water budget. It is therefore important to comprehensively study at various spatial scales ranging from a watershed (Vaheddoost and Aksoy, 2017) to a region (Ferrari et al., 2013) or to a country (Dahamsheh and Aksoy, 2007), and at temporal scales ranging from minutes to days, weeks, month and years (Unal et al. 2004). In order to capture the spatial and temporal variability in precipitation, it is essentially important to have spatially-dense and temporally-long, accurate and reliable precipitation data. It becomes more important when the issue concerned is drought, now being one of the most harmful natural hazards (McKee et al., 1993, Mishra and Singh, 2010) due to its negative effects on the society, economy and environment. Drought is mainly categorized by its impacts on various sectors. It starts with precipitation deficit as meteorological drought and propagates into agricultural and hydrological droughts, and finally into its socioeconomical and environmental types (van Loon, 2015). A significant increase is reasonably expected in the frequency, duration and intensity of natural hazards including droughts due to the foreseen climate change (IPCC, 2013) after which natural or anthropogenic disturbances become indisputable in hydrometeorological variables. Thus, a great interest has been observed on drought not only to understand the process itself or to develop methodologies but also to extend its effect on the society, economy and environment.
Precipitation data at meteorological stations on the ground are generally used in conventional drought studies. However, number of ground stations in a certain region or across a river basin is generally not at a level to reveal the spatial change in meteorological variables, and they are not homogeneously distributed nor are they available for common users in many regions around the world. Due to limitations in the spatial distribution of meteorological stations and deficiencies in the temporal availability of the data, satellite technology has emerged (Zuo et al., 2019). Accordingly, an increasing trend has been observed on the use of satellite technology in recent years (Aksu and Arikan, 2017). Among the satellite products, satellite precipitation estimates (SPEs) have gained an obvious great importance as they are used in a number of hydro-climatological applications including drought (Yilmaz et al., 2005a,b; Tote et al., 2015; Santos et al., 2017, 2019a,b; Zambrano et al., 2017; Satge et al., 2019; Li et al., 2020; Wang et al., 2020).
A variety of SPEs are available now thanks to the progress in the satellite technology. They are Precipitation Estimation from Remote Sensing Information using Artificial Neural Network (PERSIANN) (Hsu et al., 1997), Microwave/Infrared Rain Rate Algorithm (MIRAA) (Miller et al., 2001), CPC MORPHing technique (CMORPH) (Joyce et al., 2004), Climate Hazards Group InfraRed Precipitation with Stations (CHIRPS) (Funk et al., 2014, 2015a,b), Integrated Multisatellite Retrievals for GPM (IMERG) (Huffman et al., 2015), Tropical Applications of Meteorology using Satellite and ground-based observations (TAMSAT) African Rainfall Climatology and Time series (TARCAT) (Tarnavsky et al., 2014), African Rainfall Climatology Version 2 (ARC v2.0) (Novella and Thiaw, 2013), and TRMM Multi-satellite Precipitation Analysis (TMPA) (Huffman et al., 2007) among others. SPEs can provide high spatial and temporal resolution, yet a performance assessment based on ground station measurements is required before an application is made over a region. They can be substituted for ground stations in the region for which they are once validated.
Numerous studies have been performed to evaluate the accuracy of SPEs regionally or globally (Aghakouchak, 2015; AghaKouchak et al., 2015), and a great effort has been witnessed in the literature. As a few examples to mention, Khandu et al. (2016) evaluated three satellite-based products (TRMM, CMORPH, CHIRPS) to estimate rainfall in Bhutan by correcting bias with the gamma distribution method. Similarly, Diem et al. (2019), Macharia et al. (2020) and Usman et al. (2018) used four satellite products each in different countries in Africa; and Gupta et al. (2019) in India. Santos et al. (2019a, b), Neto et al. (2020) and dos Santos et al. (2020) presented case studies from Brazil, and Chen et al. (2020) from the Yangtze River Basin in China.
Among the above SPEs, CHIRPS combines the long-term infrared (IR) remote sensing data and ground-based observations. Thus, it is relatively long enough (more than 30 years) to use for drought monitoring as well as other climatological applications. The performance of CHIRPS varies from region to region. A few recent examples on CHIRPS have been made available by Katsanos et al. (2016), Guo et al. (2017), Gao et al. (2018), Perdigon-Morales et al. (2018), and Peng et al. (2020) among many others. They are used not only to replace for ground precipitation data but it has found that they also have a useful potential in streamflow forecasting (Sulugodu and Deka, 2019). It is even possible to use them for the management of environmental conflict in transboundary river basins (Minanabadi et al., 2020). The performance of CHIRPS data was evaluated over Turkey as well, and SPEs were found consistent with ground measurements in the western and southern parts of the country particularly (Aksu and Akgul, 2020) within which Kucuk Menderes River basin, the study area of this research, is located.
Besides the fast progress in the satellite products providing SPEs, there is a wide and common methodology based on indicators or indices used in research and practiced by meteorological services when the drought is concerned. The methodology uses either meteorological or hydrological variables (i.e., precipitation, streamflow, temperature) called indicators or indices calculated from the indicators themselves. Among numerous drought indices so far developed in the literature, Standardized Precipitation Index (SPI; McKee et al., 1993) is the most commonly used due to its availability for diverse time scales and needing only precipitation (Cavus and Aksoy, 2019, 2020; Eris et al., 2020) as the input. It has therefore been suggested by the World Meteorological Organization (WMO, 2012) to national meteorological organizations as a standard for drought characterization.
In the common application, drought indices are calculated based on data recorded at the ground stations. However, as in many cases, the ground stations are sparsely and unevenly distributed over the study area, and they might not have uninterrupted long records. This creates problems in performing a proper spatial and temporal drought analysis (Kalisa et al., 2020). This is the case for Kucuk Menderes River Basin in western Turkey which is greatly important for agriculture and hence requiring water to irrigate agricultural lands, i.e. the river basin is not well equipped with ground stations (Eris et al., 2020). SPEs which blend the remote-sensed data with existing ground-based measurements could be proposed as an alternative solution to overcome this difficulty. Aksu and Akgul (2020) have validated the CHIRPS SPEs for Turkey and concluded that they were performed well in the western part of the country particularly. The scarcity of ground stations combined with the availability of the well performed SPE has been the motivation for this study which aims at (i) evaluating the performance of CHIRPS at the 3-month time scale over the Kucuk Menderes River basin; and (ii) mapping the spatio-temporal variability over the river basin by using SPI. The study proceeds as follows: First, study area and meteorological stations are introduced, CHIRPS data and bias correction method are explained next. SPI as the comparison methodology is then given, followed by the assessments and comparison of the CHIRPS with ground stations. Spatio-temporal analysis of the drought is made before the conclusions are listed at the end.