Assessment of Drought Severity and Their Spatio-Temporal Variations in the Hyper Arid Regions of Kingdom of Saudi Arabia: A Case Study from Al-Lith and Khafji Watersheds

Abstract

The goal of this study is to calculate meteorological drought using the Standard Precipitation Index (SPI) and Standard Precipitation Evapotranspiration Index (SPEI) for the Al-Lith and Khafji basins of the Kingdom of Saudi Arabia (KSA) from 2001 to 2020. The in situ (rain gauges, RGs) and Integrated Multi-satellite Retrievals for GPM (IMERG) data are used in the current study. The meteorological drought is monitored across the AL-Lith and Khafji watersheds. The climate of the Khafji watershed is like the climate of Al-Lith to some extent. Still, due to complex terrain, Al-Lith receives relatively high precipitation and has a higher average temperature than the Khafji watershed. Results show that the total drought periods observed are 166 and 139 months based on SPEI and SPI on a multiple time scale (1, 3, 6, and 12 months) in the Al-Lith watershed, respectively. While, based on SPEI and SPI, the Khafji watershed experienced a drought of 129 and 72 months, respectively. This finding indicates that the SPEI-calculated drought is more severe and persistent in both watersheds than the SPI-calculated drought. Additionally, the correlation coefficient (CC) between SPI and SPEI is investigated; a very low correlation is observed at a smaller scale. CC values of 0.86 and 0.93 for Al-Lith and 0.61 and 0.79 for the Khafji watershed are observed between SPEI-1/SPI-1 and SPEI-3/SPI-3. However, the correlation is significant at high temporal scales, i.e., 6 and 12 months, with CC values of 0.95 and 0.98 for Al-Lith and 0.86 to 0.94 for the Khafji watershed. Overall, the study compared the performance of IMERG with RGs to monitor meteorological drought, and IMERG performed well across both watersheds during the study period. Therefore, the current study recommends the application of IMERG for drought monitoring across data-scarce regions of KSA. Furthermore, SPEI estimates a more severe and long-lasting drought than SPI because of the temperature factor it considers.

1. Introduction

Water scarcity is becoming more prevalent worldwide due to increased demand for water owing to the significant increase in human population, climate change, and water pollution [1]. Droughts intensify the problems of both surface and groundwater resources, resulting in decreased water availability, lower water quality, food scarcity, and flood plain disruption [2]. As a result, experts in hydrology, climatology, and agriculture have concentrated their efforts on better understanding and modelling drought. Drought is produced by inadequate precipitation over a longer duration, such as a season or a year, ending in insufficient moisture retained in the soil [3]. Droughts are among the most dangerous hazard events, wreaking havoc on people’s lives, including plants, animals, and other natural resources, such as water, biodiversity, and ecology [4,5,6]. Among various natural hazards, floods occur at 11%, while drought occurs at 7.5% and is the second-most geographically widespread disaster globally [7]. According to Día [8], drought-prone countries require an efficient early warning system capable of reliable drought prediction, accurate drought monitoring, adequate information, and plans to reduce and mitigate drought.

Precipitation and evapotranspiration are two essential parameters in monitoring drought [9]. Precipitation is one of the most complicated natural phenomena in the hydrological cycle, with enormous geographical and temporal diversity [10,11]. Precise and accurate precipitation measurement is very important. It is the primary factor for predicting climate change, environmental research, water resource assessment, and hydrological extremes such as drought monitoring and flood forecasting [11]. Several methods have been employed to measure the precipitation. Rain gauges (RGs) have traditionally been used to measure precipitation. RGs precipitation records are considered the most dependable point observations and reasonably depict precipitation’s temporal variability [12]. On the other hand, RG networks are too sparse to adequately explain the geographic variability in precipitation [12]. Therefore, there is an urgent need for high-quality satellite gridded data to provide continuous precipitation data and reflect both temporal and spatial precipitation variability in countries such as KSA, where RGs are not frequent in mountainous areas. Most RGs in KSA are centred in the lowlands [13]. Remote sensing and retrieval advancements enabled the development of alternative precipitation data sources, namely satellite precipitation products (SPDs). These SPDs calculate precipitation amounts based on cloud properties as determined by infrared (IR), visible (VI), and microwave (MW) satellite images [14,15]. For example, PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks method) [16], CMORPH (Climate Prediction Center Morphing) [17], and TRMM (Tropical Rainfall Measuring Mission) [18] combine MW and IR to take advantage of their complementary qualities. National Aeronautics Space Administration (NASA) and Japan Aerospace Exploration Agency (JAXA) recently collaborated to launch the Global Precipitation Measurement (GPM) satellite in 2014 [19], following TRMM’s spectacular accomplishment. It is composed of one primary observatory satellite and ten additional partner satellites that are equipped with an advanced Dual-frequency Precipitation Radar (DPR), a GPM Microwave Imager (GMI), and other innovative instruments [20].

GPM has four categories of data, i.e., Level 0, Level 1, Level 2, and Level 3 [19,21]. They further elaborated that Level 3 data are suitable for researchers provided by IMERG. NASA also provides three main products for IMERG, i.e., Early run (near-real-time run products), Late run, and Final run products. El Kenawy et al. [22] examined the effectiveness of TRMM, CMORPH, and PERSAINN ANN globally high-resolution remotely sensed products in simulating the principal properties of precipitation over the Middle East, focusing on extreme precipitation occurrences. Recently, a study validated the performance of IMERG and its different products over KSA [23]. The study demonstrated that IMERG’s final run product outperformed other products in reliable and accurate precipitation estimation in arid regions. Therefore, the current study uses the IMERG SPD and RGs to monitor drought over two different basins of KSA.

Many drought monitoring tools and indices are available. Svoboda and Fuchs [24] provided a detailed description of the most used drought indices. The widely used drought indices include the Palmer drought severity index (PDSI), Reconnaissance Drought Index (RDI), Standardized Precipitation Index (SPI), and Standardized Precipitation Evapotranspiration Index (SPEI) for monitoring droughts [25,26,27]. The PDSI index, despite its many positives, has significant flaws. A lack of ability to produce both short-term and long-term effects of droughts is the problem with this index [28]. The current study considers SPI and SPEI to monitor meteorological drought across two different catchments of KSA. SPI is based on monthly precipitation data and can easily calculate drought over multiple time scales. However, SPI cannot portray the whole picture, including variations in soil moisture associated with changes in groundwater, reservoirs, and streamflow responses to long-term precipitation anomalies [29]. Therefore, different new drought indices were developed to address the different aspects of drought. SPEI is an extension of the widely used SPI index, a relatively new and comprehensive drought index [29]. This index is based on the water balance calculation and requires monthly precipitation data and air temperature for its analysis. Thus, it combines the sensitivity to variation in evaporation of PDSI with SPI multi-time scale calculation and its simplicity [30].

The Middle East is one of the most drought-prone regions due to extreme heat and aridity [31]. According to Malick et al. [32] the Arab world has been going through a severe drought since 1970 because of less rain, less humidity, and worse climate conditions. KSA is tropical and occupies approximately 80% of the Arabian Peninsula [33]. KSA and the other Gulf Cooperation Council (GCC) countries are already classified as water-scarce by the United Nations [34]. Procházka et al. [35] estimate that freshwater availability in KSA has shrunk by 75% since 1950, with an additional 50% reduction predicted by 2030. Elhag and Bahrawi [36] evaluate hydrological drought indices in different locations in Saudi Arabia. The study depicted that the main reason for water stress in an arid country such as KSA is the heavy abstraction of groundwater and the high evaporation rate during long summer days. Almazroui [37] used SPI and PDSI to monitor the meteorological drought in KSA. The study reported that droughts of all types occurred more frequently in the dry season in KSA, with less in the rainy season. Prior research in Kingdom of Saudi Arabia mainly focused on hydrological drought rather than meteorological drought [38]. Alsubih et al. [31], utilized the short-term Standardized Precipitation Index (SPI-6) index to estimate meteorological drought conditions in the Asir region of Saudi Arabia between 1970 and 2017. The innovative part of this study is the application of it to a hyper-dry region that has received less attention in terms of meteorological drought evaluations. In addition to it, the application of IMERG data for monitoring drought in KSA is due to the sparse distribution of RGs. Therefore, the objectives of the current research are: (i) to evaluate the performance of IMERG data against RGs in Al-Lith and Khafji between 2001 and 2020, (ii) to investigate the comparative analysis between SPI and SPEI by monitoring the duration and severity of drought in Al-Lith and Khafji, (iii) to analyze the role of climate and topography on frequency and severity of drought, (iv) recommendation of an alternative dataset (IMERG) to the sparsely distributed RGs. Evaluating the potential of IMERG against RGs in Al-Lith and Khafji will help the water resources planner and researcher to devise a plan for drought mitigation in KSA.

2. Study Area

The study is conducted across two important regions of KSA, including the Al-Lith and Khafji watersheds, as shown in Figure 1. Al-Lith is located on the Red Sea coast, southwest of Makkah, between 20°9′00″ N longitude and 41°16′00″ E latitude, with an elevation between 0 and 2663 m above sea level. Al-Lith has an average high temperature of 41.9 °C in July, followed by June and August, having 41.3 °C and 41.2 °C, respectively. In contrast, the average lower temperature in Al-Lith is 20.0 °C in January. June and July are the driest months with 2 mm, while November, December, and January are the wettest months with an average precipitation of 21 mm.

Khafji is between 28°25′00″ N and 48°30′00″ E. In Khafji, an average high temperature of 40.9 °C occurred in August, followed by July and June, recorded at 40.8 °C and 39.4 °C, respectively. While the average lower temperatures of 8.7 °C and 9.8 °C occurred in February and January, respectively. The driest months are from June to September in the Khafji watershed with 0 mm precipitation, while the wettest months are January and November with 25 mm and 21 mm precipitation, respectively.

3. Materials and Methods

3.1. Methodology

The methodology used for this research is categorized into (i) meteorological data acquisition, (ii) comparison between RGs and SPD data, and (iii) drought monitoring using SPEI and SPI indices for (1, 3, 6, and 12 months) time scale, and (iv) finding the correlation between SPEI and SPI. The detailed flow chart used in this research is shown in Figure 2.

3.2. Meteorological Data Acquisition

RGs and IMERG data are used in Al-Lith and Khafji watersheds during the study period from 2001–2020, as shown in Figure 3. The daily precipitation and temperature records across Al-Lith and Khafji watersheds have been collected from the Ministry of Water Environment and Agriculture (MEWA), as shown in

Table 1. Name and coordinates of Stations in Al-Lith and Khafji modified from Almazroui et al. [40].

Station Name	Elevation (m)	Annual Rainfall (mm)	Variance	Standard Deviation	Skewness	Kurtosis
TA 233	2640	145.21	14,988.9	122.43	2.11	6.60
TA109	2632	167.67	3240.993	56.93	−0.63	−0.405
J107	88	56.72	2915.471	53.99	1.63	3.56
J108	20	47.56	2045.15	45.23	1.89	3.55
Meteorological section Terminal and Marine Department Khafji	16	98.95	5256.82	72.50	1.19	0.48

. The monthly SPD obtained from the IMERG GPM version 3 (V06) for 2001–2020 has been downloaded and can be freely accessed at [39]. IMERG dataset of one-month temporal resolution and a spatial resolution of 0.1° × 0.1° based on PMW (Passive Microwave) and IR (Infrared) data from the GPM constellation satellite is used in this research.

MEWA collects the majority of the precipitation records from RGs manually and may be vulnerable to various errors. External issues such as splashing and wind errors might influence the data quality of RGs at high altitudes. Furthermore, quality checks such as skewness and kurtosis were conducted on the collected RG data to ensure high-quality data. The missing data were filled using the zero-order technique described by Rahman et al. [41].

3.3. Standardized Precipitation Index (SPI)

SPI was developed by McKee et al. [3], which uses the gamma function and probability distribution of precipitation over time. With the help of SPI, individuals can monitor the severity, frequency, and duration of drought. The following equations are used to calculate the SPI index.

$$f\left(x\right)=\frac{1}{{\beta }^{\alpha }\mathrm{\Gamma }\left(\alpha \right)}{x}^{\alpha -1}{e}^{-\frac{x}{\beta }},x>0.$$

(1)

where $\alpha >0$. represents the shape factor, $\beta >0$ represents the scale parameter, the amount of precipitation is denoted by $x>0$, and $\mathrm{\Gamma }\left(\alpha \right)$ represents the gamma function.

For the precipitation $x_0$ in a given year, the chance that the random variable x is more minor than $x_0$ may be computed as follows:

$$\hat{\alpha }=\frac{1}{4\mathrm{A}}(1+\sqrt{1+\frac{4\mathrm{A}}{3}})$$

(2)

and

$$\hat{\mathsf{\beta }}=\overline{\mathrm{x}}/\hat{\alpha }$$

(3)

where $\overline{\mathrm{x}}$ represent the mean precipitation, and A is given by

$$A=\ln \left(\overline{x}\right)-n^{-1}{\displaystyle \sum }\ln \left(x\right)$$

(4)

The cumulative probability G(x) of an observed amount of precipitation for a given month and time period is given by:

$$\mathrm{G}(\mathrm{x})=\frac{1}{{\hat{\mathsf{\beta }}}^{\hat{\mathsf{\alpha }}}\mathrm{\Gamma }(\hat{\mathsf{\alpha }})}{\int }_0^{\mathrm{X}}{\mathrm{x}}^{\hat{\mathsf{\alpha }}}{\mathrm{e}}^{-\mathrm{x}/\hat{\mathsf{\beta }}}\mathrm{d}\mathrm{x}$$

(5)

Letting $\mathrm{t}=\mathrm{x}/\hat{\mathsf{\beta }}$ we reduce the function to the following equation

$$\mathrm{G}(\mathrm{x})\frac{1}{\mathrm{\Gamma }(\hat{\mathsf{\alpha }})}{\int }_0^{\mathrm{x}}{\mathrm{t}}^{\hat{\alpha }-1}{\mathrm{e}}^{-1}\mathrm{d}\mathrm{t}.$$

(6)

For x = 0, the gamma distribution is not defined, and because the probability of zero precipitation, $\mathit{q}=\mathrm{P}(\mathit{x}=0)$, is positive, the cumulative probability is

$$H\left(x\right)=q+\left(1-q\right)G\left(x\right)$$

(7)

The cumulative probability, H(x), is then transformed into a standard normal random variable Z with a mean of zero and a variance of one, obtaining values of SPI.

3.4. Standardized Precipitation Evaporation Index (SPEI)

SPEI is an extension of the widely used SPI index, a relatively new and comprehensive drought index [29]. This index is based on the water balance calculation and requires monthly precipitation data and potential evapotranspiration (PET) for its analysis. SPEI is usually calculated by normalizing the precipitation and PET differences using the log-logistic probability distribution function. Many methods are used for PET calculation, but the Hargreaves–Samani equation proposed by Hargreaves and Samani (1985) was used to compute PET in this article [42].

$${\mathrm{PET}}_{\mathrm{i}}=0{.0135\mathrm{K}}_{\mathrm{T}}\left(\mathrm{T}+17.78\right){\left({\mathrm{T}}_{\max }{-\mathrm{T}}_{\min }\right)}^{0.5}{\mathrm{R}}_{\mathrm{a}}$$

(8)

Equation (6) is suggested by many researchers in their findings when solar radiation, relative humidity, and wind speed are unavailable [28,29].

${\mathrm{D}}_{\mathrm{i}}$. denotes the monthly water balance, which is estimated by using the calculated ${\mathrm{PET}}_{\mathrm{i}}$ and precipitation (P) data

$${\mathrm{D}}_{\mathrm{i}}{=\mathrm{P}}_{\mathrm{i}}{-\mathrm{PET}}_{\mathrm{i}}.$$

(9)

The three-parameter log-logistic probability distribution with three parameters is expressed as under:

$$\mathrm{f}(\mathrm{x})=\frac{\mathsf{\beta }}{\mathsf{\alpha }}(\frac{\mathrm{x}-\mathsf{\gamma }}{\mathsf{\alpha }}{)}^{\mathsf{\beta }-1}[+(\frac{\mathsf{x}-\mathsf{\gamma }}{\mathsf{\alpha }}{)}^{\mathsf{\beta }}{]}^{-2}$$

(10)

The cumulated distribution function is calculated using the following equation:

$$\mathrm{F}\left(\mathrm{x}\right)={\left[1+{\left(\frac{\mathsf{\alpha }}{\mathrm{x}-\mathsf{\gamma }}\right)}^{\mathsf{\beta }}\right]}^{-1}.$$

(11)

where α is the scale parameter, β is the form parameter, γ is the beginning parameter, and x is the mean of the series of D values in a given period.

$$\mathrm{S}\mathrm{P}\mathrm{E}\mathrm{I}=\mathrm{W}-\frac{{\mathrm{C}}_0+{\mathrm{C}}_1\mathrm{W}+{\mathrm{C}}_2{\mathrm{W}}^2}{1+{\mathrm{d}}_1\mathrm{W}+{\mathrm{d}}_2{\mathrm{W}}^2+{\mathrm{d}}_3{\mathrm{W}}^3}$$

(12)

where W = −2Ln(P) for p ≤ 0.5 and P is the probability of exceeding a determined D value. The constants are C₀ = 2.515517, C₁ = 0.802853, C₂ = 0.010328, d₁ = 1.432788, d₂ = 0.189269 and d₃ = 0.001308.

The SPI and SPEI are frequently used indices for drought monitoring. They can be estimated on various timeframes, including 1, 3, 6 and 12 months, and can be used to characterize distinct kinds of droughts. A short timeline (1 month or 3 months) is used to assess meteorological drought; a 3-month or 6-month timescale is often used to describe agricultural drought; and a longer timescale, such as 12 or 24 months, is more appropriate for evaluating hydrological drought and water resources [2]. This research calculated SPEI and SPI at 1-, 3-, 6-, and 12-month timeframes.

The drought, determined by the SPEI/SPI, is divided into seven categories (wet/drought) based on positive and negative values of SPEI and SPI, as shown in

Table 2. Shows different categories based on the calculated values of SPEI/SPI.

SPEI/SPI Range	Categories	Probability (%)
>2	Extremely wet	2
1.50 to 1.99	Severely wet	6
1.00 to1.49	Moderately wet	10
−0.99 to 0.99	Nearly Normal	65
−1.49 to −1.0	Moderately drought	10
−1.99 to −1.5	Severe drought	5
<−2	Extreme drought	2

. In this study, a threshold of −1 is used, which is suggested by many researchers [43,44]. When the SPEI/SPI value is equal to −1 or less, a drought event starts, corresponding to a 17% probability. Drought severity increases as the values decrease below −1. When the value is between −1.00 and −1.49, the drought situation is moderate with a probability of 2%; when the value is between −1.5 and −1.99, the drought condition is severe with a probability of 5%; and when the value is more than or equal to −2.0, the drought condition is extreme with a probability of 2% (refer to Table 2).

3.5. Performance of IMERG against Rain Gauges in Drought Monitoring

Four statistical indices were used to evaluate the performance of IMERG against the RGs as a reference. The indices considered in the current study are Mean Absolute Error (MAE), Root Mean Square Error (RMSE), net Root Mean square Error and Correlation coefficient (CC) [30,31]. The normalized root-mean-squared error (NRMSE) is a non-dimensional variant of RMSE in which RMSE is normalized to the observed input value range (X). This metric is better than RMSE because it is less affected by having very high or low numbers in the data [22]. The equations used for the above statistical indices are given below.

The equation used for the above statistical tests is given below.

$$\mathrm{M}\mathrm{A}\mathrm{E}=\frac{1}{\mathrm{n}}{\displaystyle \sum _{\mathrm{t}=1}^{\mathrm{N}}}|{\mathrm{O}}_{\mathrm{t}}-{\mathrm{E}}_{\mathrm{t}}|$$

(13)

$$\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E}=\sqrt{\frac{1}{\mathrm{n}}{\displaystyle \sum _{\mathrm{t}=1}^{\mathrm{N}}}({\mathrm{O}}_{\mathrm{t}}-{\mathrm{E}}_{\mathrm{t}}{)}^2}$$

(14)

$$NRMSE=\frac{RMSE}{X_{max}-X_{min}}.$$

(15)

$$\mathrm{r}=\frac{\sum \left({\mathrm{O}}_{\mathrm{t}}{}_{\mathrm{i}}-\overline{{\mathrm{O}}_{\mathrm{t}}}\right)\left({\mathrm{E}}_{\mathrm{t}}{}_{\mathrm{i}}-\overline{{\mathrm{E}}_{\mathrm{t}}}\right)}{\sqrt{\sum {\left({\mathrm{O}}_{\mathrm{t}}{}_{\mathrm{i}}-\overline{{\mathrm{O}}_{\mathrm{t}}}\right)}^2\sum {\left({\mathrm{E}}_{\mathrm{t}}{}_{\mathrm{i}}-\overline{{\mathrm{E}}_{\mathrm{t}}}\right)}^2}}$$

(16)

where ${\mathrm{O}}_{\mathrm{t}}$ represents the observed precipitation by RGs, and ${\mathrm{E}}_{\mathrm{t}}$ is the estimated precipitation by IMERG, respectively, and n is the total number of precipitation observations.

4. Results

4.1. Comparison between IMERG and RGs Data

MAE, RMSE, NRMSE, and CC are used to evaluate the performance of IMERG against RGs from 2001 to 2020 in the Al-Lith and Khafji watersheds (

Table 3. Comparison of IMERG and RGs data in Al-Lith and Al-Khafji watersheds.

Watersheds	MAE	RMSE	NRMSE	CC
Al-Lith	5.4	6.1	0.13	0.88
Khafji	6	7.4	0.15	0.81

). Both areas received much lower precipitation throughout the year in the study period, which shows that both watersheds are in a hyper-arid environment and prone to drought. MAE for Al-Lith and Khafji watershed was 5.4 mm/month and 6 mm/month, indicating the IMERG has good performance. Similarly, RMSE in Al-Lith and Khafji was 6.1 mm/month and 7.4 mm/month, respectively. The correlation coefficient between RGs data and IMERG is 0.88 in the Al-Lith watershed, while in Khafji, CC is 0.81.

4.2. Evaluation of Drought

The SPEI and SPI at a four-time scale (1, 3, 6, and 12 months) were estimated for Al-Lith and Khafji to illustrate the drought’s temporal variability, as depicted in Figure 4 and Figure 5. Figure 4 and Figure 5 demonstrate that, for both SPEI and SPI, the wet and dry changes are more pronounced as the time scale decreases from one month to three months. As timescales grew longer, the variations in the SPEI and SPI became more gradual, the fluctuations diminished, and the characteristics of interannual and interdecadal changes became more evident. In other words, the disparity between the two indices decreased as the timescale increased. The SPEI categorises the drought into three distinct classes (moderate, severe, and extreme). Figure 4 and Figure 5 depict the spectrum of drought patterns, durations, and severity. In addition, both drought indices exhibit a similar pattern of variation for each time and watershed, but they differ in terms of drought duration and severity, respectively. This may be explained because each watershed experiences significant natural temperature variation.

4.3. Drought Monitoring Using SPEI

4.3.1. SPEI-1

Figure 4a,b show that there are greater fluctuations between dry and wet periods in both watersheds at a smaller scale (SPEI-1). According to SPEI-1 estimates, the Al-Lith watershed experienced moderate drought in 2001, 2002, 2006, 2013, 2014, and 2017 as shown in Figure 4a. Severe droughts occurred in 2004, 2015, and 2020. According to SPEI-1, the extreme drought in the Al-Lith watershed occurred in 2008 and 2018. From January 2001 to December 2020, the total drought periods (moderate, severe, and extreme) occurred in Al-Lith in 37 months based on SPEI-1, as shown in

Table 4. Yearly drought frequency in Al−Lith watershed based on SPEI for (1, 3, 6 and 12 months).

	SPEI 1				SPEI 3				SPEI 6				SPEI 12
Years	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ
2001	1	-	-	1	1	-	-	1	1	-	-	1	-	-	-	0
2002	2	1	-	3	1	2	-	3	-	1	-	1	2	-	-	2
2003	1	-	-	1	2	-	-	2	1	-	-	1	-	-	-	0
2004	-	1	-	1	2	1	-	3	2	1	-	3	-	-	-	0
2005	3	-	-	3	3	1	-	4	-	2	-	2	-	-	-	0
2006	1	-	-	1	4	-	-	4	-	2	2	4	1	-	-	1
2007	2	1	-	3	-	3	-	3	1	2	-	3	-	-	-	0
2008	1	1	1	3	1	2	1	4	2	4	1	7	3	2	-	5
2009	2	1	-	3	1	-	-	1	-	-	1	1	2	2	2	6
2010	2	1	-	3	-	2	-	2	-	1	-	1	1	-	-	1
2011	3	-	-	3	4	-	-	4	4	-	-	4	-	-	-	0
2012	-	1	-	1	-	-	-	0	-	5	-	5	-	2	-	2
2013	1	-	-	1	-	-	-	0	1	-	-	1	3	5	-	8
2014	-	-	-	0	2	-	-	2	1	-	-	1	4	-	-	4
2015	1	1	-	2	2	1	-	3	2	-	-	2	3	4	-	7
2016	2	-	-	2	-	-	-	0	-	-	-	0	-	3	-	3
2017	-	-	-	0	1	-	-	1	2	-	-	2	-	-	-	0
2018	-	-	1	1	1	1	-	2	-	3	-	3	-	-	3	3
2019	1	-	-	1	-	-	-	0	-	-	-	0	4	2	-	6
2020	-	4	-	4	-	-	-	0	-	-	-	0	-	-	-	0
Total	23	12	2	37	25	13	1	39	17	21	4	42	23	20	5	48

. Out of a total of 37 months, 23 months have experienced moderate, 12 severe, and 2 are extreme drought events.

On the other hand, the Khafji watershed experienced moderate droughts in 2002, 2005, 2006, 2007, 2013, 2018, 2019, and 2020 as shown in Figure 4b. In Khafji, severe droughts occurred in 2015 and 2017. The extreme drought in the Khafji watershed occurred in 2010 and 2017.

Table 5. Yearly drought frequency in Khafji watershed based on SPEI for (1, 3, 6 and 12 months).

	SPEI 1				SPEI 3				SPEI 6				SPEI 12
Years	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ
2001	2	2	-	4	1	2	-	3	3	1	-	4	-	-	-	0
2002	2	-	-	2	1	1	-	2	-	-	-	0	2	-	-	2
2003	-	-	-	0	1	-	-	1	2	-	-	2	1	-	-	1
2004	2	-	1	3	2	-	-	2	3	-	-	3	-	-	-	0
2005	1	-	-	1	1	-	-	1	2	-	-	2	-	-	-	0
2006	1	-	-	1	2	2	-	4	1	-	-	1	-	-	-	0
2007	1	-	-	1	-	-	-	0	-	-	-	0	-	-	-	0
2008	-	-	-	0	-	-	-	0	2	2	-	4	6	-	-	6
2009	-	-	-	0	-	-	-	0	-	-	-	0	-	-	-	0
2010	-	-	2	2	1	1	-	2	1	-	-	1	-	-	-	0
2011	-	-	-	0	-	2	-	2	-	2	-	2	3	1	1	5
2012	-	-	-	0	-	-	-	0	-	-	-	0	-	-	-	0
2013	2	-	-	2	1	-	-	1	-	-	-	0	-	-	-	0
2014	-	-	-	0	-	-	-	0	-	-	-	0	-	-	-	0
2015	1	1	-	2	4	1	-	5	6	2	-	8	5	4	-	9
2016	-	-	-	0	-	-	-	0	1	-	-	1	-	-	-	0
2017	3	1	2	6	2	1	3	6	1	6	2	9	7	1	-	8
2018	1	-	-	1	1	3	-	4	-	-	-	0	-	6	5	11
2019	1	-	-	1	-	-	-	0	-	-	-	0	-	-	-	0
2020	1	-	-	1	-	-	-	0	-	-	-	0	-	-	-	0
Total	18	4	5	27	17	13	3	33	22	13	2	37	24	12	6	42

shows that a total drought period of 27 months occurred in the Khafji watershed. Out of which 18 months are moderate, 4 months are severe, and 5 months are extreme drought conditions. More extreme and moderate drought has been observed in the Khafji watershed compared to the Al-Lith watershed.

4.3.2. SPEI-3

SPEI-3, like SPEI-1, frequently fluctuated around zero, indicating the variability of dry and wet conditions. SPEI-3 fluctuates less than SPEI-1 because the 3-month SPEI value compares accumulated precipitation during that specific 3-month period to the annual mean precipitation total acquired throughout the research period. Based on SPEI-3, the Al-Lith watershed experienced moderate drought in 2001, 2003, 2005, 2006, 2008, 2009, 2010, 2011, 2014, 2015, and 2017 as shown in Figure 4c. Severe droughts occurred in 2002, 2004, 2007, 2010, 2015, and 2018. According to SPEI-3, the extreme drought in the Al-Lith watershed occurred in 2008. According to Table 4, 39 months of drought have been observed in the Al-Lith watershed in the study period (2001–2020) based on SPEI-3 with 25 months of moderate, 13 months of severe, and 1 month of extreme drought.

On the other hand, the Khafji watershed experienced moderate droughts in 2001, 2002, 2004, 2005, 2007, 2013, 2018, 2019, and 2020 as shown in Figure 4d. In Khafji, severe droughts occurred in 2002, 2006, 2010, 2011, 2015, 2017, and 2018. The extreme drought in the Khafji watershed occurred in 2017. Table 5 shows that the total drought periods in the Khafji watershed were 33 months based on SPEI-3 from January 2001 to December 2020. Out of 33 months, 17 months are moderate, 13 months are severe, and 3 months are extreme drought periods. In short, the results show that overall drought is more severe in the Al-Lith watershed than in the Khafji watershed. However, the Khafji watershed experienced more extreme drought, while moderate and severe drought was the same in both watersheds.

4.3.3. SPEI-6

When the time frame increases from SPEI-1/SPEI-3 to SPEI-6/SPEI-12, the variability decreases, and we see prolonged wet and dry periods. According to SPEI-6, the worst drought period in Al-Lith occurred between November 2007 to August 2008 (Figure 4e). Based on SPEI-6, the Al-Lith watershed experienced moderate drought in 2001, 2003, 2004, 2007, 2008, 2009, 2011, 2013, 2014, and 2017. Severe drought events were observed in 2002, 2004, 2005, 2006, 2007, 2008, 2010, 2012, and 2009 while extreme drought in the Al-Lith watershed occurred in 2006, 2008, and 2018. Table 4 shows the total drought periods in Al-Lith are 42 months, based on which 17 months are moderate, 21 months are severe, and 4 months are extreme drought periods.

On the other hand, the worst drought is observed in Khafji from August 2017 to October 2018, as shown in Figure 4f. Khafji watershed experienced moderate droughts in 2001, 2003, 2004, 2005, 2006, 2008, 2010, 2015, 2016, and 2017. There were severe droughts in 2002, 2008, 2011, 2015, and 2017 while there was an extreme drought in the Khafji watershed in 2017. Table 5 shows that 37-month drought periods occurred in the Khafji watershed, of which 22 were moderate, 13 severe, and 2 were extreme drought periods. In short, the Al-Lith watershed experienced more severe and extreme drought than the Khafji watershed.

4.3.4. SPEI-12

Based on SPEI-12, the Al-Lith watershed experienced moderate drought in 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2010, 2011, 2014, and 2018 as shown in Figure 4g. Severe droughts occurred in 2008, 2012, 2013, 2015, 2016, and 2019. According to SPEI-12, the extreme drought in the Al-Lith watershed occurred in 2009 and 2017. Table 4 shows the total drought periods in Al-Lith in 48 months based on SPEI-12 in the study period (2001–2020), where 23 months of moderate, 20 months of severe, and 5 months of extreme drought occurred in Al-Lith based on the SPEI-12 index.

On the other hand, the worst drought occurred in Khafji between November 2017 and December 2018, as shown in Figure 4h. Khafji watershed experienced moderate droughts in 2002, 2003, 2008, 2011, 2015, and 2017. Severe droughts in 2011, 2015, 2017, and 2018. The extreme drought in the Khafji watershed occurred in 2011 and 2018.

Table 6. Yearly drought frequency in Al−Lith watershed based on SPI for (1, 3, 6, and 12 months).

	SPI 1				SPI 3				SPI 6				SPI 12
Years	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ
2001	2	1	-	3	2	-	-	2	2	2	1	5	-	-	-	0
2002	-	-	-	0	3	-	-	3	-	1	-	1	1	-	-	1
2003	-	1	-	1	1	-	-	1	-	-	-	0	-	-	-	0
2004	-	1	-	1	2	1	-	3	1	-	-	1	-	-	-	0
2005	3	1	-	4	1	-	-	1	1	-	-	1	-	-	-	0
2006	1	1	-	2	1	1	2	4	-	1	-	1	2	-	-	1
2007	2	1	-	3	-	1	-	1	2	-	-	2	-	-	-	0
2008	1	2	-	3	-	2	2	4	1	1	-	2	4	1	-	5
2009	2	-	-	2	1	-	-	1	-	-	-	0	3	-	-	3
2010	-	-	-	0	-	-	-	0	-	2	-	2	1	1	-	2
2011	2	1	-	3	4	1	-	5	-	-	-	0	-	-	-	0
2012	-	1	-	1	-	-	-	0	1	-	-	1	4	-	-	4
2013	2	-	-	2	-	-	-	0	-	-	-	0	2	4	-	6
2014	-	-	-	0	2	-	-	2	-	-	-	0	-	-	-	0
2015	-	-	-	0	-	-	-	0	1	2	2	5	4	2	-	5
2016	-	-	-	0	-	-	-	0	-	-	-	0	-	4	-	4
2017	-	-	-	0	-	-	-	0	3	-	-	3	-	-	-	0
2018	-	-	-	0	1	-	-	1	1	2	3	6	4	-	-	4
2019	1	-	-	1	1	-	-	1	3	2	-	5	-	-	-	0
2020	2	-	-	2	-	-	-	0	-	-	-	0	-	-	-	0
Total	18	10	0	28	19	6	4	29	16	13	6	35	25	12	0	37

shows the total drought periods in Khafji in 42 months based on SPEI-12 in the study period (2001–2020), of which 24 are moderate, 12 severe, and 6 are extreme drought periods. Al-Lith experienced more moderate and severe drought than the Khafji watershed based on SPEI-12, and also, the total drought period is greater in the Al-Lith watershed and Khafji watershed.

4.4. Drought Monitoring Using SPI

4.4.1. SPI-1

SPI-1 estimates show that the Al-Lith watershed experienced moderate drought in 2001, 2005, 2006, 2007, 2008, 2009, 2013, 2019, and 2020 shown in Figure 5a. Severe droughts occurred in 2001, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2011, and 2012. However, no extreme drought events were observed in the Al-Lith watershed during the study period.

Table 7. Yearly drought frequency in Khafji watershed based on SPI for (1, 3, 6, and 12 months).

	SPI 1				SPI 3				SPI 6				SPI 12
Years	Moderate	Severe	Extreme	Σ	Moderate	Σ	Extreme	Σ	Moderate	Severe	Extreme	Σ	Moderate	Severe	Extreme	Σ
2001	1	2	-	3	-	3	-	3	1	2	1	4	-	-	-	0
2002	1	-	-	1	2	-	-	2	1	-	-	1	-	-	-	0
2003	-	-	-	0	-	-	-	1	-	1	-	1	-	-	-	0
2004	-	1	-	-	-	-	-	0	1	-	-	1	-	-	-	0
2005	-	-	-	0	-	-	-	0	-	-	-	0	-	-	-	0
2006	2	-	-	2	1	-	-	1	-	-	-	0	-	-	-	0
2007	-	-	-	0	-	-	-	0	-	-	-	0	-	-	-	0
2008	1	-	1	2	2	-	-	2	2	1	-	3	-	-	-	0
2009	-	-	-	0	-	-	-	0	-	2	-	2	1	-	-	1
2010	-	-	1	-	-	-	-	0	-	-	-	0	-	-	-	0
2011	-	-	-	0	-	-	1	1	-	-	-	0	1	-	-	1
2012	-	-	-	0	-	-	-	1	-	-	-	0	-	-	-	0
2013	1	-	-	1	-	-	-	2	-	-	-	0	-	-	-	0
2014	1	-	-	1	-	-	-	0	2	3	2	7	-	-	-	0
2015	2	1	-	3	2	-	1	3	-	-	-	0	2	6	2	10
2016	-	-	-	0	-	-	-	0	2	1	-	3	-	-	-	0
2017	1	2	-	3	1	-	-	1	-	-	-	0	8	2	-	10
2018	-	2	-	2	1	-	3	4	1	-	3	4	1	9	1	11
2019	-	-	-	0	-	-	-	0	-	-	-	0	-	-	-	0
2020	-	-	-	0	-	-	-	0	-	-	-	0	-	-	-	0
Total	10	8	2	18	9	3	5	21	10	10	6	26	13	17	3	33

shows that the total drought periods in Al-Lith are 28 months, out of which 18 months are moderate, 10 months are severe, and no extreme drought periods.

Khafji watershed, On the other hand, experienced moderate droughts in 2001, 2002, 2006, 2008, 2013, 2014, 2015, and 2017 shown in Figure 5b. In Khafji, severe droughts occurred in 2001, 2004, 2015, 2017, and 2018. Further, an extreme drought occurred in 2008 and 2010. Khafji watershed experienced a total of 20 months of drought periods (Table 7), where 10 months were moderate, 8 months were severe, and 2 months were extreme drought periods. Similar to SPEI-1, SPI-1 results show that Al-Lith experienced more drought periods than the Khafji watershed.

4.4.2. SPI-3

SPI-3 results show that the Al-Lith watershed experienced moderate drought in 2001, 2002, 2003, 2004, 2005, 2006, 2009, 2011, 2013, 2014, 2018, and 2019 shown in Figure 5c. Severe droughts occurred in 2004, 2006, 2007, 2008, and 2011, while the extreme drought occurred in 2009 and 2017 in the Al-Lith watershed. Table 6 shows that the total drought periods in Al-Lith are 48 months, which consists of 23 months of moderate, 20 months of severe, and 5 months of extreme drought.

The Khafji watershed experienced moderate droughts in 2002, 2006, 2008, 2013, 2014, 2015, and 2017 shown in Figure 5d. In the Khafji watershed, severe droughts occurred in 2001, 2015, and 2017, and extreme in 2011, 2015, and 2018. Table 7 shows that the 22 months of total drought occurred in the Khafji watershed, where 11 months are moderate, 6 months are severe, and 5 months are extreme drought conditions.

4.4.3. SPI-6

Figure 5e reveals that the Al-Lith watershed experienced moderate drought in 2001, 2002, 2006, 2007, 2008, 2012, 2015, 2017, 2018, and 2019. Severe droughts occurred in 2001, 2002, 2006, 2008, 2010, 2015, 2018, and 2019 while the extreme drought occurred in 2001, 2015, and 2018 in the Al-Lith watershed. Table 6 shows that the total drought periods in Al-Lith are 33 months based on SPI-3, consisting of 14 months of moderate, 13 months of severe, and 6 months of extreme drought.

Khafji watershed experienced moderate droughts (Figure 5f) in 2001, 2002, 2004, 2008, 2014, 2016, and 2018. Severe droughts were observed in 2001, 2003, 2008, 2009, 2014, and 2016, while extreme in 2001, 2014, and 2018. Table 7 shows that the 26 months of total drought periods that occurred in the Khafji watershed out of 10 months are moderate, 6 months are severe, and 6 months are extreme drought conditions.

4.4.4. SPI-12

SPI-12 results (Figure 5g) depicted that the Al-Lith watershed experienced moderate drought in 2002, 2006, 2008, 2010, 2009, 2010, 2012, 2013, 2015, and 2018. Severe droughts occurred in 2008, 2010, 2013, 2015, and 2016, whereas there was no extreme drought in the Al-Lith watershed. Table 6 shows the total drought periods in Al-Lith are 37 months, consisting of 25 months of moderate, 12 months of severe, and no extreme drought.

Khafji watershed experienced moderate droughts in 2009, 2011, 2015, 2017, and 2018 as shown in Figure 5h based on SPI-12. In the Khafji watershed, severe droughts occurred in 2015, 2017, and 2018 and extreme in 2005 and 2018. Table 7 shows the 33 months of total drought in the Khafji watershed based on SPI-12, of which 13 are extreme drought conditions.

4.5. Correlation between SPEI and SPI

The correlation coefficient (CC) between SPEI and SPI is calculated for both Al-Lith and Khafji watersheds during 2001–2020. The finding reveals that SPEI and SPI showed a strong and significant correlation at 12 months. The correlation coefficient between SPEI-12/SPI-12 is 0.98 when p < 0.01 in the Al-Lith watershed. However, the CC drops when moving from SPEI-12/SPI-12 to SPEI-1/SPI-1. For instance, CC = 0.95, CC = 0.93 and CC = 0.86 when p < 0.001 is observed for SPEI-6/SPI-6, SPEI-3/SPI-3 and SPEI-1/SPI, respectively, as shown in Figure 6. This means that correlation increases when the timeframe between two indices increases.

Like the Al-Lith watershed, the Khafji watershed also shows a strong correlation between SPEI and SPI at a high timescale (12 months). The correlation between SPEI and SPI decreases wi th a decrease in timescale. The CC values of 0.94, 0.90, 0.79, and 0.69 when p < 0.001 are observed between SPEI-12/SPI-12, SPEI-6/SPI-6, SPEI-3/SPI-3 and SPEI-1/SPI-1, respectively, as shown in Figure 7.

5. Discussion

The study uses the Standard Precipitation Index (SPI) and Standard Precipitation Evapotranspiration Index (SPEI) to monitor meteorological drought in the Al-Lith and Khafji basins of Saudi Arabia from 2001 to 2020 (KSA). Temperature, precipitation, evaporation, etc., are just a few of the climatic variables that are affected by climate change [45]. Other studies [46,47,48,49,50], have shown some disparities between the SPI and SPEI regarding regional drought monitoring. The fluctuations of the SPI and SPEI over each period were comparable from a time series standpoint. The two indexes varied the most frequently, and their differences were the greatest throughout the shortest period. The SPI and SPEI tended to change with time, and the differences between them narrowed, although there were still minor differences in drought severity. The SPI and SPEI revealed significantly different regional drought situations due to differences in SPI and SPEI data, as demonstrated by the study. However, it is a good idea to check that the drought identification values for SPI and SPEI are the same.

The SPI can characterize drought changes, but it ignores the effect of evaporation on drought. The SPEI, on the other hand, considers both precipitation and evapotranspiration. In the context of global warming, it is more suited for drought monitoring in arid and semiarid areas [47,50]. In this research, drought monitoring based on the SPI relied entirely on Al-Lith and Khafji precipitation as the indicator. As a result, the monitoring results may have been erroneous. The temperature in KSA is increasing, the severity of the drought is worsening, and almost 70% of the total area is under drought [33,51]. As a result of the inclusion of temperature in the SPEI, the drought reflected by the SPEI exhibited a rising trend throughout multiple timelines, as shown in Figure 6.

Drought intensification in both watersheds over the last two decades is consistent with various regional and worldwide evaluations [31,47,51,52,53]. El Kenawy et al. [47] assessed the spatial distribution of drought in Oman from 1979–2014 using SPEI and SPI at 3 and 12 months, indicating a significant increase in the drought characteristics (frequency, severity, and duration) in the region. Alsubih et al. [31] used SPI-6 to monitor meteorological drought in the Asir region of Saudi Arabia from 1970 to 2017, showing that more severe and frequent droughts occurred in the region in mentioned years. Rahman et al. [54], monitored mild to severe drought conditions over the climatic regions of Pakistan from 2000–2015 using merged Satellite precipitation products. Syed et al. [51], used six different drought indices to assess meteorological drought in KSA from 1985 to 2020 and revealed that 70% of the total area was under drought. We compared and evaluated the fluctuations of the SPI and SPEI over time and space from a variety of timelines in order to gain a deeper understanding of the performance of the two indexes as shown in Figure 5 and Figure 6. Furthermore, because the SPI and SPEI use different parameters, there will inevitably be differences between the two indices; however, these differences may be stable over time [47,48,49,50].

Based on the discussion for drought monitoring in an arid region such as KSA, precipitation and temperature are crucial factors. We recommend that SPI be used when precipitation is the only crucial factor, but SPEI may be helpful in situations when evapotranspiration is most noticeable.

6. Conclusions

Drought is one of KSA’s most severe natural disasters. Droughts may develop slowly and go unnoticed. Drought categorization is necessary for governmental and non-governmental organizations to manage drought. The current study employed SPEI and SPI to monitor meteorological drought using rain gauges (RGs) and IMERG data during 2001–2020. The meteorological drought is calculated across two different watersheds (in terms of topography and climate), i.e., Al-Lith and Khafji, of the Kingdom of Saudi Arabia (KSA) at 1-, 3-, 6- and 12-month temporal scales. The performance of IMERG is also compared with RGs using different statistical indices, including Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Correlation Coefficient (CC). The significant findings of the current research are listed below:

• Both drought indices (SPEI and SPI) could identify drought and reflect their temporal variability. SPEI recognized more severe and moderate drought events as compared to SPI. The SPEI result depicted that drought was more severe and long-lasting even though they were not immediately evident. When droughts last longer, the role of potential evapotranspiration becomes more critical. Evapotranspiration has major spatial and temporal effects on the amount of water in the soil, leading to severe and intense droughts.
• The findings revealed that the connection between SPI and SPEI is minimal at a lower time scale, but it grows stronger as the time scales of both indices increase.
• Both drought indices (SPEI and SPI) could identify drought and reflect their temporal variability. SPEI recognized more severe and moderate drought events as compared to SPI. The SPEI result depicted more severe and long-lasting droughts even though they were not immediately evident.
• The findings revealed that the connection between SPI and SPEI is minimal at a lower time scale, but it grows stronger as the time scales of both indices increase.
• The total drought periods observed in the Al-Lith watershed are 166 and 139 months, while for Khafji watersheds they are 129 and 72 months on SPEI and SPI on a multiple time scale (1, 3, 6, and 12 months).
• The moderate drought of 88 months, severe drought of 66 months, and extreme drought of 12 months were experienced in the Al-Lith watershed based on different time scales of SPEI (1, 3, 6 and 12) months.
• The moderate drought of 81 months, severe drought of 42 months, and extreme drought of 16 months were experienced in the Khafji watershed based on different time scales of SPEI (1, 3, 6 and 12) months.
• The Al-Lith watershed observed a moderate drought of 65 months, a severe drought of 34 months, and an extreme drought of 10 months according to different SPI time frames.
• The moderate drought of 42 months, severe drought of 34 months, and extreme drought of 16 months are observed in Khafji watersheds according to different SPI time frames.
• The CC values between SPEI-1/SPI-1 and SPEI-3/SPI-3 in Al-Lith Watershed are 0.86 and 0.93, respectively. While for 6 and 12 months, the correlation is strong, with CC values of 0.95 and 0.98, respectively.
• In Khafji water at 1 and 3 months, the CC values are 0.61 and 0.79, respectively. While the CC values between SPEI-6/SPI-6 and SPEI-12/SPI-12 are 0.86 and 0.94, respectively.

Overall, the Al-Lith watershed experienced more severe and extreme drought than the Khafji watershed based on different time scales (1, 3, 6, and 12 months). SPEI/SPI calculations show that its complex topography receives less precipitation and more average high temperature.