Assessing Vegetation Response to Drought in the Central Part of Oltenia Plain (Romania) Using Vegetation and Drought Indices

Abstract

Drought is an extremely negative phenomenon that is becoming increasingly frequent in the southern part of Romania (Oltenia Plain). An insufficiency or lack of precipitation, especially in the warm season, induces a state of stress on the vegetation, damaging it prematurely and decreasing the agricultural yield. Integrating satellite observations into research inventories has practical applications for drought dynamics in plain regions and may significantly contribute to its agricultural sustainability. The aim of our study was to highlight the relationship between drought and vegetation health in the central parts of the Oltenia Plain, namely, the Băilești Plain and Nedeia Plain. We used four different indices (SPI/SPI-CDF-ISND, SPEI, NDVI, NDMI) in order to assess the occurrence of meteorological and agricultural drought and gained a wider picture regarding past and future trends. The results of this study contribute to a better understanding of vegetation health index trends and their implications for climate change. The selected indices were the most suitable for assessing drought according to the literature, and combining all of them helped us to obtain a full picture of drought’s impact on vegetation.

1. Introduction

Drought is a complex phenomenon that is generally defined as an extended period (a season, a year, or even several years) characterized by below-average moisture conditions, usually affecting vast areas [1]. There are four drought types [2]. These are determined based on the variables used for characterization and the component impacted—the environment, society, and economy. Thus, the first type is a meteorological drought; precipitation deficits are considered the main driver of this phenomenon [3]. Against the background of global warming, there occurs an increase in the atmospheric evaporative demand during drought events, which intensifies their severity [4]. If meteorological drought persists, there emerges a soil moisture deficit, inducing agricultural drought [5] (increased water stress impacting both cultivated and natural vegetation). Hydrological drought is perceived as the most severe type of drought as water shortages affect both surface water bodies and groundwater [6], limiting water use. When water demand is not satisfied by water supply, drought has a negative impact, not only on the environment, but also on human activities. This type of drought is defined as socio-economic drought [7].

Drought is one of the most detrimental and costliest natural hazards, affecting large areas worldwide in recent decades. In Europe, severe and persistent droughts caused substantial damage (both environmental and socio-economic) in the last 20 years as they extended over long periods, constituting so-called multi-year droughts [8]. In Europe, the Mediterranean region is especially prone to drought, but the phenomenon may occur all over the continent [9]. The period 2018–2020, for example, was marked by a drought of extremely high intensity and persistence, lasting for more than 2 years and being considered a “record-breaking event over the past 250 years” [10]. Thus, it seems that multi-year droughts are no longer limited to Southern Europe, and severe events are affecting Central and Southeastern Europe to a greater extent as well [11].

Due to its location (Southeastern Europe) and topography (the concentric presence of the Carpathian Mountains and the large southern and eastern exposure to dry continental air mases), Romania is vulnerable to drought. During winter, large-scale atmospheric circulation is considered the key drought driver, while thermodynamic factors (temperature and humidity) prevail in summer [12]. The phenomenon is often associated with severe heat waves [13], which enhance drought’s negative impact. Studies emphasize that drought trends display an inhomogeneous spatial distribution at the country level. A statistically significant evolution towards drier conditions was identified in the same regions, namely, the southwest and southeast [14,15,16,17].

The southwestern sector of the Romanian Plain, corresponding to Oltenia Plain, is one of the most drought-affected regions in the country. Besides temperature increases and heatwaves, numerous studies indicate drought is one of main climate hazards in the region [18,19,20,21]. Angearu et al. [14] identified the Oltenia Plain as the second most affected and vulnerable region to drought in Romania after the Bărăgan Plain. Ionita et al. [16] revealed that the southwestern and eastern parts of the country register statistically significant negative dryness trends (drier conditions). Prăvălie et al. [17] emphasized that Oltenia is one of the two main epicenters of drier trends in the country. The severity of drought is also enhanced by the presence of large surfaces covered by sand and sandy soils. Part of the plain used to be called “Oltenia’s Sahara” because of the presence of mobile sand dunes, which stabilized over time. However, deforestation occurring in recent decades triggered land degradation, thus increasing socio-economic vulnerability to drought in the region [22].

Taking into account the complexity of this phenomenon and its impact on the environment and various activity sectors, drought represents an important preoccupation for sustainability researchers. The difficulty related to the assessment of drought results from the multitude of criteria/indices, used worldwide, to render both its typology and impact [23]. There are more than 100 distinct drought indices mentioned in the scientific literature, thus confirming the complexity of the phenomenon [3]. The assessment of meteorological drought is most often achieved based on the standardized precipitation index (SPI), followed by the Palmer Drought Severity Index (PDSI) and the Standardized Precipitation Evapotranspiration Index (SPEI), while in the case of agricultural drought, the Normalized Difference Vegetation Index (NDVI) is the most frequently used [24]. Consequently, the selection of appropriate indices, which depends on the type of drought and the purpose of the research, is an important aspect to be considered toward the pursuit of sustainability.

In Romania, drought was mainly assessed based on the aforementioned indices, both at country and regional levels. SPI was used by Cheval et al. [12] and Ionita et al. [16,25]. Bojariu et al. [15] applied SPEI and PDSI, and Bădăluță et al. [26] only used SPEI. Angearu et al. [14] used the Drought Severity Index, which is a standardized product based on evapotranspiration (ET) and NDVI. All these studies highlighted the southwestern, eastern, and southeastern parts of Romania as the most problematic in the country. Studies have also been performed at the regional level. Onțel and Vlăduț [18] analyzed drought’s impact on agriculture in the Oltenia Plain based on SPI, de Martonne Aridity Index, and NDVI, while Lazăr et al. [27] applied NDVI, Normalized Difference Water Index (NDWI), and Normalized Drought Index (NDDI) to the Mostiștea Plain. Minea et al. [28] analyzed meteorological (SPI, SPEI), hydrologic (streamflow drought index), and hydrogeological droughts (the standardized groundwater index) in eastern Romania (Moldova) for different timescales (1, 3, 6 and 12 months), revealing a good connection between indices and large areas affected by meteorological drought after 2007. Chelu et al. [29] reveled a slight downward trend in SPEI for southeastern Romania, while Venturi et al. [30] emphasized that SPEI offer better results in terms of drought occurrence and trend estimation than SPI for the same region.

As the study region is a prominent agricultural area, the spatial and temporal variability of drought gains in importance, especially in the present climatic context, characterized by increasing mean/extreme temperatures and evapotranspiration, conditions which enhance drought’s intensity and negative impact. The novelty of this research mainly consists of the use of four distinct indices, two for the assessment of meteorological drought (SPI/SPI-CDF-ISND, SPEI) and two for the assessment of agricultural drought (NDVI, NDMI), in order to establish drought characteristics and variability within the Băilești and Nedeia Plains, parts of the Oltenia Plain. The objectives of the present study are (1) to assess the spatiotemporal variability of meteorological and agricultural drought using specific indices (SPI/SPI-CDF-ISND, SPEI, NDVI, NDMI) at different timescales (1, 3, 6, and 12 months); (2) to analyze the trends towards drier or wetter conditions in the Băilești and Nedeia Plains; (3) and to evaluate the performance of the applied drought indices.

2. Materials and Methods

2.1. Study Area

The Băilești Plain and Nedeia Plain (Figure 1), which form together the middle part of the Oltenia Plain, cover 2172 km² and have a population of over 150,000 inhabitants. They are typical fluvial terrace plains created by the Danube [31]. Their elevation ranges from 24.61 m near the Danube Floodplain (south) to 160.54 m near Bălăcița Piedmont (north). The plains are mostly flat (1–3°), except river banks, terrace scarps, and dunes, where slopes reach 3–7°. Sunny and semi-sunny slopes cover 67% of the plains, while shaded areas account for 33%. The annual mean temperature is 11.8 °C at the Băilești meteorological station and it increases westwards, reaching 12.3 °C at Calafat. In the last 50 years, the mean annual temperature and summer temperature have increased significantly [32], thus intensifying the stress induced by the lack or insufficiency of precipitation.

The annual global radiation is 126–128 k/cal/cm² and the average sunshine duration is about 2200 h [33].

The precipitation pattern corresponds to the Bsk Köppen classification type [34]. The annual precipitation amounts are generally below 550 mm and increase from east to west and from south to north. In the western sector of the plain, there is a slightly different distribution pattern during the year, with November and December displaying greater amounts compared to the previous autumn months.

Băilești and Nedeia Plains are covered with grassland, forest–steppe vegetation, intrazonal pastures, and sands [35]. The steppe and forest–steppe conditions are only preserved in small patches (on flat surfaces and within larger depressions located on the tread of the second terrace), as agricultural crops have gradually replaced natural vegetation.

2.2. Climate Data

We used annual and monthly precipitation data, but also annual and monthly mean temperatures, for the 1981–2024 period. To compute the potential evapotranspiration index, in addition to precipitation and temperature, we used data such as solar radiation, wind speed, and relative humidity. The data were provided by the National Meteorological Administration for three meteorological stations located within the Băilești and Nedeia Plains and their vicinity (

Table 1. Meteorological stations considered.

Station	Altitude (m)	Latitude	Longitude
Băilești	58	44°1′ N	23°19′ E
Calafat	60.8	43°59′ N	22°57′ E
Bechet	36	43°47′ N	23°57′ E

).

The data provided by the National Meteorological Administration referred to monthly average wind speed, soil surface temperature, relative air humidity, monthly average temperature, monthly precipitation, and net solar radiation. PET was calculated based on these data.

2.3. Methods

2.3.1. Statistical Measures

Statistical measures were used to observe if the precipitation data had a normal distribution. The coefficient of skewness (SKEW) was applied to measure the asymmetry of the probability distribution of a real-value random variable around its mean. It is based on the notion of the moment of distribution [36]. The optimal results of SKEW, considered symmetrical and acceptable for most statistical applications, should be between −0.5 and +0.5. SKEW can be computed with Microsoft Excel. If its value is not in that range, SPI-CDF-ISND should be computed.

2.3.2. Mann–Kendall Test

The Mann–Kendall test is a non-parametric test which does not require data to be normally distributed [37]. It can be applied to detect annual and monthly trends in precipitation distribution and to determine the significance of these trends [38]. To run the test, we used the Microsoft Excel template known as MAKESENS, created by the Finnish Meteorological Institute [39]. The standard significance levels, also used in the present study, are as follows: 0.001 (**), 0.01 (**), 0.05 (*), and 0.1 (+).

2.3.3. Standardized Precipitation Index (SPI)

The standardized precipitation index (SPI) was developed by T.B. McKee et al. in 1993 [40]. It serves as a simple and effective tool for quantifying and monitoring drought by analyzing precipitation data. SPI represents the probability of rainfall occurrence over a specific period in a region, eliminating temporal and spatial differences in rainfall [22]. To compute SPI, we used two methods. Firstly, we applied McKee’s Formula (1) and then determined the cumulative distribution function (CDF) using the Gamma distribution (3). Further, for standardization, we used the Inverse Standard Normal Distribution (ISND) (2). This approach is referred to in this study as SPI-CDF-ISND [38] and was applied because the skewness coefficient indicated the need for this [41].

$$SPI={\displaystyle \frac{P-\mu }{\sigma }}$$

(1)

where P represents precipitation data, μ is the mean of precipitation, and σ is the standard deviation [42].

SPIt = Φ – 1[FR(rt)]

(2)

where SPIt represents the value of the SPI at time t; Φ − 1 is ISND; FR is the fitted CDF of precipitation, which is estimated from the precipitation records registered during the same period of each year and assumed to be time-invariant over the analysis interval; and rt is the recorded precipitation at time t for a specific timescale.

The Gamma function (3), based on which CDF, is computed according to the following formula [38]:

$$\Gamma \left(\alpha \right)={\int }_0^{\infty }{x}^{\alpha -1}{e}^{-x}dx$$

(3)

More precisely, we computed SPI-CDF-ISND in Excel as the inverse normal (Gaussian) function, with a mean of 0 and a variance of 1, applied to cumulative probability [43]. SPI-CDF-ISND involved the following steps:

• Step 1: we computed the average of the precipitation data (AVERAGE function) and then we calculated the variance based on all precipitation data (VAR.P function l).
• Step 2: we calculated α and β parameters, where α is the shape parameter of the distribution, l, determined as =(AVERAGE²)/VAR.P, and β is the scale parameter of the distribution, computed in Excel as =VAR.P/AVERAGE.
• Step 3: we computed CDF as gamma distribution using the formula: CDF = GAMMA.DIST (precipitation; α; β; TRUE).
• Step 4: we computed CDF for the specified mean and standard deviation as follows =NORM.INV (CDF; 0; 1).

The obtained values can be classified into several classes, ranging from extremely wet to extremely dry, but the drought begins when the SPI values are less than zero [40]. More accurately, drought intensity, determined according to SPI values, can be classified using the following categories (

Table 2. Categories of drought and SPI values.

Drought Category	SPI Values
mild drought	0 to −0.99
moderate drought	−1.00 to −1.49
severe drought	−1.50 to −1.99
extreme draught	−2.0 and below

):

2.4. The Standardized Precipitation–Evapotranspiration Index (SPEI)

SPEI was developed by Vicente-Serrano et al. [44] and used by many researchers to compute drought, being applied to drought variability [45], to climate change [46], in spatiotemporal drought analysis [38], and to groundwater drought [47].

SPEI is an advanced drought index that considers the effects of potential evapotranspiration on establishing drought severity. This index is suited for studies focusing on the effect of global warming on climate [48]. Relative to other drought indices, such as DeMartonne or PDSI, SPEI has the advantage of being multiscale, which is essential for drought analysis and monitoring [48]. From a mathematical perspective, SPEI is the difference between monthly rainfall and evapotranspiration (4) [44].

SPEI = P − PET

(4)

where P is rainfall data for every month and PET is the evapotranspiration or the hydric deficit, which is computed as ET₀ (5) [48], using the FAO Penman–Monteith Equation [49]:

$${ET}_0={\displaystyle \frac{0.408∆\left(R_n-G\right)+\gamma \frac{900}{T+273}{u}_2(e_s-e_a)}{∆+\gamma (1+0.34u_2)}}$$

(5)

where R_n is net radiation at the surface (MJ/m²/day); G is soil heat flux density (MJ/m²/day, often negligible); T is mean air temperature (°C); u₂ is wind speed at 2 m height (m/s); e_s is saturation vapor pressure (kPa) [50]; e_a is actual vapor pressure (kPa) [51]; Δ is the slope of the vapor pressure curve (kPa/°C) [50]; and γ the psychrometric constant (kPa/°C). Five classes are used for SPEI analysis: non-drought (>−0.5), mild (−0.5 and −1), moderate (−1.5 and −1), severe (−2 and −1.5), and extreme (<−2) [52].

2.5. Sentinel Data

We used Sentinel-2 L2A data, the NIR (band 8), RED (band 4), and SWIR (band 11) bands, with a 10 m resolution. Sentinel-2 is a network of satellites orbiting the Earth, part of the European Union Copernicus program, managed by the European Space Agency (ESA) [53]. The constellation includes two operational satellites, Sentinel-L2A (launched in 2015) and Sentinel-2B (launched in 2017), which work together to provide high-resolution multispectral images (10 m).

Sentinel-2 L2A data are pre-processed, radiometric calibration is already performed, the data are bottom-of-atmosphere (BOA) reflectance values. Also, atmospheric correction was applied using the Sen2Cor algorithm, and so atmospheric effects, like scattering and absorption, were removed [54]. We chose data without clouds and so no cloud masking was needed.

2.6. Normalized Difference Vegetation Index

NDVI is a metric and is computed as the difference between near-infrared band (NIR) and red bad (RED) reflectance, divided by their summation development [55] (6). This index uses radiances or reflectance from a red channel around 0.66 μm and from a near-IR channel around 0.86 μm [56].

NDVI = (NIR − RED)/(NIR + RED)

(6)

NDVI values range between −1 and 1. Negative values, close to −1, correspond to water, and values ranging from −0.1 to 0.1 correspond to barren areas of rock, sand, or snow. Low, positive values represent shrub and grassland (approximately 0.2 to 0.4), while high positive values indicate forests [57].

2.7. Normalized Difference Moisture Index

NDMI (7) is also known as NDWI (Normalized Difference Water Index) and is an independent vegetation index. It is complementary to NDVI, but is not a substitute for it. NDMI is sensitive to changes in the liquid water content of vegetation [56]. To compute it, NIR (0.86-μm) and SWIR (1.24-μm) bands are used (7). The SWIR band reflects changes in both the vegetation water content and the spongy mesophyll structure in vegetation canopies, while the NIR reflectance is affected by the leaf internal structure and leaf dry matter content, but not by the water content [56].

NDMI = (NIR − SWIR)/(NIR + SWIR)

(7)

The NDMI raster shows the water content of vegetation and helps to detect drought stress. Values close to +1.0 indicate high moisture content and water bodies and negative values close to −1 indicate very low vegetation moisture and areas at high risk of desertification or experiencing prolonged drought stress [56].

3. Results

3.1. SPI-CDF-ISND

The SPI-CDF-ISND index was calculated for Bechet, Calafat, and Băilești meteorological stations at four distinct timescales within the period of 1981–2024. It was necessary to preprocess the data, which involved verifying the distribution of monthly and annual precipitation. The SKEW [57,58] was calculated in Microsoft Excel and the values obtained indicated slightly abnormal distributions of the precipitation amounts. As precipitation data exhibit positive skewness (

Table 3. SWEK and values at 1 year timescale.

Station	SKEW
Băilești	0.97
Calafat	0.79
Bechet	0.74

), meaning there are many small values and few large ones, we used the CDF of the SPI to handle the asymmetry better by fitting the data to a distribution that matched its characteristics before converting it to a standardized normal distribution. This approach ensures a more accurate representation of drought conditions.

Therefore, the monthly and annual SPI values were obtained by calculating the parameters of the Gamma distribution for the cumulative probability inputs and using the inverse standard normal function. After Gamma correction, the SKEW values were zero. According to the SPI-CDF-ISND values (

Table 4. SPI-CDF-ISND annual values.

Year	Băilești	Calafat	Bechet	Year	Băilești	Calafat	Bechet
1981	0.70	0.21	0.06	2003	0.15	0.14	0.88
1982	0.29	−0.27	−0.01	2004	0.26	−0.69	0.19
1983	−1.22	−1.45	−1.88	2005	2.03	1.92	2.32
1984	−0.12	−0.45	0.16	2006	0.41	0.24	0.53
1985	−0.40	−0.07	−1.31	2007	0.74	0.44	−0.03
1986	0.49	0.61	−0.08	2008	−0.31	0.08	−0.65
1987	0.12	1.01	0.10	2009	0.61	1.17	1.05
1988	−0.91	−0.64	−0.27	2010	1.60	0.48	1.48
1989	−0.71	−1.09	−0.84	2011	−0.60	−1.77	−1.28
1990	−0.58	−1.02	−0.33	2012	−0.98	−0.57	−0.56
1991	−0.28	0.06	0.75	2013	0.27	0.02	−0.21
1992	−2.66	−2.15	−0.93	2014	3.00	2.87	2.55
1993	−1.16	−0.93	−1.30	2015	0.11	0.27	0.96
1994	−0.65	−0.72	−1.04	2016	0.64	1.41	1.30
1995	0.34	0.11	0.32	2017	−0.35	0.34	0.76
1996	0.30	0.43	−0.33	2018	0.66	0.68	1.22
1997	0.17	−0.05	0.37	2019	−0.48	0.14	−0.58
1998	0.54	0.26	0.31	2020	0.07	0.16	−0.26
1999	1.15	0.64	−0.84	2021	0.00	0.42	0.53
2000	−2.55	−2.57	−2.11	2022	0.04	−0.24	−0.28
2001	−0.80	−0.40	−0.58	2023	0.39	0.77	0.58
2002	0.69	1.26	0.43	2024	−0.93	−1.01	−1.03
≥2.0 extremely wet	1.5 to 1.99 Severely wet	1.0 to 1.49 moderately wet	−0.50 to 0.99 near normal	−0.51 to −1 mild drought	−1.0 to −1.5 moderate drought	−1.51 to −2.0 severe drought	≤−2 extreme drought

), the driest year was 2000, with values lower than −2, indicating extreme drought. At Băilești station, the SPI-CDF-ISND value for the year 2000 was −2.55, while it was −2.57 at Calafat, indicating an episode of extreme drought, similar to the one registered at Bechet, where the SPI-CDF-ISND value was −2.11. These values are supported by the reduced amounts of precipitation, namely, 271.5 mm at Băilești, 303.6 mm at Calafat, and 262.8 mm at Bechet. Another extremely dry year was 1992, with SPI-CDF-ISND values lower than −2 for Băilești and Calafat stations. Severe drought episodes were isolated. One such episode occurred in 1983, when only 363 mm was registered at Bechet, indicating an SPI-CDF-ISND of −1.88. Additionally, at Calafat in 2011, the annual precipitation amount was only 373.9 mm, indicating an SPI-CDF-ISND value of −1.77. The eastern sector of the Băilești Plain and the entirety of the Nedeia Plain are significantly affected by moderate and mild drought episodes. Thus, at Bechet station, we recorded five episodes of moderate drought (1985, 1993, 1994, 2011, 2024) and sixteen episodes of mild drought, while at Băilești station, there were only two episodes of moderate drought (1983, 1993) and thirteen episodes of mild drought. In the case of the Calafat station, there were four episodes of moderate drought (1983, 1989, 1990, 2014) and eleven episodes of mild drought. SPI-CDF-ISND values greater than 2 indicate extremely rainy periods, such as in 2005 and 2014, when excess precipitation resulted in floods.

The SPI and SPI-CDF-ISND indices were calculated at 1-month, 3-month, and 6-month timescales. The results indicate extreme values, differing from the annual ones, as well as the existence of differences between the two indices (Figure 2, Figure 3 and Figure 4).

The monthly values of SPI-CDF-ISND are strongly dependent on monthly precipitation amounts; thus, if the total monthly precipitation amount is low, SPI-CDF-ISND shows negative values, indicating a certain degree of drought. For Băilești station, a SKEW value of −0.49 was obtained for the SPI-CDF-ISND at a 1-month timescale, indicating left skewness. Hence, the distribution of SPI-CDF-ISND scores tends to have more extreme negative values than extreme positive values, although the difference is not significant. The frequency of monthly extreme drought phenomena (SPI-CDF-ISND less than −2) was 5.8% and the frequency of severe drought episodes was 4.7%. At the Calafat meteorological station, the distribution of monthly SPI-CDF-ISND values did not follow a specific pattern, but was highly precipitation-dependent. The SKEW value for the data series of SPI-CDF-ISND at 1 month timescale was −0.6, highlighting left skewness as in the case of Băilești station. The frequency of extreme monthly drought episodes was 6.06% in the period of 1981–2024. Measuring the asymmetry of 1 consecutive month of SPI-CDF-ISND values for the Bechet meteorological station, we obtained a value of −0.37, indicating negative skewness in the data distribution. This means that the precipitation distribution slightly shifted towards values lower than the mean, suggesting there were more sub-average precipitation events than those above the multiannual average of the monthly precipitation amount, which was 42.6 mm. The frequency of monthly extreme drought cases was only 3.7%, indicating a normal precipitation distribution for the southeastern part of the study area.

The analysis of SPI-CDF-ISND values over a 3-month period showed minimum values approaching −4, indicating extreme drought conditions. The skewness values for SPI-CDF-ISND at a 3-month timescale were −0.38 for Băilești station, −0.41 for Calafat station, and 0.11 for Bechet station. Between 1981 and 2024, we recorded 247 cases with negative SPI-CDF-ISND values at the Băilești station.

The frequency of extreme drought episodes at Calafat, in the western part of Băilești Plain, at a 6-month timescale was 3.4% during the period of 1981–2024, while the frequency of severe drought episodes was 4.5%. Episodes with negative SPI-CDF-ISND values at a 6-month timescale were present in 51.2% of the cases. The driest month in the analyzed period was March 2002, with an SPI-CDF-ISND value of −4.07, followed by December 1992 (SPI-CDF-ISND: −3.06) and November 1992 (SPI-CDF-ISND: −2.73). Recently, November 2024 had the lowest 6-month SPI-CDF-ISND value (−1.58) since 2012, when the SPI-CDF-ISND value for November was −2.36. At Băilești station, the frequency of cumulative 6-month extreme drought episodes was 2.48% during the period of 1981–2024, while the frequency of severe drought episodes was 5.35%. Overall, the frequency of cases with negative SPI-CDF-ISND values scale was 46.2%. The driest months, with cumulative precipitation deficits over 6 months, were March 2002 (−3.82), December 1992 (−3.41), January 1993 (−3.02), April 2002 (−2.80), and November 2012 (−2.58). The analysis of extreme and severe drought episodes at a 6-month timescale shows that these do not have a continuous distribution but alternate with episodes of normality or extreme wetness.

In the southeastern part of the Băilești Plain and in the Nedeia Plain, near the Bechet meteorological station, the frequency of extreme drought episodes at a 6-month timescale was 1.72%, while the frequency of severe drought episodes was 4.39%. During the period of 1981–2024, the frequency of cases with negative SPI-CDF-ISND values was 51.2%. The distribution of SPI-CDF-ISND values at a 6-month timescale does not have a convergent trend towards 2024, but is scattered. The driest months were March 2002 (−2.79), January 1993 (−2.51), and February 2002 (−2.24).

The Mann–Kendall test and Sen’s slope test were applied to the annual and monthly precipitation data of the three meteorological stations. Z values display positive trends (

Table 5. Mann–Kendall and Sen’s slope.

Mann–Kendall Trend (First Year: 1981; Last Year: 2024, n = 44)				Sen’s Slope Estimate
	Time Series	Test Z	p-Value	Signific.	Q
BĂILEȘTI	YEAR	1	0.32		1.43
	January	1.4	0.16		0.44
	February	−0.31	1.24		−0.05
	March	0.6	0.55		0.24
	April	−1.17	1.76		−0.4
	May	1.7	0.09	+	0.62
	June	0.66	0.51		0.27
	July	0.62	0.54		0.32
	August	0.08	0.94		0.01
	September	0.93	0.35		0.27
	October	1.69	0.09	+	0.64
	November	0.96	0.34		0.32
	December	−0.29	1.23		−0.15
CALAFAT	YEAR	2.13	0.03	*	2.59
	January	1.69	0.09	+	0.56
	February	0.2	0.84		0.05
	March	0.9	0.37		0.26
	April	−1.2	1.77		−0.28
	May	1.71	0.09	+	0.59
	June	1.65	0.10	+	0.63
	July	0.48	0.63		0.14
	August	−0.09	1.07		−0.05
	September	−0.06	1.05		−0.02
	October	1.43	0.15		0.7
	November	1.17	0.24		0.44
	December	−0.02	1.02		−0.02
BECHET	YEAR	2.03	0.04	*	2.53
	January	0.85	0.40		0.26
	February	−0.53	1.40		−0.12
	March	0.89	0.37		0.27
	April	−0.86	1.61		−0.22
	May	1.33	0.18		0.52
	June	2.23	0.03	*	0.97
	July	−0.29	1.23		−0.1
	August	−0.5	1.38		−0.1
	September	1.59	0.11		0.38
	October	1.99	0.05	*	0.82
	November	0.45	0.65		0.16
	December	−0.74	1.54		−0.23

“+” indicates an increasing trend (positive) in the data. “-” indicates a decreasing trend (negative) in the data. “*” signifies that the trend is statistically significant at a certain confidence level (e.g., 0.05 or 0.01).

) for the annual amount of precipitation, and both negative and positive trends for monthly precipitations. For Băilești station, the annual trend evaluated by the Mann–Kendall test indicates a slight positive trend (Z = 1) with a p-value of 0.32 and an estimated Sen’s slope value of 1.43. The specific monthly trend indicates the following: the months of May and October show statistically significant positive trends with p-values below 0.10 (0.09 for both months) and Sen’s slope values of 0.62 and 0.64, respectively. In the case of Calafat station, the annual trend evaluated by the Mann–Kendall test indicates a significant positive trend (Z = 2.13) with a p-value of 0.03 and an estimated Sen’s slope value of 2.59. The specific monthly trend shows that the months of January, May, and June present positive significance with p-values below 0.10 (0.09 and 0.10, respectively) and Sen’s slope values of 0.56, 0.59, and 0.63.

At Bechet station, the annual trend evaluated by the Mann–Kendall test indicates a significant positive trend (Z = 2.03) with a p-value of 0.04 and an estimated Sen’s slope value of 2.53. The specific monthly trend shows that the months of June and October present statistically significant positive trends, with p-values of 0.03 and 0.05 and Sen’s slope values of 0.97 and 0.82.

Overall, the data in the table show a positive trend in annual precipitation amounts and for most months at all three meteorological stations analyzed. May shows a statistically significant increase in precipitation in the western and central sectors of the study area (0.1 level of significance in both cases), while the increase registered in June is only statistically significant in the eastern and western parts of the plain at Bechet and Calafat (levels of significance of 0.05 and 0.1, respectively). October also displays statistically significant positive trends at Băilești and Bechet (levels of significance of 0.1 and 0.05, respectively). Winter and early spring months do not show statistically significant trends, except for in January, in the western sector of the plain. The number of months characterized by negative trends (none statistically significant) increases from west to east and from north to south (3 months at Băilești, 4 months at Calafat, and 5 months at Bechet).

3.2. Standardized Precipitation Evapotranspiration Index

The SPEI was only calculated for the interval of 2005–2024 due to the unavailability of the necessary data for a longer-term calculation. The SPEI values indicate an increase in drought episodes in recent years, culminating in the year 2024, which was the driest of all analyzed years at all three stations. This can be correlated with climate change and the high temperatures mentioned in the analysis. Additionally, 2014 was an exceptionally wet year, highlighting the extreme variability of weather conditions in Băilești Plain.

The distribution of SPEI values shows great variability over the years, alternating between wet and dry periods, but with a recent trend towards severe drought conditions. According to the SPEI values calculated at the 1-year timescale (

Table 6. SPEI values at 1 year timescale.

Year	Bechet	Băilești	Calafat	Year	Bechet	Băilești	Calafat
2005	1.88	1.69	1.59	2015	0.47	−0.22	−0.14
2006	0.07	0.09	−0.18	2016	0.81	0.27	1.01
2007	−0.46	0.36	−0.05	2017	0.26	−0.62	−0.12
2008	−0.93	−0.53	−0.34	2018	0.74	0.27	0.26
2009	0.55	0.26	0.75	2019	−0.83	−0.71	−0.27
2010	1.00	1.26	0.08	2020	−0.60	−0.27	−0.26
2011	−1.32	−0.75	−1.73	2021	0.07	−0.34	−0.03
2012	−0.85	−1.05	−0.90	2022	−0.61	−0.30	−0.61
2013	−0.55	−0.06	−0.36	2023	0.10	−0.33	0.32
2014	2.15	2.88	2.79	2024	−1.94	−1.91	−1.81
≥2.0 extremely wet	1.5 to 1.99 Severely wet	1.0 to 1.49 moderately wet	−0.50 to 0.99 near normal	−0.51 to −1 mild drought	−1.0 to −1.5 moderate drought	−1.51 to −2.0 severe drought

), at Bechet station, the SPEI value in 2024 was −1.94. In the case of Calafat station, the SPEI value was −1.81, while the recorded value at Băilești was −1.91. In all three cases, drought was classified as severe. The year 2011 was also marked by drought events, but severe drought (SPEI −1.73) only emerged in the western part of the plain at Calafat. During the period analyzed, moderate drought affected the eastern sector of the plain in 2011 and its central sector in 2012. However, during the analyzed interval, analysts recorded the two rainiest years seen in the last 6 decades, 2014 and 2005. Consequently, 2014 is classified as an extremely wet year within the study area (SPEI value of 2.79 at Băilești, 2.88 at Calafat, and 2.15 at Bechet), while 2005 was a severely wet year, with SPEI values ranging between 1.59 and 1.88.

The temporal distribution of SPEI at a 3-month timescale shows (Figure 5) decreasing trends at all three stations. The distribution of SPEI values does not follow the logic of years considered drought-stricken, but is rather random. The 3-month SPEI does not indicate episodes of extreme drought, but 10 episodes of severe drought were identified at Băilești station, 5 episodes were identified at Calafat station, and 6 episodes were identified at Bechet station. The lowest 3-month SPEI value at Băilești was −1.71, recorded in October 2019, which was similar to that of Bechet (1.67), while the highest values—3.78 and 4.01, respectively—corresponded to September 2005. We mention that 4.01 was the greatest value within the entire plane for the analyzed interval. The analysis of 3-month SPEI values at Băilești station shows a distribution of negative values, especially in the spring and winter months, with significant variation over the analyzed period. For the Calafat station, the statistical analysis emphasized that the SPEI-3 values from 2005 to 2014 varied significantly, indicating either extreme drought or extreme humidity. During this interval, the SPEI-3 values fluctuated from a maximum of 2.65 in September 2005 to a minimum of −1.81 in October 2018. These data suggest considerable climatic variability in the Băilești Plain.

The coefficient of determination, R2, displays reduced values for all three stations and does not suggest a clear or significant trend in the evolution of SPEI over time, but the values probably fluctuate randomly under the influence of seasonal and cyclical factors.

The analysis of the 6-month SPEI (Figure 6) for Băilești station reveals 127 cases with negative values, indicating a precipitation deficit across a total of 234 observations. The lowest SPEI value was −2.11 (November 2012). This indicated the only month included in the extreme drought category. This low value was due to an accumulation of only 95.9 mm of precipitation and average temperatures of 19.5° C for the June–November 2012 period. The number of values between −1.51 and −2, associated with episodes of severe drought, is 7. Additionally, 21 episodes of moderate drought and 75 cases of mild drought (SPEI: −0.51 and −1) were identified. In 105 cases, SPEI ranged between −0.50 and 0.99 (almost normal). The 6-month SPEI values indicate 13 moderately wet and 6 severely wet cases. In 13 cases, SPEI values indicated an extremely wet period. The highest value was 3.84 (September 2014).

The 6-month SPEI analysis for Calafat station shows 122 cases with negative values out of a total of 234 cases. The lowest SPEI values were −2.18 (November 2012) and −2.05 (December 2011), with the respective months being included in the extreme drought category. There were 11 cases of severe drought, and 20 cases were classified as displaying moderate drought. The 6-month SPEI values indicate 39 mild drought cases and 129 almost-normal cases. The 6-month SPEI values indicate 16 moderately wet episodes, 6 severely wet episodes, and 11 extremely wet periods. The highest 6-month SPEI value was 3.72 (cumulative deficit for September 2014). At Bechet, there were 124 cases with negative values out of 234. There were no cases of extreme drought identified, but there were 13 episodes of severe drought, 19 cases of moderate drought, 51 cases of mild drought, and 120 almost-normal cases. There were 15 moderately wet episodes, 6 severely wet episodes, and 10 extremely wet episodes. The maximum SPEI value was 3.40, observed in February 2007.

Unlike SPI-CDF-ISND, which shows isolated drought episodes as it only quantifies precipitation values, SPEI is a more complex tool due to the variety of parameters it considers to determine the drought level of a period. Both SPI-CDF-ISND and SPEI were designed to quantify the precipitation deficit for multiple timescales, and as the timescale trends from 1 year to a month, the drought increases. This is an early warning for stakeholders to take action.

3.3. Normalized Difference Vegetation Index—Time Series Analysis

NDVI rasters (Figure 7) were generated based on Sentinel 2 images for each August between 2017 and 2024. The decision to analyze NDVI for August was taken to understand if there was a cumulative precipitation deficit in the spring and summer months when the vegetation characteristic of the latitude of approximately 45° N was still in the growing season. Despite NDVI minor limitations, this index has significant potential for assessing drought.

In 2017, NDVI values ranged between −0.99 and 0.99, with the pixel histogram (Figure 8) indicating the predominance of negative values, showing an unhealthy vegetation cover, but a climate regime close to normal, with a slight trend toward drought. In 2018, NDVI values ranged between −0.46 and 0.92, with a predominance of positive pixels. This indicated the presence of healthy vegetation cover, supported by a normal, even slightly excessive, rainfall regime distributed throughout the year.

In 2019, NDVI values for August ranged between −0.30 and 0.86, with the pixel histogram indicating the predominance of non-vegetative cover, although August, according to the 3-month SPEI, did not show a drought risk. However, in the spring months, SPEI values dropped to −1.14 in April at Băilești station and to −1.05 at Calafat station, indicating moderate drought.

In 2020, NDVI ranged between −0.82 and 0.92, indicating normal vegetation development, supported by a generous amount of precipitation in May (116 mm at Băilești). The pixel histogram for 2021 indicates the predominance of values between 0 and −1, indicating vegetation cover was affected by the lack of precipitation in the warm season. For August, the 3-month SPEI was −1.22 at Băilești, while at Calafat it was −1.18. Additionally, in the northwestern part of Băilești Plain, there is an expansion of dry vegetation, which was much healthier in 2018.

In August 2022, NDVI values ranged between −1 and 0.92, with drought affecting vegetation more acutely in the eastern and western sectors of the plain. In 2023, the NDVI for August had values similar to those registered in August 2022 but, being a wetter year, the vegetation remained green for a longer period. There were also positive SPEI values, which were close to normal. The NDVI calculated for August 2024 registered values ranging between −1 and 0.96, with the predominance of negative pixels indicating dryness and drought.

The spatial distribution of NDVI values in 2024 indicates an increase in the drought condition in the northwestern part of the Băilești Plain, but also in the eastern half. According to meteorological measurements, 2024 was the warmest year of all time; thus, SPEI values at a 3-month scale indicated there was moderate drought for February, March, and April at all three stations.

The persistence of drought during the vegetation growing period influenced the normal evolution, slowing development and favoring rapid drying. The situation is explained by the amount of precipitation that fell at the end of winter and the beginning of spring. In February 2024, only 8.7 mm fell at Băilești and 9.7 mm fell at Bechet and Calafat. In March and April, the cumulative amount of precipitation for the two months did not exceed 70 mm at any station. From May to July, plants and soil could not replenish their water needs because the cumulative deficit at 3 months was still high, although the rainfall regime approached normalcy in May (79.5 mm at Băilești, 86.7 at Bechet, and 91.2 mm at Calafat).

August 2024 was moderately dry at all 3 stations analyzed, with a slight increase in drought towards the southeast of the plain, at the Bechet station, where SPEI-3 had a value of −1.43. This was slightly lower than the level seen at Băilești and Calafat. September 2024 brought the phenomenon of severe drought at Calafat and Bechet, with SPEI-3 being slightly lower than −1.5. According to SPEI-3, October was severely dry in the central and eastern sectors of the plain at the Băilești and Bechet stations. The analysis of SPEI-6 values for all months of the year 2024 indicates slightly different distributions, with May, October and November being the driest months, due to the cumulative deficit over 6 months.

3.4. Normalized Difference Moisture Index—Time Series Analysis

The NDMI analysis, based on Sentinel images with a resolution of 10 m, shows fluctuations in vegetation and soil moisture content between 2017 and 2024. According to the spatial distribution of this index, the year in which vegetation and soil had sufficient water content was 2018.

The distribution of NDMI values for 2017 (Figure 9) shows an almost-normal situation in terms of of soil and plant water content, but the corresponding histogram suggests the predominance of pixels near zero, indicating moderate moisture, with vegetation undergoing a drying process. At the end of August 2019, NDMI values ranged between −0.65 and 0.68, with a predominance of values close to zero and being negative.

In August 2020, NDMI values were predominantly negative, with dry vegetation, arid soils, and some watercourses in the plain drying up. August 2021 showed NDMI values between −1 and 0.71, similar to those of August 2022, but the negative values covered larger surfaces in August 2022, indicating a more pronounced lack of water in the vegetation and soil. In August 2023, the pixel histogram (Figure 10) indicated an increase in positive pixels, suggesting a relatively normal amount of water in plants and soil. NDMI values for 2024 ranged between −1 and 0.96, but a large part of Băilești Plain and Nedeia Plain experienced a lack of water in plants and soil.

4. Discussion

Drought is a natural hazard referring to a prolonged dry period in the natural climate cycle that can occur anywhere in the world, but is also a complex natural disaster that causes serious environmental, social, and economic consequences worldwide. Southeastern Europe is currently affected by an increase in temperature, which is also associated with the increase in evapotranspiration [59] and a decrease in the precipitation amount [60]. Consequently, drier conditions [61] and increasing aridity [59] were identified in the region. In Romania, there was a significant shift towards drier conditions, mainly in the eastern and southern parts of the country [16,17,62]. A temperature increase, which is statistically significant both at annual and monthly levels, is also accompanied by a significant increase in the potential evapotranspiration in southern Romania [63]. Drought, the severity of which is enhanced by its association with severe heat waves [13], especially between 2001 and 2020, became a common occurrence in the aforementioned areas [14,16].

Drought assessment is based on a wide variety of specific indices. The presence and severity of meteorological and agricultural drought in the study region were highlighted by the application of some of the most employed indices worldwide. Drought was analyzed in different studies, the results of which indicated that it is one of the main climate hazards impacting the Băilești Plain and Nedeia Plain [14,18,21]. The analysis of climate data from the last 44 years and the applied drought indices shows the alternation of dry, wet, and normal years, with 1983, 1992, 2000, 2011, 2012, and 2024 emerging as the years most affected by severe and extreme drought episodes.

The SPI-CDF-ISND and SPEI for different timescales show results that are consistent with the results of previous studies. According to the SPI-CDF-ISND values obtained, the eastern sector of Băilești Plain and all of Nedeia Plain are affected by moderate and mild drought episodes. However, extreme drought episodes were registered corresponding to 2000–2002 (SPI-CDF-ISND value less than ≤−2 for 1-, 3-, and 6-month timescales), the respective period also being identified as a major drought event, in terms of severity and spatial extent, by Spinoni et al. [64] and Ionita et al. [16]. The Băilești Plain (central and western sectors) was also affected by extreme drought in 1992 (SPI-CDF-ISND12 ≤−2), which is consistent with the results obtained by Onțel and Vlăduț [18]. After 2000, there one severe drought episode was registered in 2011, but only in the western part of the plain.

The SPEI values indicate an increase in drought episodes in recent years, culminating with 2024, which was the driest of all years analyzed (2005–2024) at all three stations. Even though decreasing SPEI12 trends emerged, they are not statistically significant, with the same results being obtained for southeastern Romania as well [29]. Venturini et al. [30] mentions a significant negative monotonic trend for the SPEI12 for southeastern Romania, but the results are based on the analysis of climate date for only one station. For shorter timescales (1-month, 3-month), Jaagus et al. [65] indicate Romania as an area undergoing a clear decrease in SPEI, reflecting a drying trend, for the summer and summer months (evaluated period 1949–2018), even if most of the trends are not statistically significant. At a short timescale (SPEI3), there is a higher tendency for fluctuations to occur between dry and wet periods compared to a long timescale (SPEI12). This trend is also mentioned for eastern Romania [28].

Indices, calculated based on remote sensing data, were implemented in the study because they provide a comprehensive overview of the entire area, not just point-based data, as with climatic indices. Vegetation indices are typically selected to represent either vegetative greenness, such as NDVI or EVI (Enhanced Vegetation Index), or water sensitivity, such as NDMI or NDWI [66,67]. The potential of Sentinel-2 in assessing drought through vegetation characteristics; soil moisture; evapotranspiration; surface water, including wetland; and land use and land cover analysis was also used in other studies [68,69]. Similar to Varghese et al. [70] and El-Hokayem et al. [66], we found that the precision and efficiency of Sentinel-2-derived drought factors using direct and fusion methods can be considered a limitation of the study.

NDVI and NDMI show fluctuations in vegetation and soil moisture content during the period analyzed. Certain years emerged as being particularly marked by drought episodes (predominance of negative pixels) according to both indices applied—2022, 2022, and 2024—while in other years, only one of the indices emphasized drought-affected vegetation. This is the case for 2020, when NDMI indicates normal vegetation development and NDVI dry vegetation and arid soils. Prior to the period analyzed, Onțel and Vlăduț [18] identified 1985, 1993, 2000, and 2007 as years marked by agricultural drought in the region. The most severe episode corresponded to 2000, when NDVI values were between −0.2 and 0.1, indicating the presence of barren lands (sands) and grassland vegetation affected by severe drought.

The spatial distribution of NDVI and NDMI values indicates that the northwestern and eastern part of Băilești Plain and Nedeia Plain are more affected by drought compared to the other sectors of the relief unit analyzed. The persistence of drought during the vegetation growing period is due to the small amount of precipitation that fell at the end of winter and the beginning of spring, thus influencing the normal development of vegetation and favoring rapid drying. Further analysis of Oltenia Plain, focusing on the phenology of crops using Sentinel-2 data, would be useful in distinguishing characteristic vegetation changes.

5. Conclusions

All indices computed can be used individually to assess drought in a region. We wanted to combine them and identify the relationship between them, but because there were not enough meteorological stations in the studied area, it was not possible to perform an interpolation of the SPI-CDF-ISND or SPEI data to obtain raster datasets that could be statistically analyzed in correlation to NDVI or NDMI.

As we conducted our study further and read the literature, we observed that there may be some different approaches to the SPI calculation method and, after applying skewness coefficient, based on the literature suggestions, we considered that it is most suitable to apply no. 2 and no. 3 formulas and obtain SPI-CDF-ISND.

This study demonstrates that the annually calculated SPI-CDF-ISND values for Băilești, Calafat, and Bechet stations indicated episodes of extreme drought since the 1990s, but these drought episodes do not have a continuous and convergent trend towards the present. They occur sporadically, alternating with normal periods and extremely wet periods.

In drought phenomenon analysis, the most suitable indicator is SPEI, at different timescales, but also the NDVI and NDMI indices if the satellite images have good resolution. We conclude that specific bands (8, 4 and 11) in Sentinel-2 images can reliably detect changes in the vegetation index that highlight drought dynamics within the Băilești and Nedeia Plains.

The SPEI is a multiscale drought index based on climatic data. It has been used in many studies, being computed from a regional to a global level. It can be used to determine the start, duration, and magnitude of drought conditions with respect to normal conditions in a variety of natural and managed systems, such as crops, ecosystems, rivers, water resources, etc.

We calculated the NDVI and NDMI values based on Sentinel images and made observations on vegetation health and water content the end of the summer within Southwestern Romania. The working methods have highlighted the fact that the red-edge band is the most important band of the Sentinel-2, with high precision in observing the vegetation response to drought compared to conventional methods such as NDVI.

For our study area, we can conclude that drought is a hazard that occurs from time to time, and the indices we computed can show the presence of drought. However, it is difficult to consider it a trend due to the distribution of indices in time.