Frequency, Spatial Distribution, and Influence of Consecutive Dry Days on Rainfed Agriculture

Abstract

Given the climate variability of semi-arid regions, this study analysed rainfall regimes and their influence on consecutive dry days (CDDs) and cowpea (Vigna unguiculata) productivity in Ceará, Brazil. Rainfall data from 184 municipalities (1990–2019) and productivity records were used across eight homogeneous rainfall regions. Water scenarios (very dry, dry, normal, rainy, and very rainy) were defined using quantiles, and three CDD classes were considered: CDD1 (5–10 days), CDD2 (11–15 days), and CDD3 (>15 days). Statistical analyses were performed with the Kruskal–Wallis test and Spearman’s correlation, and spatial patterns were mapped with ordinary kriging. Ceará’s climate normal was 837 mm, with the Central Sertão and Inhamuns and Jaguaribana showing the lowest rainfall. A total of 39,382 CDD events were identified, with 67% as CDD1, 16% as CDD2, and 17% as CDD3. Cariri had the highest CDD1 occurrences, while Central Sertão and Inhamuns recorded the highest CDD3. Cowpea yield averaged 286 ± 85 kg ha⁻¹, with the lowest productivity in Central Sertão and Inhamuns due to reduced rainfall and frequent CDD3. Productivity correlated positively with CDD1 in one very dry scenario and negatively with CDD3 in very dry, dry, and normal conditions. The findings highlight regional vulnerabilities and the strong link between CDD and crop yield.

1. Introduction

Rainfed agriculture plays a crucial role in food production [1], corresponding to 60% of all cultivated areas and accounting for 80% of the food produced [2]. This type of agriculture is closely linked to climate conditions and faces significant challenges due to the variability and unpredictability of rainfall events [3,4,5].

This uncertainty is more obvious in arid and semi-arid regions and is related not only to the total rainfall but also to its spatial and temporal distribution [6,7,8,9]. In these regions, very dry and very rainy years often alternate with normal years [10,11,12,13,14].

Among the dry regions of the world, the semi-arid region of Brazil (SAB), which covers an area of 1,335,298 km², is the most populated in the world, with 28 million inhabitants [15,16]. Despite the uncertainties of the rainfall regime, rainfed agriculture in the SAB is found in 89% of all rural establishments, where the main crop is the cowpea (Vigna unguiculata), with a production area of 743,443 hectares [17,18]. The cowpea (Vigna unguiculata) is adapted to various types of soil and has a short cycle, together with a water requirement of between 300 and 450 mm cycle⁻¹ [19,20]. Although the water demand of this crop is low, it is sensitive to long periods of consecutive dry days (CDDs), so it is essential to understand and monitor the distribution of rainfall events to ensure satisfactory yields [21].

The impact of consecutive dry days depends on such factors as the duration of the phenomenon and the phenological stage of the crop [22,23,24,25], as seen by Cavalcante et al. [26], who observed changes in the rate of photosynthesis and in stomatal functions during CDD. Furthermore, in the current scenario of climate change, the trend is towards changes in rainfall patterns [27,28,29].

Researchers have seen an increase in extreme events, such as droughts and floods, as well as an increase in the number of consecutive dry days [27,28,30,31], which changes depending on the following regional characteristics: location, climate, and season [4,5,32,33,34].

There have been several studies on rainfall regimes and their effect on rainfed agriculture [33,35,36]. Fernandes et al. [22] found a significant reduction in the physiological functioning of the cowpea subjected to 10 consecutive dry days, whereas 2 mm of rainfall had a positive effect on the metabolic activity of the crop. Rocha et al. [24] employed dry, normal, and rainy scenarios to correlate agricultural losses with the number of CDDs. Similarly, Nogueira et al. [8] verified the occurrence of consecutive dry days under dry, normal, and rainy regimes, with a reduction of 69% in agricultural production subjected to CDDs > 15 days.

While earlier research addresses dry, normal, and rainy scenarios and their impact on agriculture, the semi-arid region goes beyond these scenarios, being characterised by its susceptibility to extreme conditions, such as severely dry and excessively rainy years. Furthermore, existing studies do not detail the exact number of CDDs in each water scenario, which is crucial for understanding the pattern of dry periods during the rainy season and their impact on rainfed agriculture. We, therefore, assumed that the amount and duration of consecutive dry days affect cowpea production under different rainfall regimes.

This study aimed to understand climate and production patterns in different water scenarios in a semi-arid region and to correlate the variables to determine the influence of each element. The specific objectives were (i) to evaluate the spatial and temporal distribution of the rainfall, (ii) to quantify and evaluate the spatial distribution of consecutive dry days in different water scenarios, (iii) to quantify and evaluate the spatial distribution of cowpea productivity under different rainfall regimes and (iv) to verify the correlation between the variables that characterise rainfall regimes and cowpea productivity.

2. Materials and Methods

2.1. Study Area

The state of Ceará is located in the northeast of Brazil and covers an area of 148,894 km² with approximately 8.8 million inhabitants [37]. The region has a unimodal rainfall regime, with 60% of the rainfall concentrated from February to April, a result of the Intertropical Convergence Zone (ITCZ), which is the main meteorological system responsible for precipitation [38]. Potential evapotranspiration ranges from 1243 to 2212 mm yr⁻¹ [39]. The predominant climate in the region is type BSh’, dry semi-arid, characterised by high evaporation, high average temperatures, and low rainfall [40].

The aridity index in 76.16% of the area of Ceará ranges from 0.21 to 0.50, classifying it as semi-arid; the extensive area includes dry sub-humid and humid sub-humid climates [39]. Given the different climate types, Xavier [41] adopted the quantile technique, subdividing Ceará into eight homogeneous rainfall regions (HRRs) with similar rainfall patterns. These zones are the Northern Coastal Region, the Coastal Region of Pecém, the Coastal Region of Fortaleza, Ibiapaba, the Baturité Massif, Jaguaribana, the Central Sertão and Inhamuns, and Cariri (Figure 1), regions monitored by the Ceará Foundation for Meteorology and Water Resources [42].

The study area exhibits a wide spatial variability of soil classes (Figure 2) [43], with emphasis on Planosols, Acrisols, Luvisols, and Leptosols, according to the World Reference Base (WRB) classes [44], which cover 22%, 19%, 18%, and 14% of the territorial area, respectively.

In soil management, rural properties are divided into two categories: those without soil tillage and those with soil tillage. In Ceará, 44% of agricultural properties do not perform soil tillage, while 27% adopt conventional tillage, 26% use minimum tillage, and 3% employ no-tillage systems,

Table 1. Tipos de sistemas de preparo do solo, por região, no Ceará.

Region	No-Tillage		Conventional Tillage
	Number of Farms	%	Number of Farms	%
R1—Cariri	20,607	2	15,294	3
R2—Ibiapaba	31,491	7	8287	4
R3—Jaguaribana	14,097	8	15,561	5
R4—Coastal Region of Fortaleza	12,704	5	4576	9
R5—Coastal Region of Pecém	18,824	3	3227	7
R6—Northern Coastal Region	28,000	3	6381	3
R7—Baturité Massif	10,035	5	7847	2
R8—Central Sertão and Inhamuns	39,751	10	47,964	21
Ceará	175,509	44	109,137	27
Region	Minimum Tillage		No-Tillage System
	Number of Farms	%	Number of Farms	%
R1—Cariri	18,649	2	1560	0.1
R2—Ibiapaba	8153	2	3297	0.8
R3—Jaguaribana	13,960	2	770	0.8
R4—Coastal Region of Fortaleza	5208	4	1059	0.4
R5—Coastal Region of Pecém	4654	4	1235	0.2
R6—Northern Coastal Region	6476	1	3149	0.3
R7—Baturité Massif	6677	1	250	0.3
R8—Central Sertão and Inhamuns	41,272	12	1811	0.4
Ceará	105,049	26	13,131	3

[17].

2.2. Hydrological Regime

2.2.1. Rainfall

Daily rainfall data (1990–2019) were obtained from the Foundation for Meteorology and Water Resources of Ceará [42] and included 184 rainfall stations (Figure 1), one in each municipality, where the criterion for using the information was the consistency of the data. The process involved analysing the historical series from 1990 to 2019 for each rainfall station. Whenever daily data were missing, the corresponding year was completely excluded.

Of the 184 stations, 88% had 30 years of data; 9% had 25 to 30 years of data; and 3% had 25 to 20 years of data. The data were homogeneous at a level of 95% [8,45].

2.2.2. Consecutive Dry Days (CDDs)

The number of consecutive dry days was calculated using the daily rainfall data, defining the CDD as a period of at least five consecutive dry days with less than 2 mm of rainfall [22], considering that rainfall values above 2 mm positively influence the physiological activities of cowpea cultivation [35]. The consecutive dry days were classified based on duration: CDD1 (5–10 days), CDD2 (11–15 days), and CDD3 (>15 days). The evaluation period lasted from January to May.

January was included in the evaluation, as this is when seed is distributed to farmers by the Rural Assistance and Extension Company of Ceará (EMATERCE), Ceará, Brazil. January in Ceará is characterised by increasing rainfall and consecutive wet days [27], with farmers planting their crops immediately after the first rains, with January to May being the period of rainfed farming in the region [46].

Finally, the CDDs were quantified using a routine written in the Python version 3.11 programming language. The daily rainfall data were then input into the program for the algorithm to count and classify the number of consecutive dry days. The script is shown in Appendix A.

2.2.3. Water Scenarios

The water scenarios were defined using Pinkayan’s Quantile Technique. The method is based on arranging the annual rainfall in ascending order and calculating the quartile position to categorise the years into their respective scenarios [41,47]. Years with the lowest 15% of the total annual rainfall were categorised as ‘very dry’, years with between 15% and 35% were classified as ‘dry’, years with between 35% and 65% were classified as ‘normal’, years with from 65% to 85% were classified as ‘rainy’, and years above 85% were classified as ‘very rainy’ [48].

2.3. Cowpea (Vigna unguiculata) Yield

Yield data (kg ha⁻¹) on the cowpea (Vigna unguiculata) between 1990 and 2019 were obtained from the IBGE Automatic Recovery System (SIDRA) [49]. The system provides statistical information on municipal agricultural production in Brazil, including data on annual and perennial crops relating to production, yield, planted area, harvested area, and production value.

Cowpea cultivation in Ceará follows a uniform pattern among farmers, with planting carried out shortly after the first rains of the year [46]. The seeds used belong to the same cultivar, being provided by the government through the Hora de Plantar program and distributed simultaneously throughout the state by the Empresa de Assistência Técnica e Extensão Rural do Ceará [50].

Additionally, to complement this information, the land use and land cover dataset from the MAPBIOMAS project for the years 1990 and 2019 was used, aiming to visualise the spatial arrangement of cultivated areas in the state of Ceará [51].

2.4. Interpolation of the Data

Ordinary kriging was applied to generate the maps of the hydrologic regime and cowpea yield in the Quantum GIS (QGIS) 3.10 software using the Processing R Provider plugin [52]. It should be noted that a semivariogram was calculated for each variable to select the best-fitted model [53], which was chosen based on the spatial dependence of the data [54].

2.5. Statistical Analysis

The return period was calculated as the inverse of the probability of occurrence of consecutive dry days (CDDs). The probability was obtained as the ratio between the number of years in which the CDD occurred within its respective duration class and the total years of the studied historical series [55].

The data (rainfall, CDD1, CDD2, CDD3, cowpea yield) were submitted to the Shapiro–Wilk test, which confirmed the lack of normality. Therefore, nonparametric statistical methods were adopted. The Kruskal–Wallis [56] test was used to verify statistical differences between the variables in the homogeneous rainfall regions in each water scenario. After rejecting the null hypothesis, the groups were submitted to Dunn’s post hoc test to identify regions that differed from each other.

Finally, Spearman’s coefficient was used to verify the degree of association between the parameters under study. The test generates values between −1 ≤ R ≤ 1, with the correlation coefficients considered significant when p < 0.01 [57,58]. The statistical tests were conducted using the SPSS Statistics software, version 23.

3. Results and Discussion

3.1. Hydrological Regime

3.1.1. Rainfall

The climate normal for the state of Ceará is 837 mm yr⁻¹_, with irregular temporal rainfall distribution and the presence of two distinct periods: the rainy season and the dry season. The rainy season, which begins in January and lasts until May (Figure 3), accounted for 62% of the total rainfall for the state, with 521 mm distributed during February, March, and April (137, 197, and 188 mm, respectively). December with 26.3 mm of rain and January with 113.7 mm represent the pre-rainy season, while June (40.4 mm) and July (19.1 mm) mark the transition between the rainy and the dry seasons.

February, March, and April correspond to the rainy season in the semi-arid region of Brazil [38,59,60,61,62]. The greater rainfall during these three months is due to the Intertropical Convergence Zone (ITCZ) being positioned to the south, which favours increased rainfall in the region. However, during May, the ITCZ shifts to the northern hemisphere, resulting in less rainfall [63,64].

In addition to the ITCZ, other atmospheric phenomena contribute to the rainfall in Ceará. From December to January, cold fronts occur due to the meeting of two air masses with different temperatures, while easterly waves occur due to changes in wind circulation, generating intense rainfall from June to August [27,65,66].

To illustrate the predominant differences in the climate of Ceará, the climate normal was verified for each homogeneous rainfall region (Figure 4). The Central Sertão and Inhamuns and Jaguaribana had rainfall of 628 and 693 mm yr⁻¹, respectively, and were, therefore, the regions with the lowest rainfall indices, remaining below the climate average for the state.

Using the Kruskal–Wallis method, the rainfall showed statistically significant differences between the homogeneous rainfall regions under the very dry [X²(2) = 73.202; p < 0.01], dry [X²(2) = 87.670; p < 0.01], normal [X²(2) = 93.229; p < 0.01], rainy [X²(2) = 103.092; p < 0.01], and very rainy [X²(2) = 96.025; p < 0.01] regimes (

Table 2. Analysis of the rainfall by homogeneous rainfall region in the state of Ceará for different water scenarios.

Homogeneous Rainfall Region	Rainfall (mm/yr)
	Very Dry	Dry	Normal	Rainy	Very Rainy
R1—Cariri	423 a	544 a	728 ab	898 ab	1155 ab
R2—Ibiapaba	388 a	565 a	783 ac	1002 ac	1226 ac
R3—Jaguaribana	264 b	387 bc	570 bd	744 bd	939 bd
R4—Coastal Region of Fortaleza	491 a	706 a	985 c	1268 c	1562 c
R5—Coastal Region of Pecém	325 ab	505 ab	703 abc	971 abc	1188 abc
R6—Northern Coastal Region	388 a	579 a	782 ac	1040 ac	1326 ac
R7—Baturité Massif	430 a	608 a	790 ac	998 ac	1232 ac
R8—Central Sertão and Inhamuns	263 b	377 c	533 d	693 d	886 d

Median values followed by the same letter in a column do not differ according to Dunn’s test at 1%.

).

The results of the Kruskal–Wallis analysis (Table 2) are corroborated by the spatial distribution of the rainfall (Figure 5), where, in the very dry (Figure 5a) and dry (Figure 5b) scenarios, the lowest rainfall indices were seen in municipalities located in the regions of Jaguaribana and Central Sertão and Inhamuns. These locations are noteworthy due to the reduction in rainfall.

Under the very dry regime, the regions of Jaguaribana and Central Sertão and Inhamuns recorded 246 and 263 mm, respectively, while under the dry regime, the values were 387 and 377 mm, respectively, differing from the other regions. As a result, the inhabitants of these areas are susceptible to problems caused by water scarcity, compromising agricultural production and the recharge of bodies of water [21,67]. Emergency measures should be adopted by the government to reduce the effects of drought and ensure the survival of the population [12].

Given the increase in water levels under the normal (Figure 5c), rainy (Figure 5d), and very rainy (Figure 5e) regimes, rainfall distribution tended to be similar, with the lowest values recorded in the centre and east of the state and the highest values along the coast and in the mountains and plateaus.

In this respect, of all the regions, Central Sertão and Inhamuns and Jaguaribana differed the most, with the lowest rainfall indices in each of the scenarios. Similar results were found by Andrade et al. [32], Rocha et al. [24], and Nogueira et al. [8] and are associated with the geomorphological model characteristic of these regions, known as the Sertaneja Depression [68].

The Sertaneja Depression is characterised by flat to undulating terrain with crystalline bedrock and low altitudes, and it is surrounded by massifs and plateaus [68]. Since altitude is a major factor in the formation of rainfall, areas at higher altitudes have higher rainfall indices, while areas at lower altitudes are associated with less rainfall [69].

It should be noted that interannual rainfall variability in Ceará is influenced by the sea surface temperature (SST) of the Equatorial Pacific Ocean and the Tropical Atlantic [11,59,60]. An increase in the SST of the Pacific Ocean is known as the El Niño Southern Oscillation (ENSO), and it interferes with the global circulation of air masses, resulting in limited cloud formation in the northeast of Brazil. This means that dry years in the semi-arid region of Brazil are associated with the ENSO phenomenon [70,71].

Furthermore, the SST of the South Atlantic, which is warmer than that further north, also affects the rainfall regime in Ceará [27]. This condition, known as a negative dipole, causes the Intertropical Convergence Zone to migrate to the southern hemisphere, leaving it in the right position to produce rainfall in the region [13,61,72]. Dry and very dry years result from a medium- to high-intensity El Niño associated with a neutral to positive dipole [54], while normal and rainy years occur when a neutral to weak El Niño is associated with a negative dipole.

In addition to these phenomena, La Niña also influences rainfall patterns in the region, producing rainy years when occurring with moderate or weak intensity together with a negative dipole, whereas a neutral to weak La Niña associated with a neutral to positive dipole tends to result in average or below-average rainfall [13,73].

Ceará is, therefore, characterised by temporal and spatial rainfall uncertainty (Table 2; Figure 3, Figure 4 and Figure 5). Despite the total annual rainfall being an important factor, its regular distribution plays a critical role, especially since consecutive dry days are common during the rainy season [3,24,74].

3.1.2. Consecutive Dry Days

There was a marked number of consecutive dry days in Ceará throughout the period under evaluation (1990 to 2019). A total of 39,382 CDDs were recorded, classified as follows: 26,386 (67%) lasting from 5 to 10 days (CDD1); 6219 (16%) lasting from 11 to 15 days (CDD2); and 6777 (17%) lasting more than 15 days (CDD3) (

Table 3. Total CDDs (1990–2019) by homogeneous rainfall region in the state of Ceará.

Homogeneous Rainfall Region	CDD Class 1			Total	Mean CDD (Municipality Year−1)
	CDD1	CDD2	CDD3
Cariri	4496	1216	1022	6734	8.0
Ibiapaba	3541	762	783	5086	6.5
Jaguaribana	3540	899	1136	5575	7.7
Coastal Region of Fortaleza	1862	389	345	2596	6.2
Coastal Region of Pecém	2112	447	536	3095	6.3
Northern Coastal Region	3006	574	659	4239	6.4
Baturité Massif	1841	331	323	2495	6.0
Central Sertão and Inhamuns	5988	1601	1973	9562	8.0
Total	26,386	6219	6777	39,382
Homogeneous Rainfall Region	CDD Return Period (Years)			Rainfall (mm year−1)
	CDD1	CDD2	CDD3
Cariri	1	1	1	843
Ibiapaba	1	2	2	916
Jaguaribana	1	1	1	695
Coastal Region of Fortaleza	1	2	2	1173
Coastal Region of Pecém	1	2	2	861
Northern Coastal Region	1	2	1	925
Baturité Massif	1	2	2	1025
Central Sertão and Inhamuns	1	1	1	628

¹ CDD1: 5 to 10 days; CDD2: 11 to 15 days; CDD3: >15 days.

). Consecutive dry days are caused by environmental conditions that prevent the formation of rain clouds, such as the presence of warm, dry air, where a lack of humidity prevents the air from rising into the atmosphere to form clouds [5].

The highest incidences of consecutive dry days were recorded in the Central Sertão and Inhamuns, Cariri, and Jaguaribana regions, with 9562, 6734, and 5575 phenomena, respectively. It should be noted that the regions with the lowest rainfall indices recorded a greater frequency of consecutive dry days. Furthermore, consecutive dry days lasting 11 to 15 days and over 15 days showed an annual return period in the semi-arid regions such as Central Sertão and Inhamuns, Cariri, and Jaguaribana. Meanwhile, in the mountainous regions, such as the Baturité Massif and Ibiapaba, as well as in the coastal areas, including the Coastal Region of Fortaleza and the Northern Coast, medium and long CDDs recorded a return period of 2 years (Table 3).

Oliveira et al. [75] and Fernandes [35] analysed rainfall data from 1974 to 2012 for Ceará, obtaining approximate values for the frequency of consecutive dry days of 66% for CDD1, 20% for CDD2, and 14% for CDD3. However, according to the authors, there was a total of 12,061 CDD events, fewer than the 39,382 CDDs in the present study.

The difference in the number of events may be associated with two factors: sample size and the period under evaluation. Despite Oliveira et al. [75] and Fernandes [35] considering a period of 39 years, they used only 77 rainfall stations distributed throughout Ceará. In this study, we used 184 rainfall stations with a historical series from 1990 to 2019, including the droughts of 2012 and 2019, factors that may have directly influenced the count of drought events.

Corroborating the results in

Table 4. Number of CDDs lasting from 5 to 10 days by homogeneous rainfall region for different water scenarios.

Homogeneous Rainfall Region	CDD1
	Very Dry			Dry
	Minimum	Maximum	Median	Minimum	Maximum	Median
R1—Cariri	2	7	5 a	4	7	5 a
R2—Ibiapaba	2	7	4 ab	3	6	4 b
R3—Jaguaribana	2	5	4 b	4	6	5 ab
R4—Coastal Region of Fortaleza	3	7	5 a	3	6	5 ab
R5—Coastal Region of Pecém	2	6	4 ab	3	6	4 b
R6—Northern Coastal Region	1	7	4 ab	3	7	5 ab
R7—Baturité Massif	2	7	4 ab	3	7	4 ab
R8—Central Sertão and Inhamuns	2	6	4 ab	3	6	5 ab
Homogeneous Rainfall Region	Normal			Rainy
	Minimum	Maximum	Median	Minimum	Maximum	Median
R1—Cariri	4	7	6 a	4	8	6 a
R2—Ibiapaba	3	7	4 b	2	7	5 abc
R3—Jaguaribana	3	6	5 abc	4	8	5 ab
R4—Coastal Region of Fortaleza	3	6	5 bc	2	6	4 bc
R5—Coastal Region of Pecém	4	6	5 abc	3	6	5 abc
R6—Northern Coastal Region	4	7	5 abc	3	6	5 abc
R7—Baturité Massif	4	6	5 abc	3	5	4 c
R8—Central Sertão and Inhamuns	4	7	5 ac	4	7	6 a
Homogeneous Rainfall Region	Very Rainy
	Minimum	Maximum	Median
R1—Cariri	4	9	5 ac
R2—Ibiapaba	2	6	4 ab
R3—Jaguaribana	4	7	5 c
R4—Coastal Region of Fortaleza	2	5	3 b
R5—Coastal Region of Pecém	2	6	4 ab
R6—Northern Coastal Region	2	7	3 b
R7—Baturité Massif	2	5	3 b
R8—Central Sertão and Inhamuns	1	9	5 c

Median values followed by the same letter in a column do not differ according to Dunn’s test at 1%.

, the Kruskal–Wallis test showed that the rainfall regions differed in the number of CDDs lasting from 5 to 10 days in the very dry [X²(2) = 28.092; p < 0.01]; dry [X²(2) = 22.965; p < 0.01]; normal [X²(2) = 33.056; p < 0.01]; rainy [X²(2) = 45.029; p < 0.01]; and very rainy [X²(2) = 69.512; p < 0.01] water scenarios (Table 4).

In addition to the differences between the homogeneous rainfall regions, the spatial positioning of CDDs is important for visualising areas of climate risk that require interventions and alternative actions to guarantee a water supply and increase agricultural production [4,21,75,76].

In the very dry scenario, there were between three and five CDD1 events, which were more frequent in the coastal area of Fortaleza and in Cariri (Figure 6a), regions that differed statistically from Jaguaribana (Table 4). This can be attributed to the fact that in these regions, the greatest volume of rainfall is concentrated in the very dry scenario (Figure 5a). The smallest number of CDD1 events was recorded in the central, eastern, and northwestern regions of the state, or more precisely, in parts of the Central Sertão and Inhamuns and Jaguaribana. It is important to remember that in years of reduced rainfall, the lower occurrence of short CDDs suggests an increase in prolonged drought events [24,36,76].

In the dry regime, CDD1 had between four and five events, with Cariri standing out in relation to Ibiapaba and the Coastal Region of Pecém (Figure 6b). Cariri had the greatest occurrence of consecutive dry days (5 to 10 days), while in Ibiapaba and the Coastal Region of Pecém, the number of CDD1 events was lower (Table 4). It should also be noted that in scenarios with a water restriction, the presence of CDD1 indicates shorter periods with no rainfall, i.e., more uniform rainfall even in the face of water scarcity [4].

In normal years, there were four to six CDD1 phenomena (Figure 6c), with an increase in consecutive dry days when approaching the southern part of Ceará. The frequency of CDD1 events in the Cariri region was higher than recorded in the coastal region and mountainous areas (Table 4). This may be associated with the reduced influence of ocean humidity in the Cariri region due to being further from the coast, which intensifies the effects of phenomena such as the positive dipole and El Niño [65,77].

Dunn’s test showed that for CDD1, there were few differences between the regions, especially in the very dry, dry, and normal water scenarios (Table 4). The slight variation between the homogeneous rainfall regions shows that the entire state of Ceará is affected by consecutive dry days in extreme drought scenarios.

During rainy years, there were three to seven CDD1 events (Figure 6d). The lowest frequencies were generally seen in the coastal region of Fortaleza, the Baturité Massif, and in the northern and southern regions of Ibiapaba. This may be due to the coastal area being the most influenced by the Intertropical Convergence Zone, which results in regular and intense rainfall [70], whereas mountainous regions experience orographic rainfall, where obstacles, such as mountains, cause the humid air mass to rise, leading to vapour condensation, cloud formation, and rainfall [3,5,31,78].

Finally, under the very rainy regime, there were between three and six CDD1 events (Figure 6e), with a greater difference between locations (Table 4). The Cariri, Jaguaribana, and Central Sertão and Inhamuns regions recorded a higher frequency of CDDs lasting from 5 to 10 days compared to the other regions.

As to the behaviour of CDDs lasting from 5 to 10 days, it should be noted that the Central Sertão and Inhamuns and Jaguaribana regions saw an increase in the number of short-duration CDDs as the rainfall increased, meaning a reduction in events of longer duration. According to Nogueira et al. [8], CDDs of from 5 to 10 days are the most frequent all over Ceará; however, the number of short-duration CDDs tends to decrease as the number of consecutive dry days increases.

In all the water scenarios, ranging from extremely dry to very rainy conditions, Cariri stood out with a predominance of CDDs lasting from 5 to 10 days (CDD1). It should be noted that the Cariri region is characterised by the presence of the Araripe Plateau, a geomorphological structure that acts as a barrier to the movement of the humid winds of the Intertropical Convergence Zone [68], such that the spatial distribution of the rainfall depends on the location of the municipalities in the region in relation to the plateau [79]. As a result, the number of CDD1 events between leeward and windward regions can vary by ±2 events.

For consecutive dry days lasting from 11 to 15 days, the Kruskal–Wallis test showed significant statistical differences between the homogeneous rainfall regions under the very dry [X²(2) = 21.916; p < 0.01], normal [X²(2) = 64.887; p < 0.01], rainy [X²(2) = 56.549; p < 0.01], and very rainy [X²(2) = 98.849; p < 0.01] regimes (

Table 5. Number of CDDs lasting from 11 to 15 days by homogeneous rainfall region for different water scenarios.

Homogeneous Rainfall Region	CDD2
	Very Dry			Dry
	Minimum	Maximum	Median	Minimum	Maximum	Median
R1—Cariri	0	3	2 a	1	3	2 a
R2—Ibiapaba	0	2	2 ab	1	2	1 a
R3—Jaguaribana	0	3	1 ab	1	3	2 a
R4—Coastal Region of Fortaleza	1	2	2 ab	1	2	1 a
R5—Coastal Region of Pecém	0	2	1 ab	0	2	1 a
R6—Northern Coastal Region	0	3	1 ab	0	2	1 a
R7—Baturité Massif	0	2	1 b	0	3	1 a
R8—Central Sertão and Inhamuns	0	3	1 b	0	3	1 a
Homogeneous Rainfall Region	Normal			Rainy
	Minimum	Maximum	Median	Minimum	Maximum	Median
R1—Cariri	1	3	1 a	1	3	1 a
R2—Ibiapaba	0	2	1 b	0	2	1 bc
R3—Jaguaribana	0	2	1 ab	1	2	1 ab
R4—Coastal Region of Fortaleza	1	1	1 b	1	1	1 bc
R5—Coastal Region of Pecém	0	2	1 ab	0	2	1 bc
R6—Northern Coastal Region	0	1	1 b	0	2	1 bc
R7—Baturité Massif	0	1	1 b	0	2	1 c
R8—Central Sertão and Inhamuns	1	2	1 a	0	3	1 a
Homogeneous Rainfall Region	Very Rainy
	Minimum	Maximum	Median
R1—Cariri	0	3	1 a
R2—Ibiapaba	0	2	1 b
R3—Jaguaribana	0	2	1 a
R4—Coastal Region of Fortaleza	0	1	0 b
R5—Coastal Region of Pecém	0	1	1 b
R6—Northern Coastal Region	0	1	0 b
R7—Baturité Massif	0	1	0 b
R8—Central Sertão and Inhamuns	1	2	1 a

Median values followed by the same letter in a column do not differ according to Dunn’s test at 1%.

).

CDD2 varied from one to two phenomena under very dry (Figure 7a), dry (Figure 7b), and normal conditions (Figure 7c). Under the very dry regime, the only difference was seen in Cariri, which stood out for the number of CDD2 events compared to the Baturité Massif and the Central Sertão and Inhamuns regions (Table 5), while under the dry regime, there were no differences between the homogeneous rainfall regions (Table 5). Under the normal regime, Cariri and Central Sertão and Inhamuns differed from the other locations, and they were found to have the highest number of CDDs.

Under the rainy (Figure 7d) and very rainy (Figure 7e) regimes, there were from zero to two CDD2 events. Cariri and Central Sertão and Inhamuns differed from the other homogeneous rainfall regions, with the exception of Jaguaribana. There was a reduction in the number of CDD2 events in the coastal and mountainous regions as water availability increased, with the Central Sertão and Inhamuns and Cariri regions being the most affected by CDDs lasting from 11 to 15 days.

Despite the increase in rainfall, the Central Sertão and Inhamuns region continues to experience difficulties with cloud formation and poor rainfall distribution. Similar results were found by Rocha et al. [18], who studied the occurrence of CDDs and found that CDD2-type events were limited to the southern part of the Central Sertão and Inhamuns region.

Furthermore, Sakamoto et al. [46] observed the same pattern and correlated CDDs of more than 10 days with a loss of up to 73% in temporary-crop production. For Mbanyele et al. [23], medium and long dry periods had a negative impact on rainfed agriculture, resulting in a significant loss in crop productivity.

For CDDs of more than 15 days, the Kruskal–Wallis test showed differences in the homogeneous rainfall regions for each of the scenarios, with CDD3 differing statistically under the very dry [X²(2) = 58.660; p < 0.01], dry [X²(2) = 67.310; p < 0.01], normal [X²(2) = 71.085; p < 0.01], rainy [X²(2) = 82.537; p < 0.01], and very rainy [X²(2) = 44.516; p < 0.01] regimes (

Table 6. Number of CDDs lasting longer than 15 days by homogeneous rainfall region for different water scenarios.

Homogeneous Rainfall Region	CDD3
	Very Dry			Dry
	Minimum	Maximum	Median	Minimum	Maximum	Median
R1—Cariri	1	3	2 b	1	3	2 ab
R2—Ibiapaba	0	4	2 b	1	3	2 b
R3—Jaguaribana	2	4	3 a	2	3	2 a
R4—Coastal Region of Fortaleza	1	3	2 b	1	2	1 b
R5—Coastal Region of Pecém	1	4	2 ab	1	2	2 ab
R6—Northern Coastal Region	1	3	2 b	0	2	2 b
R7—Baturité Massif	1	3	2 b	0	2	1 b
R8—Central Sertão and Inhamuns	1	5	3 a	1	3	2 a
Homogeneous Rainfall Region	Normal			Rainy
	Minimum	Maximum	Median	Minimum	Maximum	Median
R1—Cariri	0	2	1 bc	0	2	1 ac
R2—Ibiapaba	0	3	1 c	0	1	0 b
R3—Jaguaribana	1	2	1 ab	0	2	1 ac
R4—Coastal Region of Fortaleza	0	2	1 c	0	1	0 b
R5—Coastal Region of Pecém	0	2	1 abc	0	1	1 bc
R6—Northern Coastal Region	0	1	1 c	0	1	1 b
R7—Baturité Massif	0	2	1 c	0	1	1 b
R8—Central Sertão and Inhamuns	1	3	2 a	0	2	1 a
Homogeneous Rainfall Region	Very Rainy
	Minimum	Maximum	Median
R1—Cariri	0	2	1 ab
R2—Ibiapaba	0	1	1 b
R3—Jaguaribana	0	2	1 ab
R4—Coastal Region of Fortaleza	0	1	0 b
R5—Coastal Region of Pecém	0	2	1 ab
R6—Northern Coastal Region	0	1	0 b
R7—Baturité Massif	0	1	0 b
R8—Central Sertão and Inhamuns	0	2	1 a

Median values followed by the same letter in a column do not differ according to Dunn’s test at 1%.

).

When evaluating the spatial distribution of CDDs of more than 15 days (CDD3), the highest occurrence was recorded during very dry years, with values of one to four phenomena (Figure 8). The highest frequencies of CDD3 were concentrated in the Central Sertão and Inhamuns and Jaguaribana regions (Figure 8a), compared to the other areas (Table 4). It should be noted that locations with a greater number of CDD3 events had fewer CDD1 events, indicating that the lower the number of short CDDs, the greater the possibility of CDDs of longer duration.

In the dry scenario, the frequency of CDD3 varied from one to three events (Figure 8b). In addition, the spatial distribution under the dry regime followed the same behaviour as in the previous scenario. The regions most affected by CDD3 were Jaguaribana and Central Sertão and Inhamuns, characterised by negative rainfall and the frequent occurrence of CDD3 [18].

Under the normal regime, CDD3 varied between zero and two events (Figure 8c), following the same pattern as the previous regimes, in which the Central Sertão and Inhamuns and Jaguaribana regions differed from the other locations. These regions are characterised by poor rainfall distribution, even in years of normal rainfall, where the irregularity of the rainfall during the rainy season results in long dry periods at these locations [24]. As such, even in scenarios where the water supply is greater, these regions experience no temporal changes in rainfall, only an increase in rainfall depth [8].

Under rainy conditions, the number of CDD3 events varied from zero to two (Figure 8d). In this scenario, the Central Sertão and Inhamuns region stood out for the high frequency of drought events lasting more than 15 days. On the other hand, the lowest number of CDD3 events was recorded in the coastal region, the mountains (Ibiapaba and the Baturité Massif), southern Jaguaribana, and central Cariri.

Under the very rainy regime, there were from zero to two CDD3 events (Figure 8e), with weaker events in more than half the state. However, even in the high rainfall scenario, some municipalities are still likely to face problems due to the uneven rainfall distribution. Similar issues were found by Sakamoto et al. [46], Andrade et al. [32], Rocha et al. [24], and Souza et al. [36].

The Central Sertão and Inhamuns region had the greatest number of CDDs of more than 15 days in each of the water scenarios. In addition to the above issues, this is associated with problems of cloud formation and rainfall distribution in the region [80]. As a result, the location is particularly susceptible to longer-term CDDs [8]. According to Seigerman et al. [67], the Sertão is the region most impacted by natural disasters: of the 3438 drought events recorded between 1991 and 2019, approximately 1480 occurred at this location, corresponding to around 40% of the natural disasters in the state of Ceará during this period [13].

The increase in prolonged periods of drought in the Central Sertão and Inhamuns region shows how difficult it is to develop rainfed agriculture in the area [18,24,35,46]. It should also be noted that the increase in consecutive dry days increases the possibility of affecting different phases of the phenological cycle, especially those most sensitive to water shortage [23,81]. The longer the dry period, the less the soil water is replenished, and the smaller the guarantee of an adequate supply of water for the plant, affecting both metabolic activity and crop productivity [7,22].

3.2. Cowpea (Vigna unguiculata) Productivity

The average area planted with cowpea (Vigna unguiculata) was 529,000 hectares per year, distributed across the eight pluviometrically homogeneous regions. In this context, the Central Sertão and Inhamuns region stands out, with approximately 20% of its area dedicated to the crop each year, while the other regions allocate around 2 to 5% of their territory (Figure 9). The low values are associated with the fact that cowpea is predominantly grown for family consumption [17], playing an important role in combating malnutrition and supporting the food sovereignty of the population [82].

Average cowpea (Vigna unguiculata) productivity in the state of Ceará was 286 ± 85 kg ha⁻¹, a value below the productive potential of the species. Similar results were obtained in research carried out in the semi-arid region of Brazil, where Vasconcelos et al. [83] found an average yield of 273 ± 93 kg ha⁻¹ for Ceará, and Silva et al. [20] found a productivity of 360 kg ha⁻¹ in Paraíba. Compare this to other regions of Brazil, such as the Central-West, where productivity reached 1008 kg ha⁻¹ under a rainfed regime [84].

The low yield in the region may be associated with extensive crop management, the poor use of technology by producers, and climate variability [7,85]. It should also be noted that Ceará had 39,382 CDDs between 1990 and 2019, which, together with the above factors, affects the productive capacity of the state. Other contributing factors are the low level of education of the farmers and the lack of technical assistance and rural extension [86,87].

The yield of temporary crops, such as Vigna unguiculata, changes each year depending on the spatial and temporal distribution of the rainfall [83]. Based on the results for rainfall distribution and consecutive dry days, cowpea yields differed between the homogeneous rainfall regions in each of the water scenarios: very dry [X²(2) = 46.458; p < 0.01], dry [X²(2) = 43.185; p < 0.01], normal [X²(2) = 49.990; p < 0.01], rainy [X²(2) = 73.103; p < 0.01], and very rainy [X²(2) = 87.054; p < 0.01] (

Table 7. Analysis of cowpea (Vigna unguiculata) productivity by homogeneous rainfall region for different water scenarios in the state of Ceará.

Homogeneous Rainfall Region	Cowpea Productivity (kg ha−1)
	Very Dry	Dry	Normal	Rainy	Very Rainy
R1—Cariri	169 ab	223 ab	313 ab	411 a	398 a
R2—Ibiapaba	176 a	207 ab	278 b	286 c	272 bc
R3—Jaguaribana	304 a	271 a	389 a	433 a	411 a
R4—Coastal Region of Fortaleza	274 a	235 a	298 ab	308 abc	324 abc
R5—Coastal Region of Pecém	140 ab	177 ab	259 b	301 bc	256 bc
R6—Northern Coastal Region	146 ab	181 ab	258 b	250 c	191 c
R7—Baturité Massif	241 a	307 a	355 a	379 ab	335 b
R8—Central Sertão and Inhamuns	112 b	168 b	270 b	331 b	288 bc

Mean values followed by the same letter do not differ according to Dunn’s test at 1%.

).

From a spatial perspective, in one very dry (Figure 10a) and one dry (Figure 10b) scenario, the productivity recorded in the Central Sertão and Inhamuns region was lower than in Jaguaribana, the Coastal Region of Fortaleza, or the Baturité Massif, as shown in Table 7. In addition to the factors already discussed, approximately 25% of the Central Sertão and Inhamuns region is occupied by Leptosols (Figure 2). These soils are characterised by low depth, high stoniness, and reduced water-holding capacity, which exacerbate the effects of water scarcity in the region [44].

In a normal year (Figure 10e), the homogeneous rainfall regions of Jaguaribana and the Baturité Massif achieved the highest values for cowpea productivity (Table 7). It should be noted that the Baturité Massif benefits from deep soils that allow water to be stored and reduce water stress during consecutive dry days [17]. Thus, the presence of Acrisols, covering 36% and 25% of the Baturité Massif and Jaguaribana areas, respectively (Figure 2), contributes to more favourable agricultural conditions in these regions. Furthermore, Jaguaribana relies on the River Jaguaribe to supply water to the region [82].

During the rainy season (Figure 10d), cowpea productivity increased in Jaguaribana, Cariri, and the Baturité Massif, while during the very rainy season (Figure 7e), productivity in the Cariri and Jaguaribana regions was higher than that of the other regions (Table 7).

It should be noted that the positive performance seen in Jaguaribana is dependent on the presence of the Jaguaribe Apodi Irrigated Perimeter, where the water supply is guaranteed by the River Jaguaribe [88]. Conversely, Cariri experiences a reduced number of long CDDs (Figure 6), guaranteeing the water supply throughout the crop cycle.

Furthermore, the very rainy scenario saw a fall in productivity, especially in municipalities located in the Central Sertão and Inhamuns region, as well as in regions along the coast; this is linked to the high intensity of the rainfall throughout this scenario, resulting in torrential downpours that can saturate and erode the soil. Excess water in the soil can carry away nutrients, cause the plants to rot, and encourage the proliferation of pests and diseases [12,20,24].

3.3. Correlation Between Variables of the Rainfall Regime

Under very dry to normal water conditions, CDD3 showed a negative correlation with CDD1, CDD2, rainfall, and cowpea productivity (

Table 8. Spearman’s correlation coefficient between the variables CDD1, CDD2, CDD3, rainfall, and cowpea productivity under the very dry, dry, normal, rainy, and very rainy regimes.

Very Dry
	CDD1	CDD2	CDD3	Productivity	Rainfall
CDD1	1.00
CDD2	0.01	1.00
CDD3	−0.53 *	−0.39 *	1.00
Productivity	0.11 *	0.01	−0.18 *	1.00
Rainfall	0.32 *	0.16 *	−0.55 *	0.20 *	1.00
Dry
	CDD1	CDD2	CDD3	Productivity	Rainfall
CDD1	1.00
CDD2	−0.14 *	1.00
CDD3	−0.39 *	−0.22 *	1.00
Productivity	0.10 *	0.01	−0.19 *	1.00
Rainfall	0.05	−0.10 *	−0.52 *	0.18 *	1.00
Normal
	CDD1	CDD2	CDD3	Productivity	Rainfall
CDD1	1.00
CDD2	−0.07 *	1.00
CDD3	−0.40 *	−0.09 *	1.00
Productivity	0.09 *	−0.03	−0.20 *	1.00
Rainfall	−0.06	−0.21 *	−0.49 *	0.09 *	1.00
Rainy
	CDD1	CDD2	CDD3	Productivity	Rainfall
CDD1	1.00
CDD2	−0.05	1.00
CDD3	−0.21 *	−0.03	1.00
Productivity	0.06	0.09 *	0.01	1.00
Rainfall	−0.23 *	−0.23 *	−0.43 *	−0.18 *	1.00
Very Rainy
	CDD1	CDD2	CDD3	Productivity	Rainfall
CDD1	1.00
CDD2	0.12 *	1.00
CDD3	−0.21 *	−0.01	1.00
Productivity	0.27 *	0.13 *	−0.07	1.00
Rainfall	−0.36 *	−0.37 *	−0.37 *	−0.20 *	1.00

DD1: 5 to 10 days; CDD2: 11 to 15 days; CDD3: >15 days; cowpea productivity. * Significant correlation at a level of 1%.

). The occurrence of long dry spells is, therefore, linked to reduced rainfall and has a negative impact on productivity. It should also be noted that the occurrence of CDD3 indicates a reduction in the number of short and moderate events, such as CDD1 and CDD2.

Under very dry conditions, CDD1 showed a positive correlation with rainfall. This suggests that the occurrence of CDDs of up to 10 days under conditions of water scarcity is a positive factor, leading to more regular rainfall distribution due to the small number of days with no rainfall [8,22]. However, under the rainy and very rainy regimes, the negative correlation between rainfall and consecutive dry days intensified regardless of the duration. This shows that scenarios with more rainfall are linked to fewer consecutive dry days.

Productivity and rainfall showed a positive association in the very dry, dry, and normal water scenarios; however, in the rainy and very rainy scenarios, the correlations between the variables were negative. This is related to the greater regularity and intensity of the rainfall, which often results in torrential downpours and flooding [13], while the cowpea is sensitive to water surplus [20,89]. The effects of extreme rainfall on agriculture can be as damaging as those caused by dry periods [90].

In this respect, consecutive dry days can help mitigate the effects of heavy rainfall on agriculture given the positive correlation between cowpea productivity and both CDD1 and CDD2 (Table 8). This shows that the crop benefited from CDDs of up to 15 days during its development cycle under extremely rainy conditions. This is linked to the potential for CDDs to contribute to soil drainage and water evaporation, helping to reduce flooding and minimise prolonged crop damage.

Extremely dry or rainy years are, therefore, subject to a marked fall in productivity. In the very dry scenario, this is due to prolonged CDDs and lower rainfall, while during rainy years, there is an excess of water [31], which hampers the establishment and growth of the crops [24]. Under such conditions, it is essential to adopt measures that are adapted to climate variability to reduce the uncertainty surrounding rainfed agriculture [3,7,29].

Correlations that showed low yet significant values reflect the limited influence of other factors not directly considered, such as soil classes, temperature, topography, and the presence of pests and diseases.

4. Conclusions

This study allowed for a comprehensive understanding of rainfall dynamics, consecutive dry days, and cowpea productivity under different water-availability scenarios in the semi-arid region of Ceará. The results demonstrated that the distribution of rainfall and the duration of consecutive dry days are spatially heterogeneous, directly influencing agricultural productivity.

The Central Sertão and Inhamuns region showed greater susceptibility to water deficits and lower cowpea productivity, highlighting the need for preventive planning and adaptive management strategies to mitigate the impacts of drought. Under extreme conditions, the relationship between consecutive dry days and rainfall revealed complex patterns, with positive correlations in very dry years and behavioural shifts under wetter regimes.

Furthermore, the correlations between rainfall regime variables and cowpea productivity confirmed that both water deficit and surplus negatively affect yields. Future studies may include additional variables, such as temperature and soil characteristics, or apply this methodology to other crops, thereby expanding its analytical potential and contributing to improving the climate resilience of agricultural systems in the semi-arid region.