Dry spells and their impacts on reservoir inflow in semiarid regions: hypothesis outline in a case study of Orós hydrographic basin, Ceará, Brazil

Veranicos e seus impactos na recarga de reservatórios em regiões semiáridas: delineando hipóteses em um estudo de caso para a bacia hidrográfica do Orós, Ceará, Brasil

ABSTRACT

The occurrence of subseasonal phenomena, such as dry spells and flash droughts, are potentially detrimental to the agricultural sector and water resources. The impacts on agricultural sector have been well explored by the technical-scientific community, attesting to the importance of subseasonal variability for this sector, especially for low-tech farming areas. However, when it comes to water resources, there is a large gap regarding the hydrological impacts of these phenomena, in a way that pertinent questions about the importance of subseasonal variability for water resources management remain still unexplored or poorly understood. This study evaluated the impacts of dry spells on reservoir inflow in semiarid regions, taking the Orós reservoir (Northeast Brazil) as a case study. The relationship between the dry spell index and the inflow of Orós reservoir during the rainy season was analyzed, as well as the daily time series of precipitation and storage volume during two long-term dry spells. The two formulated hypotheses suggest that dry spells are indeed capable of reducing the reservoir inflow efficiency in semiarid regions during rainy seasons. These hypotheses serve as a starting point for a deeper discussion on the impacts of dry spells on water resource management in arid and semiarid regions.

Keywords:
Dry spell; Semi-arid; Reservoir inflow; Subseasonal variability

RESUMO

A ocorrência de fenômenos subsazonais, tais como veranicos e secas repentinas, são potencialmente prejudiciais para os setores agrícolas e dos recursos hídricos. Os impactos no setor agrícola foram bem explorados pela comunidade técnico-científica, atestando a importância da variabilidade subsazonal para esse setor, especialmente no cultivo com baixo nível tecnológico. Entretanto, em relação aos recursos hídricos, há uma grande lacuna em relação aos impactos hidrológicos desses fenômenos, de forma que questões pertinentes sobre a importância da variabilidade subsazonal para a gestão desses recursos permanecem inexplorados ou pouco entendidas. O presente estudo objetivou avaliar os impactos dos veranicos na recarga de reservatórios em regiões semiáridas tomando como estudo de caso o reservatório Orós (Nordeste brasileiro). A relação entre o índice de veranico e a recarga desse reservatório durante a estação chuvosa foi avaliada, assim como séries temporais diárias de precipitação e volume armazenado ao longo de dois veranicos de longa duração. As duas hipóteses formuladas sugerem que os veranicos são capazes de reduzir a eficiência da recarga nos reservatórios situados em regiões semiáridas durante a estação chuvosa. Essas hipóteses servem como ponto de partida para uma discussão mais aprofundada sobre os impactos dos veranicos na gestão dos recursos hídricos em regiões áridas e semiáridas.

Palavras-chave:
Veranicos; Semiárido; Recarga; Variabilidade Subsazonal

INTRODUCTION

Droughts are recurring climate extremes and the main natural phenomenon responsible for impacting water and food security in several regions of the planet. The characteristics of drought events can vary in temporal and spatial scale, as well as in severity, but they have a potential multisectoral impact in all cases (Barbosa, 2024; Marengo et al., 2022; Pendergrass et al., 2020; Marengo et al., 2020; Otkin et al., 2018).

Drought events and the dynamic processes responsible for generating them can vary from a subseasonal (weeks) to a decadal scale (Barbosa, 2024; Otkin et al., 2018). Droughts on subseasonal scales have been less explored by the scientific community than events on other temporal scales (e.g., seasonal or interannual). As a result, there is still a large gap in the understanding of these phenomena. However, in recent years, several studies have sought to understand the physical processes that cause subseasonal droughts and their potential impacts in various geographic regions, such as: Sungmin & Park (2024), Rodrigues et al. (2024), Otkin et al. (2022), Christian et al. (2021), Parker et al. (2021) and Pendergrass et al. (2020).

Historically, semiarid areas in Northeast Brazil (NEB) have been marked by major social and economic losses associated with the occurrence of droughts, reported since the 16^th century. As a consequence of recurring droughts and a precipitation pattern that is concentrated in February-May, multiple reservoirs have been built in NEB as a government strategy to guarantee water supply for different uses, alleviating the effects of precipitation seasonality and droughts (Marengo et al., 2022; Campos, 2015).

The occurrence of droughts in NEB is associated with climate variability patterns on multiple time scales. On the seasonal and interannual scale, for example, both the El-Niño Southern Oscillation (ENSO), in the Pacific Ocean, and the Tropical Atlantic Interhemispheric Gradient (TAIG), in the Atlantic Ocean, are of substantial influence. On the subseasonal scale, the Madden-Julian Oscillation (MJO) stands out, for it is capable of suppressing or intensifying precipitation meteorological systems that influence the region, depending on what phase is active (Rodrigues et al., 2024; Andrade et al., 2023; Marengo et al., 2020).

Subseasonal phenomena, such as dry spells and flash droughts, can negatively impact the agricultural sector and others that depend on water resources (Rodrigues et al., 2024; Jarrett et al., 2023; Rocha et al., 2021; Vasconcelos et al., 2019). Dry spells are defined as prolonged periods of abnormally low rainfall during the rainy season, which can last from days to weeks (Jarrett et al., 2023). Flash droughts, on the other hand, are climatic extremes that develop on a subseasonal-to-seasonal scale (weeks to months) and have the remarkable characteristic of sudden onset and rapid intensification of drought conditions (Pendergrass et al., 2020).

Despite the multisectoral impacts of these phenomena, much of the focus has been solely on agriculture, certainly due to its significant economic implications (Otkin et al., 2022). In fact, several studies have investigated the impact of dry spells and flash droughts on agricultural production in different regions of the planet: Jarrett et al. (2023), Forstner et al. (2023), Fall et al. (2021), Hoell et al. (2020), Hunt et al. (2021) and Rocha et al. (2021).

Nevertheless, studies concerning the hydrological impacts of subseasonal-scale phenomena on water supply and water-dependent sectors are less abundant. Otkin et al. (2022) argued that this scarcity comes from a lack of understanding of possible relationships and feedbacks between those phenomena, stakeholders, and environmental involved, which may greatly vary according to the analyzed region. Thus, some issues remain unexplored or poorly understood, such as: what is the relevance of events on subseasonal scale for water resources management; is it possible that improving the predictability of these events would benefit the decision-making process in water resource management?

Given these issues, this study evaluated the impacts of dry spells on the reservoir inflows in semiarid regions during the rainy season. For this purpose, the Orós reservoir, located in the State of Ceará, Northeast Brazil, was chosen as the study area. Daily time series of reservoir storage and precipitation were analyzed, seeking to establish relationships between the occurrence and characteristics of dry spells and the replenishment of Orós during the rainy season. In addition, two years with long dry spells (2016 and 2018) were separately evaluated, in order to understand the influence of these events on the water inflow behavior of the Orós.

MATERIALS AND METHODS

Study area

As study area, this research adopted the hydrographic basin of the Orós reservoir, located in the State of Ceará, Northeast Brazil (Figure 1), in a sub-basin of the Jaguaribe River basin (Alto Jaguaribe). Orós is the second largest reservoir in Ceará and part of the Jaguaribe-Metropolitan System,

Figure 1
Location of Orós Hydrographic Basin in relation to the State of Ceará and Brazil.

With a storage capacity of approximately 1,940.00 hm³, Orós is the main reservoir in Alto Jaguaribe river basin, located in the Southwestern portion of Ceará, bordering the States of Piauí (West) and Pernambuco (South). The headwaters of Jaguaribe River are located within this basin (325 km away from Orós). The predominant vegetation is caatinga, which is currently very degraded in the region (Ceará, 2009).

Due to its large storage capacity, the Orós reservoir is an important structure in guaranteeing water security in the state of Ceará, acting as an inter-annual water reserve that transfers the volumes accumulated in wet periods to dry periods. As of recently, water storage structures in Ceará were severely impacted by the multi-year (2012-2018) drought that happened in NEB. As pointed out by Pontes Filho et al. (2020), between 2012 and 2016, the volume stored in the State of Ceará decreased by approximately 63%.

In addition to the wide availability of high-quality rainfall information, as seen in Figure 1, the Orós Reservoir was chosen as a study case because it is the most upstream reservoir in the Jaguaribe-Metropolitano system (Estácio et al., 2022). This decision was made in order to avoid the presence of a large upstream storage structure that could perennialize the streamflows to the analyzed reservoir. This perennialization could mitigate/mask the effects of dry spells on the water inflow.

Reservoir storage data

Daily time series of Orós reservoir storage were obtained from Ceará Hydrological Portal, ran by Ceará Foundation for Meteorology and Water Resources (FUNCEME)¹. This information is generated from daily water level measurements, obtaining the stored volume of the reservoir through its Elevation-Area-Volume (EAV) relationship. Daily water level measurements are carried out for all reservoirs in the state of Ceará by the Ceará State Water Supplies Management Company (COGERH). The time series of stored volumes in reservoirs in the state of Ceará began in 2004 and has been updated daily ever since (Fundação Cearense de Meteorologia e Recursos Hídricos, 2024b).

The present study considered the historical series of storage volume for Orós reservoir between 2004 and 2021. Although data from more recent years were available, 2021 was chosen in order to match the end of the time series for observed precipitation. The daily inflow of the Orós reservoir was defined as the daily fluctuation of its storage volume. In this way, it is possible that the daily reservoir inflow takes on positive values, in the case of filling, or negative values, in the case of depletion.

Precipitation data – acquistion and processing

Daily precipitation data were obtained from the database provided by FUNCEME², which has an extensive network of pluviometric stations distributed throughout the state (Fundação Cearense de Meteorologia e Recursos Hídricos, 2024b). The information from these stations was used to define the average precipitation of the Orós hydrographic basin, as well as for a spatialized assessment of long-term dry spell events in 2016 and 2018 for the entire state of Ceará.

Average precipitation

The average rainfall of the Orós watershed was determined using the Thiessen method, considering the rain gauge stations shown in Figure 1. In summary, the Thiessen method consists of a weighted average of the precipitation of these stations by their influence areas. Because it only takes into account the station's influence area, the Thiessen method faces challenges in accurately representing the average precipitation in areas with complex topography (Goovaerts, 2000). Furthermore, when applied to a rainfall network that is poorly distributed relative to the study area, it can produce biased average rainfall information because it does not adequately address the spatial heterogeneity of rainfall.

As shown in Figure 1, the selected rain gauge stations to perform the Thiessen method are well distributed in the Orós hydrographic basin. In addition, this watershed has a simple topography, with the highest elevations concentrated in a small part of the upstream region. For these reasons, and due to its simplicity of application, we considered it appropriate to use the Thiessen method to determine the average precipitation of the Orós hydrographic basin. The daily precipitation time series obtained covered the period from January 01, 1974 to December 31, 2021 uninterruptedly (i.e., with no data failures in the considered interval).

Definition of rainy season

Figure 2 presents the climatological precipitation of Orós hydrographic basin, obtained through observed data ranging between January 1^st, 1974 and December 31^st, 2021.

Figure 2
Climatological precipitation of Orós hydrographic basin, obtained from observed precipitation between 1974 and 2021.

Medians of the boxplots in Figure 2 show that there is a well-defined seasonality of precipitation, which clearly differentiates the Dry Season (DS) from the Wet Season (WS) in the basin. On average, the rainy season goes from January to April, with peak precipitation occurring between February and April. The dry season happens between May and December.

However, the lower and upper limits of the boxplots demonstrated that both beginning and end of rainy season may vary, reducing or extending its duration each year. In this context, we decided to define the beginning and end of the WS for each analyzed year, since fixing it in months of climatological maximum could lead to erroneous detection of dry spells, for it could consider periods prior to its real beginning in years in which the rainy season was reduced. The time series for Orós’ water inflow began in 2004, so this analysis was performed from that year onwards.

The definition of beginning and end of the WS was performed using the methodology presented by Liebmann et al. (2007), which considers the accumulated anomaly of daily precipitation for a more precise delimitation of the WS as presented in Equation 1:

$A\left(d\right)={\displaystyle \sum }_{n=1}^d\left[P\left(n\right)-\overline{P}\right]$ (1)

in which: A(d) is the deviation of accumulated daily precipitation up to a certain day, d; P(n) is the daily precipitation on a certain day, n; and $\overline{P}$ is the climatological average daily precipitation within the climatological year.

As recommended by Liebmann et al. (2007), the start of a climatological year should be 10 days before the month of climatological minimum. In this study, the month of climatological minimum was August (as seen in Figure 2), so the climatological year was designated to start on August 22 of any considered year until the same date of the following year.

In this methodology, the beginning of the WS corresponds to the day following the beginning of the longest period in which the accumulated anomaly of daily precipitation remains with a positive trend. The end corresponds to the day in the climatological year in which the accumulated anomaly reaches its maximum value (Liebmann et al., 2007). In this study, these definitions were maintained and the 5-day moving average has been applied to accumulated anomaly of daily precipitation to simplify its daily fluctuations.

Table 1 presents starting and end days, as well as the duration of the WS (in number of days) for each evaluated year (2004 to 2021). Supplementary Material A presents the accumulated anomalies of daily precipitation for these years, highlighting the start and end of the rainy season.

Table 1
Beginning, end, and duration (in number of days) of the rainy season for the hydrographic basin of Orós reservoir between 2004 and 2021.

Year	Beginning	End	Duration (in days)
2004	Jan. 11	April 11	91
2005	Jan. 14	May 09	115
2006	Jan. 31	May 19	108
2007	Feb. 01	May 06	94
2008	Jan. 14	May 10	117
2009	Jan. 17	May 22	125
2010	Dec. 31	May 18	138
2011	Jan. 20	May 23	123
2012	Jan. 18	April 13	86
2013	Jan. 03	June 27	175
2014	Jan. 07	May 09	122
2015	Feb. 03	June 11	128
2016	Jan. 06	April 20	105
2017	Jan. 16	April 17	91
2018	Jan. 13	May 05	112
2019	Jan. 09	May 04	115
2020	Dec. 31	May 01	122
2021	Jan. 25	May 10	105

Characterization and dry spell index

We defined dry spells as the occurrence of periods of three or more days with a precipitation of less than 2 mm during the rainy season, as presented by Rocha et al. (2021).

The impacts of dry spells are directly proportional to their duration. Thus, their detection alone is not sufficient to establish a relationship between their occurrence and resulting impacts. Furthermore, a single rainy season may contain multiple occurrences of these events, so that the total impact of the dry spell on the rainy season is given by the combination of impacts of each event.

In this context, a dry spell index was determined for each evaluated rainy season (Table 1), seeking to put together (i) the duration of dry spells and (ii) the multiple occurrences of dry spells within the WS. We adopted an adaptation of the Dry Spell Index (DSI) proposed by Ferijal et al. (2021), as shown by Equation 2:

$DSI=\frac{1}{RSL}{\displaystyle \sum }_{j=1}^k{\displaystyle \sum }_{i=1}^n{w}_i$ (2)

in which: $RSL$ is the duration of the rainy season (in days); $w_i$ is the weight associated with the i^th day of a dry spell event; n is the duration of the j^th dry spell occurred in the rainy season; and k is the number of dry spells occurred in a certain analyzed rainy season.

As presented by Ferijal et al. (2021), the weights $\left(w_i\right)$ in Equation 2 increase linearly throughout the dry spell event, so that the first day of the dry spell has a weight equal to 1 $\left(w_1=1\right)$, whereas the n^th day has a weight $n$ $\left(w_n=n\right)$. This means that the term ${\displaystyle \sum }_{i=1}^n{w}_i$ represents the sum of an arithmetic progression of unitary ratio.

Thus, longer-lasting dry spells, which have greater potential impacts, will have a higher sum of weights (that is, a greater influence on DSI). Besides, this is temporally independent, so it is possible to account for multiple dry spell events occurring during the rainy season.

Analysis of dry spell events in 2016 and 2018

While evaluating the occurrence of dry spells in the daily precipitation series of Orós reservoir basin between 2004 and 2021 (period in which daily reservoir inflow data are available), it was noticed that long-term dry spell events occurred in 2016 and 2018. The dry spell in 2016 lasted for 34 days (Feb. 3^rd to March 8^th), and in 2018, for 19 days (March 4^th to March 22^nd), as shown in Figure 3.

Figure 3
Daily precipitation data for the Orós watershed between (a) January 1, 2016, and April 30, 2016, and (b) January 1, 2018, and April 30, 2018. The vertical, dotted lines comprise periods when the analyzed dry spell events occurred in 2016 and 2018. The red solid lines represent the daily precipitation during these events.

The behavior of the daily replenishment volume of Orós reservoir in response to the occurrence of these two dry spells was separately evaluated. For this purpose, the daily precipitation and reservoir inflow series were analyzed from January 1st to April 30th in both years, highlighting the beginning and end of the dry spell events. A 5-day moving average was applied to the daily series, filtering out daily fluctuations. In addition, this moving average was used to filter – at least in part – the influence of the propagation time of precipitated water in the river basin on the daily reservoir inflow series.

Complementarily, the dry spells of 2016 and 2018 were also evaluated in spatial terms within the State of Ceará, using data from the extensive network of pluviometric stations operated by FUNCEME (Fundação Cearense de Meteorologia e Recursos Hídricos, 2024b). Information on daily precipitation was spatialized for the entire state using the Kriging spatial interpolation method, implemented using a Python library called pykrige (Murphy, 2014).

To perform kriging, the daily precipitation series along the WS of each evaluated year were accumulated in pentads. Pluviometric stations with absence of daily precipitation data in the considered period were disregarded. In addition, the accumulated precipitation in the pentads in each of the selected pluviometric stations was normalized using the expression shown in Equation 3:

$P_{ac,norm}=\frac{\left(P_{ac}-P_{ac,min}\right)}{\left(P_{ac,max}-P_{ac,min}\right)}$ (3)

where: $P_{ac}$ is the accumulated precipitation in the pentad; $P_{ac,norm}$ is the normalized accumulated precipitation in the pentad; $P_{ac,max}$ and $P_{ac,min}$ are the maximum and minimum accumulated precipitation in a pentad during the WS of the analyzed year.

Normalization allowed the values of accumulated precipitation in the pentads to be set between 0 and 1, making it possible to compare pluviometric stations with different magnitudes of accumulated precipitation. For the spatial representation of the dry spell events in 2016 and 2018, we selected pentads that encompassed the occurrence of dry spells events and another two pentads prior to this period. Thus, 9 pentads were selected for 2016, and 6 pentads for 2018.

Table 2 shows the period in which daily precipitation was accumulated for the spatial evaluation of dry spell events in 2016 and 2018, aiming to define the accumulated precipitation in each pentad.

Table 2
Pentads used for spatial evaluation of the 2016 and 2018 dry spell events, listing the periods in which daily precipitation was accumulated to determine the accumulated precipitation of each pentad.

2016		2018
Pentad 1	24 Jan. - 28 Jan.	Pentad 1	22 Feb. - 26 Feb.
Pentad 2	29 Jan. - 02 Feb.	Pentad 2	27 March - 03 March
Pentad 3*	03 Feb. - 07 Feb.	Pentad 3*	04 March - 08 March
Pentad 4*	08 Feb. - 12 Feb.	Pentad 4*	09 March - 13 March
Pentad 5*	13 Feb. - 17 Feb.	Pentad 5*	14 March - 18 March
Pentad 6*	18 Feb. - 22 Feb.	Pentad 6*	19 March - 23 March
Pentad 7*	23 Feb. - 27 Feb.
Pentad 8*	28 Feb. - 03 March
Pentad 9*	04 March - 08 March

• Pentads marked with (*) comprise the period when dry spells occurred in 2016 and 2018.

Soil moisture data

Soil Moisture (SM) data for the Orós reservoir catchment were obtained from Zeri et al. (2020). Zeri et al. (2020) presented a soil moisture dataset over the Brazilian semiarid region from August 2015 to April 2019. In this data set, soil volumetric water content was obtained by measuring the dielectric constant of the soil at a frequency of 70 MHz, taking hourly measurements at a depth of 0.1 m and 0.2 m. A total of 360 ground stations are available throughout the Brazilian semi-arid region, which have undergone quality control and had their consistency checked. It should be noted that gaps due to malfunctions or communication failures are occasional in the data, and stations with at least 2 months of measurements were kept.

Out of 360 ground stations in the Zeri et al. (2020) database, only eight are located in the Orós reservoir catchment area, as can be seen in Figure 1. Table 3 shows the coordinates and station ID of the 8 ground stations in the Orós reservoir watershed.

Table 3
Coordinates and ID of ground stations located in the Orós reservoir hydrographic basin.

Station ID	Longitude	Latitude
230160401C	−39.78°	−6.92°
230360001C	−39.87°	−6.14°
230550601C	−39.36°	−6.27°
230740301C	−39.45°	−6.43°
231135501C	−39.20°	−6.25°
231210601C	−39.78°	−7.16°
230030901U	−39.51°	−6.12°
231195901U	−40.45°	−7.26°

Based on these 8 stations (Table 3), the behavior of soil moisture during the 2016 and 2018 dry spells was evaluated. As can be seen in Figure 1, these ground stations are poorly distributed throughout the Orós basin, making it impossible to represent soil moisture in average terms or even to interpolate to a regular grid across the basin. Therefore, we decided to analyze soil moisture behavior at a point scale – that is, separately for each station – during these dry spell events.

As mentioned above, the soil volumetric water content measurements are hourly and are taken at depths of 0.1 m and 0.2 m. In order to evaluate the dry spells events that occurred in 2016 and 2018, however, the average value of all hourly measurements contained in one day was used, giving rise to a daily SM time series. These daily series were determined for the period from January 1st to April 30th in both years.

We noted that the measurements at 0.2 m depth had a large absence of data for 2018, making it impossible to assess the behavior of the SM during the dry spell of that year. Therefore, we decided to only consider measurements at 0.1 m depth for 2018 and, for standardization, also for 2016.

For 2016, a total of 7 out of the 8 stations shown in Table 3 had enough data to make it possible to assess the behavior of soil moisture during that year's dry spell. For 2018, only 2 of the 8 stations provided enough data to allow such an analysis.

RESULTS

DSI-reservoir inflow relation

Figure 4 presents a scatter plot of DSI and Orós reservoir inflow percentage (ratio of inflow volume and storage capacity) for the WS in each evaluated year (see Table 1). Points in the scatter plot were colored according to the SPI12³ value of the last month of the adopted climatological year for each evaluated year (i.e., the month of August).

Figure 4
Scatter plot of DSI and percentual inflow of the Orós reservoir during the rainy season of years 2004-2021. The SPI12 of the last month of the adopted hydrological year (August) is represented by the color of each point.

In Figure 4, reservoir inflow decreases with the increase in DSI, but without a very clear pattern. As for example, for DSI values between 1.5 and 2.0, reservoir inflow values vary between −3.21% and 14.97%, a similar fluctuation is observed for DSI values between 0.5 and 1.0. When evaluating in relation to SPI12, it can be seen that reservoir inflows of more than 20% only occurs for SPI12 values greater than approximately +0.5. For SPI12 values below zero, there is no clear pattern between these two variables, with percentage inflows ranging from −3.16 (SPI12 = −0.99) to +19.92% (SPI12 = −0.05). Thus, a directly proportional relationship between SPI12 values and reservoir inflow during the WS became fairly evident only for values greater than ~+0.5.

The relationships described above led to the conclusion that the seasonal component, represented by SPI12, is predominant for the reservoir inflow during the WS. The relationship observed between DSI and the reservoir inflow may arise from the inversely proportional correlation between SPI12 and this index, that is, a relationship between multiple time scales of climate variability.

This inversely proportional correlation between DSI and SPI12 can be explained by the definition of DSI. The DSI measures the level of aridity of a rainy season, based on the ratio between the sum of the weighted durations of dry spells and the duration of the considered period. Therefore, high DSI values represent a rainy season with longer sequences of dry days, which could impact the precipitation volumes of that climatological year and, consequently, reduce its SPI12. However, it is worth noting that high precipitation before or after the suppression of precipitation can increase the SPI12, even for high DSI values.

2016 dry spell event

Figure 5 shows the normalized accumulated precipitation of the pentads between Jan.24, 2016 and Mar. 08, 2016 (as shown in Table 2), spatially distributed within the State of Ceará, highlighting the region where Orós basin is located. The 2016 dry spell occurred between February 03 and March 08, totaling 34 days, corresponding to Pentads 3 to 9.

Figure 5
Normalized Accumulated Precipitation in the pentads between January 24, 2016 and March 08, 2016 for the State of Ceará, highlighting the region where Orós basin is located. The 2016 dry spell occurred from Pentad 3 to Pentad 9.

The evolution of the normalized accumulated precipitation in the evaluated pentads showed the magnitude of precipitation suppression during the dry spell of 2016, both for the Ceará State and Orós basin. This suppression was significant in spatial and quantitative terms.

In pentad 2, despite observing accumulated precipitation around 50% and between 80% and 100% of the maximum accumulated precipitation during the WS of 2016 in Eastern and Southern Ceará, it was possible to evidence a significant reduction in precipitation volumes in the other regions in comparison with Pentad 1, mainly in Orós basin.

From Pentad 3 to Pentad 9, the suppression of precipitation in Ceará was almost total, except for some regions in the North, which presented accumulated precipitation around 20% of the maximum accumulated precipitation during the 2016 WS. In addition, precipitation around 50% of this maximum accumulated precipitation was observed in the Northeastern portion of the state (Pentad 6), as well as small regions with precipitation between 80% and 100% in Eastern areas (Pentad 7).

Specifically for Orós basin, it is possible to say that the suppression of precipitation during the dry spell of 2016 was constant and spatially uniform throughout the entire basin, except for some areas with normalized accumulated precipitation around 20% of the maximum accumulated precipitation in the region during the WS of 2016 (Pentads 6, 7, and 9).

Figure 6 presents the 5-day moving averages of observed daily precipitation and daily Orós reservoir inflow between January 01, 2016 and April 30, 2016. The vertical dotted lines in Figure 6 represent the period when the dry spell happened in 2016.

Figure 6
5-day moving averages concerning (a) daily precipitation observed in the hydrographic basin of Orós reservoir and (b) Orós reservoir inflow during the 2016 dry spell (Jan. 01-April 30). The vertical, dotted lines delimitate the period when the dry spell occurred.

The WS in 2016 happened between January 06 and April 20 (see Table 1). During that period, there was a predominance of moving average values below climatology, except for the initial periods, as seen in Figure 6a. This was mainly, but not exclusively, due to the occurrence of the dry spell, from February 03 to March 08.

The occurrence of the dry spell quickly led to a period of depletion of Orós reservoir, which – as expected – lasted throughout the duration of the event. By continuously evaluating Orós’ inflow, it was observed that the above-average precipitation before the dry spell ensured a considerable daily filling of the reservoir

Given the onset of the dry spell, a small filling was still observed in the first days of the event, probably sustained by subsurface runoff from the previous above-average precipitation. However, this residual runoff quickly ceased, giving room a long period of depletion of the reservoir during the rainy season.

At the end of the dry spell, a short period of above-average precipitation occurred, inducing a low filling, quickly terminated by a period of below-average precipitation. Then, the following sequence of precipitation events was observed: i) a short period of above-average precipitation; ii) a period of below-average precipitation, yet above the dry spell threshold; and iii) a short period of above-average precipitation. This sequence triggered a season of high reservoir inflow, with a daily filling peak similar to the inflow levels observed prior to the dry spell event.

Figure 7 shows the daily Soil Moisture (SM) measurements at 10 cm depth between January 01, 2016 and April 30, 2016 in the ground stations located in the hydrographic basin of Orós (Figure 1). The vertical dotted lines in Figure 7 represent the period when the dry spell happened in 2016.

Figure 7
Daily SM at 10 cm depth between January 01, 2016 and April 30, 2016 in the ground stations located in the hydrographic basin of Orós. The vertical, dotted lines delimitate the period when the dry spell occurred.

It can be seen that, despite the different magnitude of station 231210601C, the behavior of SM at all the considered stations is similar. There was a rapid increase in SM after the start of the 2016 rainy season (January 6), except for stations 230030901U and 231135501C, which showed a more gradual increase. Between January 20 and January 25, all the stations showed SM peaks prior to the occurrence of the 2016 dry spell event.

At the beginning of February, all the stations showed decreasing behavior in SM, probably due to the below-average rainfall during this period (Figure 6a). From the start of the dry spell event (February 03), all the stations showed sharp reductions in observed SM at 10 cm depth, except for stations 230030901U and 231135501C, where this reduction was more gradual. Moreover, SM was decreasing throughout the summer, except for stations 231210601C and 230160401C, which showed a positive oscillation during this period.

At the end of the dry spell event (March 08), all the stations showed significant reductions in SM in relation to the values reached between Jan-20 and Jan-25, in some cases dropping to SM levels similar to or lower than those observed in the period prior to the start of the rainy season. It is worth noting that the stations that showed a gradual increase in SM also showed more gradual reductions during the summer event. This behavior may be associated with the hydraulic properties of the soil in the region of these ground stations

2018 dry spell event

Figure 8 shows the normalized accumulated precipitation of Pentads between February 22, 2018 and March 23, 2018 (as shown in Table 2), spatially distributed within the State of Ceará, highlighting the region where Orós basin is located. The 2018 dry spell occurred between March 04 and March 22 (19 days), corresponding to Pentads 3 to 6.

Figure 8
Normalized Accumulated Precipitation in the pentads between February 22, 2018 and March 23, 2018 for the State of Ceará, highlighting the region where Orós basin is located. The 2018 dry spell occurred from Pentad 3 to Pentad 6.

Similarly to the 2016 dry spell, the evolution of normalized accumulated precipitation in Pentads 3 to 6 confirmed the magnitude of the precipitation suppression during the 2018 dry spell, both for State of Ceará and Orós basin. This precipitation suppression was significant in quantitative and spatial terms.

Pentads 1 and 2, which happened before the dry spell, showed several regions of the State with accumulated precipitation that represented more than 50% of the maximum accumulated precipitation during the WS of 2018. From Pentad 3 (when dry spell begins), a suppression of precipitation was observed, and it increased until Pentad 5 throughout Ceará, both in quantitative and spatial terms.

This suppression was alleviated in Pentad 6, even though a low value of accumulated precipitation was still maintained in most of the State – around 20% of the maximum accumulated precipitation in the WS of 2018 –, except for a few regions that reached around 50% of this maximum accumulated precipitation.

For Orós basin, Pentad 1 showed accumulated precipitation equivalent to 50% of the maximum accumulated precipitation in the WS of 2018 in most areas. In Pentad 2, this value was intensified, reaching between 80% and 100% of maximum precipitation.

From the beginning of the dry spell event (Pentad 3), a strong and spatially uniform suppression of precipitation was observed, which persisted until Pentad 5. In Pentad 6, this suppression was relieved, causing the accumulated precipitation in the region to reach approximately 20% of the maximum accumulated precipitation in the rainy season of 2018.

Figure 9 shows the 5-day moving averages of observed daily precipitation and daily Orós reservoir inflow from January 01, 2018 to April 30, 2018. The vertical dotted lines represent the period when the 2018 dry spell occurred.

Figure 9
5-day moving averages concerning (a) daily precipitation observed in the hydrographic basin of Orós reservoir and (b) Orós reservoir inflow during the 2018 dry spell (January 01-April 30). The vertical, dotted lines delimitate the period when the dry spell occurred.

The WS of 2018 happened between January 13 and May 05, as shown in Table 1. It can be seen that this rainy season presented a greater precipitation volume than that of 2016, with longer periods of precipitation above climatology.

Similar to 2016, a small filling was observed in the initial days of the dry spell event, which can be attributed to the previous period of above-average precipitation. After this filling ceased, a phase of reservoir depletion began, which continued with the end of the dry spell event. After that, a period of below-average precipitation began, followed by a considerable period of above-average precipitation, which generated the largest reservoir inflow in the analyzed period.

Figure 10 shows the daily Soil Moisture (SM) measurements at 10 cm depth between January 01, 2018 and April 30, 2018 in the ground stations located in the hydrographic basin of Orós (Figure 1). The vertical dotted lines in Figure 10 represent the period when the dry spell happened in 2018.

Figure 10
Daily SM at 10 cm depth between January 01, 2018 and April 30, 2018 in the ground stations located in the hydrographic basin of Orós. The vertical, dotted lines delimitate the period when the dry spell occurred.

The only two ground stations in Table 3 with available data for the evaluated period in 2018 showed very similar behavior. Although the start of the 2018 rainy season occurred on January 13, the increasing behavior of SM in both stations did not begin until just before the start of February. This behavior can be explained by the below-average rainfall at the start of the rainy season, including long periods in which the 5-day moving average showed values below the dryspell threshold (Figure 9a).

Shortly before the beginning of February, there is a period with precipitation above the dry spell threshold. This period, although below average, is sufficient to promote some gain in SM. With the onset of the prolonged period of above-average precipitation, a rapid increase in SM occurs in both stations, reaching the highest levels prior to the occurrence of the dry spell.

With the onset of the dry spell event (March 04), there was a rapid decline in SM. This decline continued throughout the dry spell and, by its end (March 22), had resulted in a significant reduction in SM. For the station 230360001, despite the absence of measurements at 10 cm depth during the event, it is observed that it reaches levels similar to those observed in the days prior to the start of the 2018 rainy season.

DISCUSSIONS

Relationship between dry spells and the Orós reservoir inflow: a hypothesis design in light of the surface runoff generation mechanisms

The behavior of the Orós reservoir inflow given the occurrence of long-term dry spells, such as those that occurred in 2016 (Figure 6) and 2018 (Figure 8), allows two hypotheses to be raised about the impact of dry spells on the reservoir inflow in semi-arid regions in light of the hydrometeorological processes responsible for the generation of surface runoff and the climatic characteristics of these regions

The first hypothesis suggests that the suppression of precipitation during the dry spell leads to a reduced inflow to the reservoir. When inflow is lower than the sum of regularized streamflow and lake’s evaporation and infiltration, the reservoir volume gets depleted, as can be seen in Figures 6b and 8b. This hypothesis originates from the simple water balance relationship present in a water storage reservoir.

The seasonality of precipitation – characteristic of Brazilian semiarid regions – gives great importance to this first hypothesis, since the annual reservoir inflows in these areas occurs mainly in the few months when annual precipitation is concentrated, i.e., dry spell periods can reduce the reservoir inflows or even lead to the depletion of the reserved volume during the main period of the hydrological year when replenishment occurs

Complementary to the first hypothesis, the second hypothesis proposes that dry spells, as long as they last long enough, are capable of influencing the efficiency of the basin’s rainfall-runoff conversion in precipitation events that occur after dry spells. This hypothesis can be justified based on the high Evaporative Demand (ED), characteristic of semiarid regions, which generate a rapid decline in available Soil Moisture (SM) due to successive days with no precipitation. This rapid decline is supported by the behavior of SM during the dry spell events seen in Figure 7 (2016 dry spell) and Figure 10 (2018 dry spell). A reduction in available SM compromises the efficiency of runoff generation processes, both through to saturation-excess mechanism (Dunne & Black, 1970) and to infiltration-excess mechanism (Horton, 1945).

In the Infiltration-Excess Mechanism (IEM), also known as Hortonian processes, the surface runoff occurs when the rainfall intensity surpasses the infiltration capacity of the soil, regardless whether the underlying soil are wet or dry (Horton, 1945). The Saturation-Excess Mechanism (SEM), in turn, requires the complete saturation of the soil profile, either by the elevation of the wetting front or by the existence of an impermeable layer that results in a return flow of subsurface runoff. It is only after this complete saturation that subsequent precipitation will generate surface runoff and thus effectively contribute to channel flow (Dunne & Black, 1970; Kidron, 2021). Another crucial difference between these two mechanisms is that in IEM the soil is saturated from above by infiltration, while in SEM the soil is saturated from below by subsurface runoff.

Regarding IEM, the abovementioned reduction in available SM makes it difficult to generate runoff by this process in the rainfall events subsequent to the end of the dry spell event. When rainwater encounters relatively dry soil, due to successive days of rainfall suppression and the high ED of semi-arid regions, it is quickly absorbed as the soil has many voids and a high matric potential, reducing the efficiency of rainfall-runoff conversion. Successive precipitation events, or even a very intense precipitation, are capable of filling these soil voids, limiting infiltration capacity to the soil's ability to transfer water to its deeper layers. This reduction/limitation of the soil's infiltration capacity, in turn, can favor the generation of IEM runoff.

In relation to the generation of runoff by the SEM, the need for precipitation on saturated soil to generate surface runoff supports the hypothesis of partial contributing areas (Betson, 1964), which can expand or shrink over time. In this context, the greater the contributing area, i.e. the greater the saturated area of the catchment, the greater the volume of runoff generated by the precipitation event. By reducing the available SM due to the high DE in semi-arid regions, dry spells events reduce the basin's partial contribution area, resulting in lower runoff generation, i.e. reducing the efficiency of rainfall-runoff conversion. The successive occurrences of rainfall after the dry spell event, however, gradually restore the basin's partial contribution areas, recovering the efficiency of rainfall-runoff conversion and favoring the generation of runoff by the SEM.

Figures 6b and 8b support the second hypothesis and the impact of dry spells on runoff generation mechanisms. For the 2016 dry spell (Figure 6b), the three subsequent periods with precipitations above climatological average generated increasing reservoir inflow, according to their order of occurrence. Therefore, it is possible to say that the efficiency in the rainfall-runoff conversion was low in the first period but increased as the following precipitation events occurred (even when they were below climatological average). Hence, the last precipitation period showed a high efficiency in the rainfall-runoff conversion, generating more surface runoff and, consequently, greater replenishment volume for the reservoir. The 2018 dry spell (Figure 8b) was similar to 2016, although the precipitation below climatological average after the occurrence of the dry spell allowed efficiency in the rainfall-runoff conversion to be reestablished before the period of precipitation above climatological average, resulting in the peak reservoir inflow of the evaluated period.

Relationship between dry spells and transmission losses

A relationship can also be established between the rapid decline in available Soil Moisture (SM) of the second hypothesis and losses in runoff transmission. Transmission losses can be defined as reductions in runoff volume, as water moves downstream in ephemeral rivers in arid and semiarid regions, which could consist of several processes, such as: initial filling of cavities, secondary channels, and floodplains along the channel during the rise of the hydrograph; infiltration on riverbanks and riverbed; evaporation of water in the channel; and evapotranspiration in riparian areas directly connected to it. This process plays an important role in outflow rates and groundwater recharge and, as a consequence, in the supply of freshwater and also in water-dependent ecosystems (McMahon & Nathan, 2021; Costa et al., 2012).

Technical literature reports runoff transmission losses as mostly concentrated in the infiltration process occurring in the riverbed, as well as in the sum of water evaporation (E) occurring in the river channel and the evapotranspiration (ET) of riparian areas directly connected to the river (McMahon & Nathan, 2021).

For the Diamantina River, in the central part of Australia (an arid region), Jarihani et al. (2015) estimated that E and ET components totaled 61% of runoff transmission losses, whereas Costelloe et al. (2003) pointed out 72% as its total. Nevertheless, as stated by McMahon & Nathan (2021), the representativeness of these components in total transmission losses could be distinct, since there are different types of soils and rocky beds that can favor the infiltration process in detriment of E and ET.

Here, we address the components of transmission loss separately: i) Evapotranspiration components, encompassing E in the river channel and ET in the riparian areas, and ii) Infiltration Components, which includes initial filling of cavities, secondary channels and floodplains and infiltration on riverbanks and riverbed

Evapotranspiration components of transmission losses

In semiarid regions, the occurrence of dry spells can favor E and ET in runoff transmission losses. This favoring effect can be explained by the relationship between ED and ET in these regions, as presented by Pendergrass et al. (2020). Starting from a scenario of sufficient soil moisture (ET limited by ED – energy limited conditions), the suppression of precipitation during dry spells results in a rapid drop in this moisture due to the high ED in semiarid regions.

During this drop, a sensible heat flux emerges from the excess ED and intensifies air temperature, producing an amplifying feedback in ED and, consequently, in ET as well. Under these conditions, the continued suppression of precipitation causes soil moisture to become insufficient for the existing ED (ET limited by the available soil moisture – water limited conditions). From this point on, ET decreases along with the available moisture, increasing the excess ED and its amplifying feedback.

Given the above and assuming the second hypothesis as true, precipitation events after a dry spell would encounter a high ED due to the aforementioned amplifying feedback. This high ED would increase transmission losses resulting from the E of water in the channel and the ET of the riparian areas directly connected to the river, leading to reductions in streamflow.

Infiltration components of transmission losses

SM loss during dry spell events due to high ED in semiarid regions can also increase the infiltration component of transmission loss: initial filling of cavities, secondary channels and floodplains and infiltration on riverbanks and riverbed.

Especially for the infiltration process in the river banks and riverbed., the reduction in available SM can lead to a reduction in the water table which, when lower than the river stage, causes streamwater to seep into the aquifer, reducing the river's runoff and turning it into a losing river.

Indeed, large-scale studies suggest that most rivers in the Brazilian drylands are predominantly losing rivers because the water table in these regions is usually below the local topography, which favors the infiltration of streamwater into aquifers (Uchôa et al., 2024).

Another factor also influences river-aquifer iterations in dryland rivers. As pointed out by Azevedo Toné et al. (2023), during the dry season the water table drops below the river channel and a clogging layer form in the river bed, with transmission losses predominating even during the early stages of the rainy season (< 10 mm/day). However, after more intense rainfall events (> 20 mm/day), the clogging layer is removed and subsequent rainfall events cause the water table to rise, initiating a period of predominantly transmission gains.

In this context, the occurrence of a dry spell event long enough to affect the available SM could lead to an interruption of the rainy season period with a predominance of transmission gains, as pointed out by Azevedo Toné et al. (2023). During this interruption, there could be a predominance of transmission losses through infiltration, as mentioned above, reducing the river's runoff and extending the period of the year in which the river is classified as a losing-river.

Study limitations and further analysis

Influence of upstream reservoir network

Although we tried to minimize the effects of large upstream regularizations by choosing the Orós reservoir as a case study, this methodological framework does not take into account the complex upstream network of small and medium reservoirs. It is estimated that the upper Jaguaribe River basin, where the Orós reservoir is located, contains about 3,500 reservoirs with an area of more than 1 ha (Meira Neto et al., 2024).

The extensive network of reservoirs upstream of the Orós reservoir could influence the actual impact of dry spells on its replenishment. As reported by Meira Neto et al. (2024), the expansion of reservoirs in the upper Jaguaribe River basin has led to a significant reduction in streamflows at the basin outlet, where the Orós reservoir is located, and an increase in evaporated volume at the basin scale due to the continuous exposure of open water surfaces at various locations.

Furthermore, there is an extensive literature documenting the effects of dense reservoir networks in dryland catchments: modifying the occurrence patterns of hydrological droughts (Ribeiro Neto et al., 2022, 2024); accelerating the transition from meteorological to hydrological droughts (Colombo et al., 2024) and shifting water storage upstream during meteorological droughts., intensifying hydrological droughts downstream and alleviating them upstream (van Langen et al., 2021; van Oel et al., 2018).

Due to the recurrent periods of drought in the Brazilian semi-arid region, the construction of small and medium reservoirs to supply water to small rural communities is quite common (Rabelo et al., 2022). Therefore, we recommend that future studies consider the effect of the upstream reservoir network in modulating the impacts of dry spell on the water inflows of strategic reservoirs in semi-arid regions, such as the Orós reservoir.

Influence of irrigation water demand

Despite the predominance of corn and bean cultivation through rainfed family farming in the agricultural production of the state of Ceará, which depends on the amount of water available during the wet season (Rocha et al., 2021), water demand for irrigation is not negligible and plays an important role in the water inflow to the state's reservoirs.

Regarding the occurrence of dry spells, it is worth noting that there is a close relationship between the suppression of precipitation in these events and the demands for irrigation. Irrigation systems installed in agricultural production areas generally remain unused during the period when the natural water supply from rainfall is sufficient to meet the physiological water demands of the crops.

Once the dry spell event begins, this natural supply of water through rainfall ceases and an artificial supply of water through irrigation must be introduced to meet these physiological needs and mitigate the climatic risk to agricultural production. The increase in water demand for irrigation during this period can accelerate the propagation of the effects of precipitation suppression on surface runoff, intensifying the impact of dry spells on reservoir inflows.

Despite the aforementioned relationship, this methodological framework did not allow for an assessment of how irrigation demands modulate the impacts of dry spells on the Orós reservoir inflows. Therefore, we suggest that future studies consider the effects of irrigation demand on modulating the effects of dry spell events on reservoir inflows in semi-arid regions.

Dry spell and flash drought

It is worth highlighting the relationship between dry spells and flash droughts in drylands catchments. One of the characteristics of drylands catchments is that Evaporative Demand (ED) exceeds Evapotranspiration (ET), so that flash droughts can be caused only by a precipitation deficit. This will generate a rapid decrease in soil moisture, which will be intensified by the increase in air temperature due to the excessive sensible heat flux (Pendergrass et al., 2020; Wang & Yuan, 2018; Mo & Lettenmaier, 2016).

Despite the long duration of the analyzed dry spell events, characterizing them as flash droughts would require further analyses to assess the sudden onset and rapid intensification of the drought state throughout these events.

SUMMARY AND RECOMMENDATIONS

This study evaluated the impact of dry spells on the water inflows of reservoirs located in semiarid regions. For this purpose, the hydrographic basin of Orós reservoir was taken as the study area, considering its daily time series of precipitation and storage volume. The results and presented discussion indicated that the occurrence of long-term dry spells can indeed impact the reservoir inflows in semiarid regions with well-defined precipitation seasonality. However, this influence (subseasonal component) is not predominant in the replenishment during the rainy season.

The analysis of the dry spell events occurred in 2016 and 2018, considering the daily precipitation and storage volume series, made the impacts of dry spells on the Orós reservoir inflow more evident, allowing two hypotheses (complementary to each other) to be outlined in light of the hydrometeorological processes responsible for the generation of surface runoff and the climatic characteristics of these regions

• Hypothesis 1: the suppression of precipitation during dry spell events leads to a reduction in the inflows to the reservoir, which can generate a depletion of reservoir volume during the rainy season, if the inflows are lower than the sum of regularized streamflow and lake’s evaporation; and

• Hypothesis 2: dry spells, as long as they last long enough, are capable of influencing the efficiency of the basin’s rainfall-runoff conversion in precipitation events following their occurrence.

Results presented in this article provided evidence in favor of these hypotheses. For example, the periods of depletion of the stored volume during the dry spells events of 2016 (Figure 6) and 2018 (Figure 8) argue in favor of the first hypothesis. The second hypothesis, on the other hand, is supported by the behavior of the reservoir inflow after the dry spell events, which starts low, possibly as a result of a low efficiency in rainfall-runoff conversion, and tends to increase with the occurrence of rainfall events, indicating that this efficiency is gradually restored.

Also for the second hypothesis, the assumption of a reduction in the efficiency of rainfall-runoff conversion was supported by the reduction in available soil moisture due to the suppression of precipitation during the dry spells and the high evaporative demand of semi-arid regions. This reduction was observed in Figure 7, for the 2016 dry spell, and in Figure 10, for the 2018 dry spell.

Despite the evidence presented in favour of these hypotheses, we suggest that additional analyses be carried out to validate them more robustly, overcoming the limitations of the current methodological structure used, especially those highlighted in the topic “Study limitations and further analysis”.

However, it is evident that the two hypotheses presented in this study serve as a starting point for a deeper understanding of the impacts of dry spells on water resources management in arid and semiarid regions. Assuming both to be true, one can state that dry spells are indeed capable of reducing reservoir inflow efficiency during the rainy season.

In this context, water resources management and planning should outline strategies to mitigate the impacts of dry spells on reservoir replenishment, seeking to maximize their efficiency during rainy seasons. As an example, these strategies could focus on ways to contribute to maintaining soil moisture levels during dry spells. In addition, strategies could appoint changes in regularized flows, ensuring that the reservoir volume is not depleted during dry spells. In both cited examples, improvements in the ability to predict these events are of great value, as they can aid stakeholders in early decision-making processes.

DATA AVAILABILITY STATEMENT

The data supporting the findings of this study are freely available from Fundação Cearense de Meteorologia e Recursos Hídricos at the following links: i) http://www.funceme.br/?page_id=2694 and ii) http://www.hidro.ce.gov.br/.

Supplementary Material

Supplementary material accompanies this paper.

Figure A. 1 –

Accumulated daily rainfall anomalies for the climatological year of 2004, highlighting the beginning and end of the rainy season.

Figure A. 2 –

Accumulated daily rainfall anomalies for the climatological year of 2005, highlighting the beginning and end of the rainy season.

Figure A. 3 –

Accumulated daily rainfall anomalies for the climatological year of 2006, highlighting the beginning and end of the rainy season.

Figure A. 4 –

Accumulated daily rainfall anomalies for the climatological year of 2007, highlighting the beginning and end of the rainy season.

Figure A. 5 –

Accumulated daily rainfall anomalies for the climatological year of 2008, highlighting the beginning and end of the rainy season.

Figure A. 6 –

Accumulated daily rainfall anomalies for the climatological year of 2009, highlighting the beginning and end of the rainy season.

Figure A. 7 –

Accumulated daily rainfall anomalies for the climatological year of 2010, highlighting the beginning and end of the rainy season.

Figure A. 8 –

Accumulated daily rainfall anomalies for the climatological year of 2011, highlighting the beginning and end of the rainy season.

Figure A. 9 –

Accumulated daily rainfall anomalies for the climatological year of 2012, highlighting the beginning and end of the rainy season.

Figure A. 10 –

Accumulated daily rainfall anomalies for the climatological year of 2013, highlighting the beginning and end of the rainy season.

Figure A. 11 –

Accumulated daily rainfall anomalies for the climatological year of 2014, highlighting the beginning and end of the rainy season.

Figure A. 12 –

Accumulated daily rainfall anomalies for the climatological year of 2015, highlighting the beginning and end of the rainy season.

Figure A. 13 –

Accumulated daily rainfall anomalies for the climatological year of 2016, highlighting the beginning and end of the rainy season.

Figure A. 14 –

Accumulated daily rainfall anomalies for the climatological year of 2017, highlighting the beginning and end of the rainy season.

Figure A. 15 –

Accumulated daily rainfall anomalies for the climatological year of 2018, highlighting the beginning and end of the rainy season.

Figure A. 16 –

Accumulated daily rainfall anomalies for the climatological year of 2019, highlighting the beginning and end of the rainy season.

Figure A. 17 –

Accumulated daily rainfall anomalies for the climatological year of 2020, highlighting the beginning and end of the rainy season.

Figure A. 18 –

Accumulated daily rainfall anomalies for the climatological year of 2021, highlighting the beginning and end of the rainy season.

This material is available as part of the online article from https://doi.org/10.1590/2318-0331.302520240131

• 1
  Data from Ceará Hydrological Portal can be found at Fundação Cearense de Meteorologia e Recursos Hídricos (2024a).
• 2
  All data collected by FUNCEME’s pluviometric stations is shared free of charge and can be found at Fundação Cearense de Meteorologia e Recursos Hídricos (2024b).
• 3
  SPI12 is the 12-month Standardized Precipitation Index, a meteorological drought indicator that measures precipitation anomalies within a 12-month period.

REFERENCES

• Andrade, F. M., Godoi, V. A., & Aravéquia, J. A. (2023). Why above-average rainfall occurred in northern Northeast Brazil during the 2019 El Niño? Meteorology, 2(3), 307-328. http://doi.org/10.3390/meteorology2030019
  » http://doi.org/10.3390/meteorology2030019
• Azevedo Toné, A. J., Costa, A. C., Gonçalves Barros, M. U., Barros de Freitas, H., Vasconcelos Cavalcante, J. R., & Lima Neto, I. E. (2023). Spatiotemporal river-aquifer interactions of a large tropical dryland river with high anthropic intervention. Hydrological Sciences Journal, 68(14), 2009-2026. http://doi.org/10.1080/02626667.2023.2252816
  » http://doi.org/10.1080/02626667.2023.2252816
• Barbosa, H. A. (2024). Understanding the rapid increase in drought stress and its connections with climate desertification since the early 1990s over the Brazilian semi-arid region. Journal of Arid Environments, 222, 105142. http://doi.org/10.1016/j.jaridenv.2024.105142
  » http://doi.org/10.1016/j.jaridenv.2024.105142
• Betson, R. P. (1964). What is watershed runoff? Journal of Geophysical Research, 69(8), 1541-1552. http://doi.org/10.1029/JZ069i008p01541
  » http://doi.org/10.1029/JZ069i008p01541
• Campos, J. N. B. (2015). Paradigms and public policies on drought in northeast Brazil: A historical perspective. Environmental Management, 55(5), 1052-1063. http://doi.org/10.1007/s00267-015-0444-x
  » http://doi.org/10.1007/s00267-015-0444-x
• Ceará. (2009). Caderno Regional da Sub-Bacia do Alto Jaguaribe Fortaleza: INESP, Conselho de Altos Estudos e Assuntos Estratégicos, Assembleia Legislativa do Estado do Ceará.
• Christian, J. I., Basara, J. B., Hunt, E. D., Otkin, J. A., Furtado, J. C., Mishra, V., Xiao, X., & Randall, R. M. (2021). Global distribution, trends, and drivers of flash drought occurrence. Nature Communications, 12(1), 6330. http://doi.org/10.1038/s41467-021-26692-z
  » http://doi.org/10.1038/s41467-021-26692-z
• Colombo, P., Ribeiro Neto, G. G., Costa, A. C., Mamede, G. L., & Van Oel, P. R. (2024). Modeling the influence of small reservoirs on hydrological drought propagation in space and time. Journal of Hydrology, 629, 130640. http://doi.org/10.1016/j.jhydrol.2024.130640
  » http://doi.org/10.1016/j.jhydrol.2024.130640
• Costa, A. C., Bronstert, A., & Araújo, J. C. (2012). A channel transmission losses model for different dryland rivers. Hydrology and Earth System Sciences, 16(4), 1111-1135. http://doi.org/10.5194/hess-16-1111-2012
  » http://doi.org/10.5194/hess-16-1111-2012
• Costelloe, J. F., Grayson, R. B., Argent, R. M., & McMahon, T. A. (2003). Modelling the flow regime of an arid zone floodplain river, Diamantina River, Australia. Environmental Modelling & Software, 18(8-9), 693-703. http://doi.org/10.1016/S1364-8152(03)00071-9
  » http://doi.org/10.1016/S1364-8152(03)00071-9
• Dunne, T., & Black, R. D. (1970). An experimental investigation of runoff production in permeable soils. Water Resources Research, 6(2), 478-490. http://doi.org/10.1029/WR006i002p00478
  » http://doi.org/10.1029/WR006i002p00478
• Estácio, A. B. S., Rocha, M. A. M., Oliveira, M. C., Silva, S. M. O., & Souza Filho, F. (2022). Priority of water allocation during drought periods: the case of Jaguaribe Metropolitan inter-basin water transfer in semiarid Brazil. Sustainability, 14(11), 6876. http://doi.org/10.3390/su14116876
  » http://doi.org/10.3390/su14116876
• Fall, C. M. N., Lavaysse, C., Kerdiles, H., Dramé, M. S., Roudier, P., & Gaye, A. T. (2021). Performance of dry and wet spells combined with remote sensing indicators for crop yield prediction in Senegal. Climate Risk Management, 33, 100331. http://doi.org/10.1016/j.crm.2021.100331
  » http://doi.org/10.1016/j.crm.2021.100331
• Ferijal, T., Batelaan, O., & Shanafield, M. (2021). Spatial and temporal variation in rainy season droughts in the Indonesian Maritime Continent. Journal of Hydrology, 603, 126999. http://doi.org/10.1016/j.jhydrol.2021.126999
  » http://doi.org/10.1016/j.jhydrol.2021.126999
• Forstner, V., Vremec, M., Herndl, M., & Birk, S. (2023). Effects of dry spells on soil moisture and yield anomalies at a montane managed grassland site: a lysimeter climate experiment. Ecohydrology: Ecosystems, Land and Water Process Interactions. Ecohydrology, 16(3), 1. http://doi.org/10.1002/eco.2518
  » http://doi.org/10.1002/eco.2518
• Fundação Cearense de Meteorologia e Recursos Hídricos – FUNCEME. (2024a). Portal hidrológico do Estado do Ceará. Retrieved in 2024, July 29, from http://www.hidro.ce.gov.br/
  » http://www.hidro.ce.gov.br/
• Fundação Cearense de Meteorologia e Recursos Hídricos – FUNCEME. (2024b). Postos pluviométricos do Estado do Ceará. Retrieved in 2024, July 29, from http://www.funceme.br/?page_id=2694
  » http://www.funceme.br/?page_id=2694
• Goovaerts, P. (2000). Geostatistical approaches for incorporating elevation into the spatial interpolation of rainfall. Journal of Hydrology, 228(1-2), 113-129. http://doi.org/10.1016/S0022-1694(00)00144-X
  » http://doi.org/10.1016/S0022-1694(00)00144-X
• Hoell, A., Parker, B.-A., Downey, M., Umphlett, N., Jencso, K., Akyuz, F. A., Peck, D., Hadwen, T., Fuchs, B., Kluck, D., Edwards, L., Perlwitz, J., Eischeid, J., Deheza, V., Pulwarty, R., & Bevington, K. (2020). Lessons learned from the 2017 flash drought across the U.S. Northern Great Plains and Canadian Prairies. Bulletin of the American Meteorological Society, 101(12), E2171-E2185. http://doi.org/10.1175/BAMS-D-19-0272.1
  » http://doi.org/10.1175/BAMS-D-19-0272.1
• Horton, R. E. (1945). Erosional development of streams and their drainage basins: hydrophysical approach to quantitative morphology. Geological Society of America Bulletin, 56(3), 275-370. http://doi.org/10.1130/0016-7606(1945)56[275:EDOSAT]2.0.CO;2
  » http://doi.org/10.1130/0016-7606(1945)56[275:EDOSAT]2.0.CO;2
• Hunt, E., Femia, F., Werrell, C., Christian, J. I., Otkin, J. A., Basara, J., Anderson, M., White, T., Hain, C., Randall, R., & McGaughey, K. (2021). Agricultural and food security impacts from the 2010 Russia flash drought. Weather and Climate Extremes, 34, 100383. http://doi.org/10.1016/j.wace.2021.100383
  » http://doi.org/10.1016/j.wace.2021.100383
• Jarihani, A. A., Larsen, J. R., Callow, J. N., McVicar, T. R., & Johansen, K. (2015). Where does all the water go? Partitioning water transmission losses in a data-sparse, multi-channel and low-gradient dryland river system using modelling and remote sensing. Journal of Hydrology, 529, 1511-1529. http://doi.org/10.1016/j.jhydrol.2015.08.030
  » http://doi.org/10.1016/j.jhydrol.2015.08.030
• Jarrett, U., Miller, S., & Mohtadi, H. (2023). Dry spells and global crop production: A multi-stressor and multi-timescale analysis. Ecological Economics, 203, 107627. http://doi.org/10.1016/j.ecolecon.2022.107627
  » http://doi.org/10.1016/j.ecolecon.2022.107627
• Kidron, J. G. (2021). Comparing overland flow processes between semiarid and humid regions: does saturation overland flow take place in semiarid regions? Journal of Hydrology, 593, 125624. http://doi.org/10.1016/j.jhydrol.2020.125624
  » http://doi.org/10.1016/j.jhydrol.2020.125624
• Liebmann, B., Camargo, S. J., Seth, A., Marengo, J. A., Carvalho, L. M. V., Allured, D., Fu, R., & Vera, C. S. (2007). Onset and end of the rainy season in South America in observations and the ECHAM 4.5 atmospheric general circulation model. Journal of Climate, 20(10), 2037-2050. http://doi.org/10.1175/JCLI4122.1
  » http://doi.org/10.1175/JCLI4122.1
• Marengo, J. A., Cunha, A. P. M. A., Nobre, C. A., Ribeiro Neto, G. G., Magalhaes, A. R., Torres, R. R., Sampaio, G., Alexandre, F., Alves, L. M., Cuartas, L. A., Deusdará, K. R. L., & Álvala, R. C. S. (2020). Assessing drought in the drylands of northeast Brazil under regional warming exceeding 4 °C. Natural Hazards, 103(2), 2589-2611. http://doi.org/10.1007/s11069-020-04097-3
  » http://doi.org/10.1007/s11069-020-04097-3
• Marengo, J. A., Galdos, M. V., Challinor, A., Cunha, A. P., Marin, F. R., Vianna, M. dos S., Alvala, R. C. S., Alves, L. M., Moraes, O. L., & Bender, F. (2022). Drought in Northeast Brazil: a review of agricultural and policy adaptation options for food security. Climate Resilience and Sustainability, 1(1), e17. http://doi.org/10.1002/cli2.17
  » http://doi.org/10.1002/cli2.17
• McMahon, T. A., & Nathan, R. J. (2021). Baseflow and transmission loss: A review. WIREs. Water, 8(4), e1527. http://doi.org/10.1002/wat2.1527
  » http://doi.org/10.1002/wat2.1527
• Meira Neto, A. A., Medeiros, P., de Araújo, J. C., Pereira, B., & Sivapalan, M. (2024). Evolution of drought mitigation and water security through 100 years of reservoir expansion in semi‐arid Brazil. Water Resources Research, 60(9), e2023WR036411. http://doi.org/10.1029/2023WR036411
  » http://doi.org/10.1029/2023WR036411
• Mo, K. C., & Lettenmaier, D. P. (2016). Precipitation deficit flash droughts over the United States. Journal of Hydrometeorology, 17(4), 1169-1184. http://doi.org/10.1175/JHM-D-15-0158.1
  » http://doi.org/10.1175/JHM-D-15-0158.1
• Murphy, B. S. (2014). PyKrige: development of a kriging toolkit for Python. In American Geophysical Union, Fall Meeting 2014 (H51K-0753), San Francisco, CA. Retrieved in 2024, September 24, from https://ui.adsabs.harvard.edu/abs/2014AGUFM.H51K0753M/abstract
  » https://ui.adsabs.harvard.edu/abs/2014AGUFM.H51K0753M/abstract
• Otkin, J. A., Svoboda, M., Hunt, E. D., Ford, T. W., Anderson, M. C., Hain, C., & Basara, J. B. (2018). Flash droughts: a review and assessment of the challenges imposed by rapid-onset droughts in the United States. Bulletin of the American Meteorological Society, 99(5), 911-919. http://doi.org/10.1175/BAMS-D-17-0149.1
  » http://doi.org/10.1175/BAMS-D-17-0149.1
• Otkin, J. A., Woloszyn, M., Wang, H., Svoboda, M., Skumanich, M., Pulwarty, R., Lisonbee, J., Hoell, A., Hobbins, M., Haigh, T., & Cravens, A. E. (2022). Getting ahead of flash drought: from early warning to early action. Bulletin of the American Meteorological Society, 103(10), E2188-E2202. http://doi.org/10.1175/BAMS-D-21-0288.1
  » http://doi.org/10.1175/BAMS-D-21-0288.1
• Parker, T., Gallant, A., Hobbins, M., & Hoffmann, D. (2021). Flash drought in Australia and its relationship to evaporative demand. Environmental Research Letters, 16(6), 064033. http://doi.org/10.1088/1748-9326/abfe2c
  » http://doi.org/10.1088/1748-9326/abfe2c
• Pendergrass, A. G., Meehl, G. A., Pulwarty, R., Hobbins, M., Hoell, A., AghaKouchak, A., Bonfils, C. J. W., Gallant, A. J. E., Hoerling, M., Hoffmann, D., Kaatz, L., Lehner, F., Llewellyn, D., Mote, P., Neale, R. B., Overpeck, J. T., Sheffield, A., Stahl, K., Svoboda, M., Wheeler, M. C., Wood, A. W., & Woodhouse, C. A. (2020). Flash droughts present a new challenge for subseasonal-to-seasonal prediction. Nature Climate Change, 10(3), 191-199. http://doi.org/10.1038/s41558-020-0709-0
  » http://doi.org/10.1038/s41558-020-0709-0
• Pontes Filho, J. D., Souza Filho, F., Martins, E. S. P. R., & Studart, T. M. C. (2020). Copula-based multivariate frequency analysis of the 2012–2018 drought in Northeast Brazil. Water, 12(3), 834. http://doi.org/10.3390/w12030834
  » http://doi.org/10.3390/w12030834
• Rabelo, U. P., Costa, A. C., Dietrich, J., Fallah-Mehdipour, E., Van Oel, P., & Lima Neto, I. E. (2022). Impact of dense networks of reservoirs on streamflows at dryland catchments. Sustainability, 14(21), 14117. http://doi.org/10.3390/su142114117
  » http://doi.org/10.3390/su142114117
• Ribeiro Neto, G. G., Melsen, L. A., Martins, E. S. P. R., Walker, D. W., & van Oel, P. R. (2022). Drought Cycle Analysis to evaluate the influence of a dense network of small reservoirs on drought evolution. Water Resources Research, 58(1), e2021WR030799. http://doi.org/10.1029/2021WR030799
  » http://doi.org/10.1029/2021WR030799
• Ribeiro Neto, G. G., Melsen, L. A., Costa, A. C., Walker, D. W., Cavalcante, L., Kchouk, S., Brêda, J. P., Martins, E. S. P. R., & van Oel, P. R. (2024). Clash of drought narratives: a study on the role of Small Reservoirs in the emergence of drought impacts. Earth’s Future, 12(7), e2023EF004311. http://doi.org/10.1029/2023EF004311
  » http://doi.org/10.1029/2023EF004311
• Rocha, T. B. C., Vasconcelos Júnior, F. C., Silveira, C. S., Martins, E. S. P. R., Gonçalves, S. T. N., Silva, E. M., Alves, J. M. B., & Sakamoto, M. S. (2021). Indicadores de veranicos e de distribuição de Chuva no Ceará e os impactos na agricultura de sequeiro. Revista Brasileira de Meteorologia, 36(3, Suppl.), 579-589. http://doi.org/10.1590/0102-77863630041
  » http://doi.org/10.1590/0102-77863630041
• Rodrigues, B. D., Silveira, C. S., Vasconcelos Júnior, F. C., Brito Neto, F. A., Silva, I. A., Sakamoto, M. S., & Martins, E. S. P. R. (2024). Atmospheric and oceanic mechanisms in precipitation in March 2018 in Ceará, Brazil. Theoretical and Applied Climatology, 155(9), 8633-8650. http://doi.org/10.1007/s00704-024-05143-x
  » http://doi.org/10.1007/s00704-024-05143-x
• Sungmin, O., & Park, S. K. (2024). Global ecosystem responses to flash droughts are modulated by background climate and vegetation conditions. Communications Earth & Environment, 5(1), 88. http://doi.org/10.1038/s43247-024-01247-4
  » http://doi.org/10.1038/s43247-024-01247-4
• Uchôa, J. G. S. M., Oliveira, P. T. S., Ballarin, A. S., Meira Neto, A. A., Gastmans, D., Jasechko, S., Fan, Y., & Wendland, E. C. (2024). Widespread potential for streamflow leakage across Brazil. Nature Communications, 15(1), 10211. http://doi.org/10.1038/s41467-024-54370-3
  » http://doi.org/10.1038/s41467-024-54370-3
• van Langen, S. C. H., Costa, A. C., Ribeiro Neto, G. G., & van Oel, P. R. (2021). Effect of a reservoir network on drought propagation in a semi-arid catchment in Brazil. Hydrological Sciences Journal, 66(10), 1567-1583. http://doi.org/10.1080/02626667.2021.1955891
  » http://doi.org/10.1080/02626667.2021.1955891
• van Oel, P. R., Martins, E. S. P. R., Costa, A. C., Wanders, N., & van Lanen, H. A. J. (2018). Diagnosing drought using the downstreamness concept: the effect of reservoir networks on drought evolution. Hydrological Sciences Journal, 63(7), 979-990. http://doi.org/10.1080/02626667.2018.1470632
  » http://doi.org/10.1080/02626667.2018.1470632
• Vasconcelos, T. S., Moraes, J. G. L., Alves, J. M. B., Jacinto Júnior, S. G., Oliveira, L. L. B., Silva, E. M., & Sousa, G. G. (2019). Variabilidade pluviométrica no Ceará e suas relações com o cultivo de milho, feijão-caupi e mandioca (1987-2016). Revista Brasileira de Meteorologia, 34(3), 431-438. http://doi.org/10.1590/0102-7786343053
  » http://doi.org/10.1590/0102-7786343053
• Wang, L., & Yuan, X. (2018). Two types of flash drought and their connections with seasonal drought. Advances in Atmospheric Sciences, 35(12), 1478-1490. http://doi.org/10.1007/s00376-018-8047-0
  » http://doi.org/10.1007/s00376-018-8047-0
• Zeri, M., Costa, J. M., Urbano, D., Cuartas, L. A., Ivo, A., Marengo, J., & Alvalá, R. C. S. (2020). A soil moisture dataset over the Brazilian semiarid region. Mendeley Data, V2 http://doi.org/10.17632/xrk5rfcpvg.2
  » http://doi.org/10.17632/xrk5rfcpvg.2

Edited by

• Editor-in-Chief:
  Adilson Pinheiro

• Associated Editor:
  Edson Cezar Wendland

Publication Dates

• Publication in this collection
  15 Aug 2025
• Date of issue
  2025

History

• Received
  02 Dec 2024
• Reviewed
  08 Apr 2025
• Accepted
  21 May 2025