PLUVIOMETRIC VARIABILITY, WATER BALANCE AND RISK OF DRY SPELLS IN THE BASIN OF PARANÁ RIVER III, BRAZIL

For the agriculture context, the water balance and precipitation analysis are essential for planning and decision-making. The objective of this work was to carry out the analyse of pluviometric variability, climatological water balance (CLIMWB) and the occurrence of dry spells in the Basin of Paraná River III, Paraná State, Brazil. For this purpose, 43 meteorological data from 43 stations, from 1976 to 2018, were used. Geoprocessing techniques were applied to regionalize rain data, in addition to box plots and probabilities to analyze precipitation and the occurrence of dry spells. A signficant precipitation variability was identified with regional and temporal discrepancies. Despite the Basin of Paraná River III is a rainy region in the Paraná State, the occurrence of dry spells was identified. Periods of 20 to 30 days with no precipitation event in the region they were also frequent, due the annual occurrence risks ranging from 80 to 50 %, respectively. The risk of 40 consecutive days without rain has already proved to be nil. The water balance exhibited sufficient values for agricultural practice with water surplus along the Basin. However, when analyzing dry years, a water deficit of more than 100 mm in a single month can occur.


INTRODUCTION
Global climate change is a scientific consensus due the climatic evidences and several studies published across the world (MARENGO, 2009). These changes have caused atypical and global extreme events, as occurrences of extreme temperatures, heat waves, drought events, an increase in the heights of daily precipitation, culminating in severe heat or cold risks, in addition, severe storms and floods. The warmest periods in the entire planet, were seen successively in the last three decades, and for precipitation the uncertainties are significant, where the trends point to increase or decrease, according to the geographic position (IPCC, 2014). Rainfall has a key importance for characterizing the climate, interfering with various aspects and activities related to society. Excess or scarcity of rainfall can cause significant losses for environment and for economics (DINIZ, 2013;DA SILVA DIAS, 2014;CALDANA et al., 2018).
The water availability of a region can be quantified by the climatic water balance (CLIMWB).
This method exhibits the seasonal variations of water necessities and deficiencies, thus, it is crucial for agricultural planning and decision-making (SOUZA et al., 2013). The studies should be developed with the purpose to aid and improve the relations soil -plant -atmosphere, allowing an adjustment of agricultural practices to the climatic conditions, in addition, exhibiting agricultural applications for the definition of agroclimatic risk zoning, irrigation projects, hydrology, reservoir sizing, drainage, among others (PEREIRA et al., 2002;DANTAS et al., 2007).
For urban space, natural disasters can occur as a consequence of extreme scenarios of precipitation and are defined as nature phenomena which can change the landscape and geographical space (CALDANA et al., 2018). The floods are the most common among hydroclimatological events, with 59 % of the records, thus, the key cause of climatic catastrophes. These climatic events affect the humanity due the way societies organize considering the rhythm and variability of the atmospheric system (VICENTE, 2005, MARCELINO, 2007. Severe atmospheric instabilities generate large amounts of rain in a short time, in addition, the low drainage of urban spaces and the occupation of inadequate areas are responsible for such events (CIDADE et al., 2013;FREIRE et al., 2014;GRIGOLETTI et al. 2018). In cases of drought, low relative humidity can cause damage to human, animal and crops (SATO et al., 2003).
The Paraná state, South of Brazil, is located in an area of climatic transition, with wide variations in altitude and latitude, thus conditioning large discrepancies in the climate. The purpose of this work was to analyze rainfall variability, water balance and the occurrence of dry spells in the basin of Paraná River III.

Climate Variability
The hydrographic basin of Paraná River III is located in a Cfa climate, which means that it has a humid subtropical climate according to the Köppen climate classification (NITSCHE et al., 2019). This is characterized by the absence of drought seasons and by summers with higher average temperatures. This climate is controlled by air masses from tropical regions (the Atlantic Tropical Mass and the Continental Tropical Mass) and the Atlantic Polar Mass. In addition, the Continental Equatorial Mass can influence the Cfa climate zone during the summer season. Due to the temperature and humidity differences in these climatic masses, the area of the basin is a convergence zone for these climatic front systems, particularly in the winter season period (NITSCHE et al., 2019). The spatialization of these data was carry out by interpolation, which is an effective method for spatial visualization of climate data. This was done using isohyets and/or by spatially filling the values through adjusted regression statistics, using the inverse distance weighted spatial interpolation algorithm (LEM et al., 2013). The maps were created using QGIS software.
We used Box Plot graphs or box diagrams to complement the analysis of rainfall variability and detection of climate extremes. The key resource obtained in its use is to provide a quick view of the data distribution. If the distribution is symmetrical, the box is balanced with the median positioned in the center of the box. For asymmetric distributions, there is an imbalance in the box with respect to the median (SILVESTRE et al., 2014). We created the graphs using the Statistica® software.
The Box plots represent five classifications of values. Discrepancies are divided into outliers (values above what is considered maximum, however not extreme) and extremes, with any values greater than Q3 + 1.5 (Q3 -Q1) or less than Q1 -1.5 (Q3 -Q1). The highs and lows are considered the highest values in the series, however they are not extreme or outliers. Inside the box, three quartiles are classified with 25 % of the data each, in addition to the median value, equivalent to the second quartile, or 50 % of the data (LEM et al., 2013;SCHNEIDER and DA SILVA, 2014).
For the analysis using Box plot, we used data from the municipalities of Matelândia, Santa Lucia, Terra Roxa and Toledo, taking the different rainfall values as a parameter from the area under the study (Figure 4).
In order to identify the probability of occurrence of dry spell, the frequencies of the number of consecutive days with precipitation equal to or less than 1 mm. day -1 or 10 mm. day -1 , with duration of at least 10 days were determined. Frequency analyzes were carried out using mobile decendial (1-10 / 01, 2-11 / 01, 3-12 / 01, and subsequently). This procedure avoids the omission of consecutive December periods without rain that can occur when considering only the ten days 1-10, 11-20 and 20-30 of each month. Only values above than or equal to 1 mm we considered as rainfall.
In the data series, we verified the occurrences of consecutive periods without rain lasting 20 days or more and the longest dry period of each year. Rainy days were those in which the precipitation was equal to or greater than 1.0 mm. A descriptive analysis of the mean of the events that had 20 days or more without rain was also made, in addition to the error and standard deviation. The frequencies of the dry periods and the adjustment of the largest annual dry period to the distribution of extreme values were calculated, whose probability density function f (X) and the cumulative function of probability F (X) have the following format (ASSIS et al., 1996;COSTA et al., 2009): We accepted "X" as the random variable, α is the parameter that controls the position of the curve on the abscissa axis and β is the parameter that controls the dimensions of the curve, given a constant shape. We also used the Lieblein method to estimate the α and β parameters and the Kolmogorov-Smirnov adherence test, as described by Assis (1996) Where Tn is the average temperature of month n (n = 1 is January, n = 2 is February, etc.) in °C, and I is an index that expresses the heat level of the region.
The value of I depends on the annual temperature cycle, integrating the thermal effect of each month, and is calculated using the formula The exponent "a", being a function of I, is also a regional thermal index, and is calculated using the expression 0.49239 1.7912 x 10 7.71 x 10 6.75 x 10 .
The PET value represents the total monthly potential evapotranspiration that would occur under the thermal conditions of a standard 30-day month, and with a 12-hour photoperiod (N) each day. Therefore, PET should be corrected for N and the number of days in the period.

RESULTS AND DISCUSSIONS
Precipitation in the basin of Paraná River III exhibited a regional discrepancy ( Figure 02). We identified the lowest average annual rainfall was 1,550 mm, in the Northern of the Basin (Guaíra), while the highest average values were verified in the municipality of Cascavel region and in part of the Southern region of the basin with precipitation ranging from 2,050 to 2,125 mm.  to 960 mm, respectively. The latter was the lowest annual rainfall verified among the analyzed stations. The interval between Q1 and Q3 in Terra Roxa was from 1,798 to 1,385 mm, respectively.
Even with the discrepancy between the locations, the basin of Paraná River III was rainy by the annual average, not being a limiting factor for the crops production, however its variability and regional distribution can be limiting aspects.
Monthly, different from other regions of Paraná state that present the rainiest month in January (CALDANA et al., 2018;CALDANA et al., 2019;NITSCHE, 2019), the basin presented the month of October with the highest pluviometric heights and the highest monthly median (Figure 04).
The Matelândia station showed great variation in the month of May, with an oscillation from 557mm in 1992 to 2.5mm in 2006. Only August 1999 was verified without rain at this point. Were identified seven discrepancies and one extreme, concentrated mainly in the winter and spring months.
Was verified in municipality of Santa Lucia, the highest monthly rainfall, which occurred in June 2014 with 608mm. There were five discrepancies and two extremes in this location.
Comparatively, there was a difference in the distribution of rainfall between the municipalities of Matelândia and Santa Lucia, as the month of December was rainier than November, in the latter, and February was rainier than May, while in municipality of Terra Roxa there was the smallest monthly variation between Q1 and Q3, mainly in the month of July with a 43mm fluctuation. In contrast, there was a greater concentration of outliers (13)     It is should be noted that even considering the Basin of Paraná River III as a rainy location, the risks of dry days reach at least 20 days, ranging from 80 to 90 % of risks.
The risk of more than 35 days without rain is practically inexistent, while from 40 consecutive days without precipitation, it is zero. For the context agriculture and crops productions, 30 days without rain can be determine the crop failure, in the field. Species, mainly fruit and arboreal species are more resistant due they have an effective depth of the deepest roots (CALDANA et al., 2020). On the other hand, it is should be noted that also dry periods can be favorable for fruit species, due the development of the profitability of fruit commercialization.While for the climatological water balance of the region (Figure 08), it was identified that every month at the Cascavel station, São Miguel do Iguaçu and Toledo presented water surplus, demonstrating that for the annual average, for species cultivated with deep roots of at least 30 cm, there is no water stress caused by disability. However, atypical years, with rainfall well below the average, as in 1979 (Figure 02), may present discrepancies in the water balance.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License

CONCLUSIONS
• Precipitation in the basin of Paraná River III exhibited a local climate discrepancy. The annual average variability verified was from 1,550 mm in the far Northern of the basin to 2,125 mm in the East. The topography, altitude and formation of atmospheric instabilities justify the differences observed in this study.
• Monthly, the basin necessities of appropriate agricultural planning management and other agricultural activities. The precipitation may variate, in the same month, from 2.5 to 557 mm.
• Even though it is a rainy region of the Paraná state, the occurrence of dry spells is significant during certain times of the year, ranging from zero to more than 60 % of risk of occurrences in the basin of Paraná river III. The periods with the lowest risk verified are October, followed by December 10 to January 5.
• Periods of 20 to 30 days without rain in the region were also frequent, with a risk of annual occurrence ranging from 80 to 50 %, respectively. However, the risk of 40 consecutive days without rain has already proved to be insignificant.
• The annual average water balance exhibited sufficient values for agricultural practice with water, in the basin of Paraná River III. However, when analyzing dry years, a water deficit of more than 100 mm in a single month can occur, demonstrating the importance of this study and agricultural planning and decision-making.