Rainfall and Extreme Drought Detection: An Analysis for a Potential Agricultural Region in the Southern Brazilian Amazon

Abstract

In recent decades, the main commercial crops of Mato Grosso, such as soybeans, corn, and cotton, have been undergoing transformations regarding the adoption of new technologies to increase production. However, regardless of the technological level, the climate of the region, including the rainfall regime, can influence the success of crops and facilitate, or not, the maximum production efficiency. This study aimed to define the behavior of the variability in monthly and annual rainfall and its probability of monthly occurrence and calculate the drought index for the northwestern region of Mato Grosso, in the southern region of the Brazilian Amazon. To carry out the study, daily rainfall records were collected, calculating the totals for each month of the historical series for each of the four National Water and Sanitation Agency (ANA) rain gauge stations, Aripuanã (1985–2020), Colniza (2001–2020), Cotriguaçu (2004–2020), and Juína (1985–2020), representing the northwestern region. The annual distribution of rainfall during the periods studied ranged from 1376.2 to 3017.3 mm. The monthly distribution indicated a typical water shortage in the months of June, July, and August. The probability of rainfall near the average for each month was more than 50%. The monthly SPI-1 index revealed a total of 56 months affected by very dry events and 34 extreme dry events. The annual SPI-12 index pointed to seven very dry years and five extremely dry years. Therefore, the region presented high rainfall rates in most years; however, a significant process of drought was also observed, including in rainy months, which are the periods with the greatest demand for the main agricultural crops.

1. Introduction

Brazil’s gross domestic product had a 25% share of agribusiness in 2022, with this segment being leveraged mainly by the record harvest in the field in the respective year, with a significant share of soy and corn commodities [1]. In the same period, Brazil exported USD 334 billion in products, in which soybeans and corn together accounted for 17.6% of the volume [2]. Mato Grosso’s performance in terms of agricultural production makes the state an important ally in the country’s trade balance [3].

The state of Mato Grosso is the third largest among the states of Brazil; the state leads the national production of soybeans, corn, and cotton due to the large expanse of agricultural land associated with high production technology, thus contributing significantly to the country being an important supplier of grain and fiber in the world market [4]. The northwestern region of Mato Grosso is one of the locations that has areas available to further increase the state’s agricultural production. It has seven municipalities, all of which are considered agricultural frontiers, i.e., where the economy is based on agricultural expansion, with soybeans and corn being the most prominent crops [5].

In recent years, there have been significant increases in forest fires and frequent reductions in rainfall for the proper development of agricultural crops [6,7], including in the state of Mato Grosso. The El Niño and La Niña phenomena have been more frequent in the Amazon region, altering the rainfall regime and directly affecting the intensity of droughts and floods [8]. Regions with transitions between biomes, such as the Cerrado and the Amazon Rainforest, are more susceptible to climate change, which justifies studies that seek to understand the climatic fluctuations of these locations [9].

The influence of biophysical characteristics, such as relief and vegetation, among others, tends to cause variations in the rainfall indices between the internal regions of the state [10,11]. In addition, global climate change may intensify the length of the dry season and the spatial and temporal distribution of rainfall, which will threaten agricultural yields [12,13,14,15]. Regional knowledge about the distribution of rainfall in monthly and annual periods, as well as its probability of monthly occurrence, according to the history of accumulated rainfall (climatology), contributes to the implementation of good agricultural practices that aim at gains in productivity [16].

Agricultural drought is caused by an imbalance due to excess evapotranspiration or a lack of rainfall for a certain time in a location [17]. Although drought cannot be avoided, it is agreed that quality information, through the application of drought indices, which characterize the anomaly, is essential for planning in order to reduce its impact [17]. The standardized precipitation index (SPI) is based on the amount of rainfall and is used to assess the volume of water received within a certain time scale [18,19]. This index has the advantage of characterizing not only drought events but also excess rainfall [20].

This study addresses a gap in the literature regarding the use of a drought index applied in a regionalized way to a location with agricultural crops. As a contribution, the study seeks to provide an analysis, especially of drought events in the Amazon biome, in order to improve the monitoring and management strategies applied to agriculture and forest preservation.

Considering the importance of knowledge regarding rainfall behavior in agricultural planning, this study aimed to characterize the annual and monthly variability of rainfall in the northwestern region of Mato Grosso (southern region of the Brazilian Amazon) and define its different levels of probability of occurrence for monthly periods. In addition, we sought to categorize the intensity of the droughts that occurred in the intervals studied, based on the standardized precipitation index (SPI).

2. Materials and Methods

2.1. Study Area

The experimental area encompasses the northwestern region of Mato Grosso, which is part of the legal Brazilian Amazon and has an area of approximately 108,013 km². This region is located between latitudes 8°30′ S and 12°30′ S, longitudes 58°00′ W and 62°00′ W [21,22]. The local climate, according to the Köppen classification, is of type Aw, i.e., it has an average temperature higher than 18 °C in the coldest month, low temperatures are absent, and the annual rainfall is higher than the annual potential evapotranspiration [23]. The predominant soil is of the dystrophic Red–Yellow Argisol type [24].

2.2. Acquisition and Processing of Data

The data were extracted from four meteorological stations belonging to the National Water and Sanitation Agency (ANA) [25,26]. The reduced number of weather stations is justified by the low density of rain gauges in the area. In addition, some of the few units that exist had considerable temporal data gaps. To carry out the study, the daily rainfall records were collected, and the totals for each month were calculated. The recorded intervals available in the histories of each station were as follows: Aripuanã station—1985 to 2020 (36 years), Colniza station—2000 to 2020 (21 years), Cotriguaçu station—2004 to 2020 (17 years), and Juína station—1985 to 2020 (36 years). For the calculation of climatological normals, the World Meteorological Organization (WMO) recommends records of 30 years; however, according to [27,28], the WMO recommends the calculation of provisional climatological normals (PCNs), which can also be simply called averages, for minimum intervals of 10 years for stations that do not have long historical series of atmospheric variables, as is the case for the Colniza and Cotriguaçu stations (Figure 1).

Table 1. Rainfall data of the stations in the northwestern region of Mato Grosso—indicator (ID), latitude (°), longitude (°), altitude (m), and average annual rainfall (mm).

ID	Station and Code	Municipality	Lat. (°)	Long. (°)	Alt. (m)	Average Annual Rainfall (mm)
1	Aripuanã_C_01058005	Aripuanã	−10.59	−58.87	255	2080
2	Colniza_C_00958002	Cotriguaçu	−9.46	−58.22	224	2299
3	Cotriguaçu_C_00958004	Cotriguaçu	−9.91	−58.56	258	2279
4	Juína C_01058003	Juína	−11.41	−58.72	331	2027

Source: [29].

contains the codes, the municipality in which the station is installed, the geographical location, and the average amount of annual rainfall.

After the tabulation of the data, the statistical parameters of the mean and standard deviation were determined [30]. The missing or discontinued information in certain periods was completed using the mean of that month from the studied interval.

The choice of time scale was based on the type of impact being studied. In the case of SPI-1, the parameter can be used, for example, to analyze the effect of agricultural stress, while SPI-12 is more associated with studying the levels of water bodies and reservoirs, among others [31]. The respective scales can provide a framework for agricultural planning that involves, respectively, monitoring the vegetative growth of plants and irrigation projects.

The annual and monthly rainfall distributions were calculated using the monthly rainfall records of each station [32], using the Microsoft Excel datashed. The annual values were obtained from the sum of the amount of rainfall for each year within the analyzed intervals. The monthly values obtained were based on the averages ($X$) of each month over the years in the temporal period of each station. Then, the calculation of the standard precipitation index (SPI) was performed.

The SPI evaluates rainfall anomalies for all environments using records from several years on a given time scale [33]. For this research, the SPI-1 and SPI-12 were estimated, respectively, for monthly and annual scales. All operations were performed via the SPEI package, available in the Comprehensive R Archive Network (CRAN—https://CRAN.R-project.org/package=SPEI accessed 20 January 2023) [34] and found in the RStudio program, version 4.1.3 [35]. For the determination of the index, initially, the calculation of the incomplete gamma distribution was used:

$$f \left(x\right)\frac{{\left(\frac{x}{\beta }\right)}^{\alpha -1 }\exp \left(-\frac{\mathrm{x}}{\mathsf{\beta }}\right)}{\beta \mathsf{\Gamma } \left(\alpha \right)}$$

(1)

where

β > 0, (β) = scale parameter (mm);

α > 0, (α) = shape parameter (dimensionless);

x > 0, (α) (x) = amount of rainfall (mm);

Γ (α) = gamma function.

Thus, the gamma function is presented as follows:

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

(2)

All parameters, as well as the gamma probability density function, were adjusted for the frequency distribution of the accumulated rainfall used in the study. The α and β measurements of the function were estimated for each station analyzed. The maximum likelihood method (MLM) was used to estimate the parameters [36].

$$\alpha =\frac{1+\sqrt{1+4A/3}}{4A}$$

(3)

$$\beta =\frac{\overline{x}}{\alpha }$$

(4)

$$A=\mathrm{Ln}\left(\overline{x}\right)-\frac{1}{n}{{\displaystyle \sum }}_{i=1}^{n}ln\left(x_i\right)$$

(5)

where

$\overline{x }$= arithmetic mean of the rainfall (mm);

$Ln$ = Nepierian logarithm;

$n$ = number of observations.

The results of the equation were used to determine the cumulative probability of rainfall for a defined period [32] at the levels of 10%, 25%, 40%, 50%, 60%, 75%, and 90%. Therefore, the gamma probability function is

$$F \left(x\right)={{\displaystyle \int }}_0^{x}f\left(x\right)dx=\frac{1}{\mathsf{\Gamma }\left(\mathrm{a}\right){\beta }^{a }}{{\displaystyle \int }}_0^x{x}^{a-1}{e}^{-\frac{x}{\beta }}dx$$

(6)

By replacing the value of $t=\frac{x}{\beta }$ in Equation (6), this can be reduced to

$$F \left(\beta t\right)=\frac{1}{\mathsf{\Gamma }\left(\mathrm{a}\right){\beta }^{a }}{{\displaystyle \int }}_0^{\beta t}{t}^{a-1}{e}^{-t}dt$$

(7)

The samples contain rainfall data equal to zero; however, the incomplete gamma function does not define null values, so the cumulative probability is determined by the equation

$$f\left(x\right)=P_{0 }+\left(1-P_0\right)G\left(x\right)$$

(8)

where

$P_0$= probability of occurrence of null values (mm);

$G\left(x\right)$ = subjective cumulative distribution, the parameters of which are assumed for rainy days.

The function F(x) was converted into a normal distribution for the random variable Z with the mean as zero and variance of one, whose variable in question is equivalent to the SPI value [37].

The definition of the sample size is given by

$$P_0=\frac{m}{n+1}$$

(9)

where

$m$ = order number of values of zeros in a series;

$n$ = sample size.

An approximate calculation, optimizes the SPI or Z value by transforming the cumulative probability into a normal distribution:

$$Z=SPI=- (t-\frac{c+c_1^t+c_2{t}^2}{1+d_1^t+d_2{t}^2+d_3{t}^3})$$

(10)

$$Z=SPI=+ (t-\frac{c+c_1^t+c_2{t}^2}{1+d_1^t+d_2{t}^2+d_3{t}^3})$$

(11)

$$t=\sqrt{\ln \left[\frac{1}{{\left(\left(P9x\right)\right)}^2}\right]},\mathrm{for}0<\mathrm{P}(\mathrm{x})\le 0.5$$

(12)

$$t=\sqrt{\ln \left[\frac{1}{1-{\left(\left(P9x\right)\right)}^2}\right]},\mathrm{for}0.5<\mathrm{P}(\mathrm{x})\le 1$$

(13)

where $C_0=2.515517;C_1=0.802853;C_2=0.010328;d_1=1.432788;d_2=0.189269;\mathrm{and}{d}_3=0.001308$.

Table 2. Classification of dry and humid periods using the SPI.

SPI	Category
≥2.00	Extremely humid
1.99 to 1.50	Very humid
1.49 to 1.00	Moderately humid
0.99 to −0.99	Close to normal
−1.00 to −1.49	Moderately dry
−1.50 to −1.99	Very dry
≤−2.00	Extremely dry

Source: [33].

indicates the numerical ranges of the standardized precipitation index and their respective classification categories.

3. Results

3.1. Annual Distribution of Rainfall

In 36 years, the Aripuanã station (

Table 3. Annual distribution of rainfall at the Aripuanã, Colniza, Cotriguaçu, and Juína stations located in the northwestern of the state of Mato Grosso.

Years	Annual Amount of Rainfall (mm)
	Aripuanã	Colniza	Cotriguaçu	Juína
	1985–2020	2000–2020	2004–2020	1985–2020
1985	2304.7			2057.4
1986	2427.2			1877.3
1987	1719.4			2257.9
1988	2042.7			2185.5
1989	2128.6			2162.4
1990	1764.7			1741.7
1991	1898.8			2033.2
1992	1821.9			1961.0
1993	1966.7			1891.8
1994	2418.1			2418.3
1995	2388.4			2141.0
1996	1923.8			2535.1
1997	2235.9			2093.6
1998	2069.3			2136.7
1999	1978.4			1947.1
2000	1790.1	2361.1		1936.4
2001	1916.6	2563.8		2248.7
2002	2078.0	1601.1		1376.2
2003	2168.0	1898.0		2461.3
2004	2079.9	2141.3	2136.3	1893.9
2005	2036.4	2086.6	1931.1	2061.3
2006	2349.2	2332.7	2388.1	2628.0
2007	1790.8	2096.7	2094.6	1833.8
2008	2208.6	2605.6	2296.4	1953.2
2009	2034.7	2478.5	2395.2	1908.9
2010	2373.1	1879.0	2039.9	2126.2
2011	2524.2	2121.6	2406.2	1732.1
2012	1598.4	2379.9	2363.8	1433.0
2013	2456.6	3017.3	2718.3	2824.4
2014	1795.3	2715.7	2482.4	1654.8
2015	2050.7	1974	1832.8	1766.1
2016	2503.7	2269.8	2154.9	1889.8
2017	2059.2	2335.5	1977.9	1488.3
2018	2046.1	2829.6	2723.4	2458.7
2019	1914.5	2609.6	2441.0	2015.4
2020	2013.9	1973.9	2354.2	1828.8
Mean	2079.9	2298.6	2278.6	2026.6
Standard deviation	240.9	350.4	256.2	320.9

) presented an average of 2079.9 mm of rainfall and a standard deviation of 240.9 mm, which was the smallest variation in the amount of rainfall amongst the stations. Between the years 2000 and 2020, the Colniza station (Table 3) registered an average of 2298.6 mm of rainfall, with a standard deviation of 350.4 mm, thus exhibiting the highest variability in rainfall when compared to the other stations. In the period between 2004 and 2020, the Cotriguaçu station (Table 3) presented mean precipitation of 2278.6 mm and a standard deviation of 256.2 mm. It was observed that 59% of the cycles (10 years) exceeded the mean, while 41% (7 years) did not exceed the mean. The value of 2718.3 mm stands out as the maximum amount of rainfall in 2013, while the minimum volume was 1832.8 mm in 2015, a difference of 885.5 mm, which was the smallest variation between the evaluated locations. The Juína station (Table 3) reached a mean of 2026.6 and a standard deviation of 320.9 mm in the interval of 36 years. In the historical series, 47% of the years (17 years) exceeded the mean; in contrast, 53% of the cycles (19 years) were below the mean magnitude. The maximum and minimum amounts, respectively, were 2824.4 mm in 2013 and 1376.2 mm in 2002, with a variation of 1448.2 mm.

3.2. Monthly Distribution of Rainfall

The monthly average rainfall for the four stations (Figure 2) that represent the northwestern region of Mato Grosso indicate that the months of June, July, and August seasonally presented the lowest rainfall indices, with the lowest value of 3.2 mm and a standard deviation of 6.7 mm for the month of July in Cotriguaçu. The highest mean accumulation of 44.1 mm and standard deviation of 37.6 mm occurred in August at the Colniza station (Figure 2B).

3.3. Monthly Rainfall Probability Levels

The values for the probability of occurrence of rainfall per month for the calculation of the gamma function and the data of the beta-scale parameter, per station, are presented in

Table 4. Values of α, β and probability of rainfall for the municipality of Aripuanã, state of Mato Grosso.

Aripuanã Station
Month	$\overline{\mathit{x}}$	S²	α	β	Probability Levels
					90%	75%	60%	50%	40%	25%	10%
Jan	351.6	65.2	29.0	12.0	437.3	393.1	364.3	347.6	331.4	305.7	271.1
Feb	330.7	96.2	11.8	27.9	458.3	389.5	345.9	321.4	298.0	261.8	214.9
Mar	331.0	106.5	9.6	34.2	472.6	395.2	346.8	319.6	293.9	254.3	203.9
Apr	199.0	82.9	5.7	34.6	310.0	247.0	208.6	187.6	168.0	138.6	102.7
May	60.0	44.7	1.8	33.4	119.7	81.3	60.1	49.3	39.9	27.2	14.3
Jun	9.5	15.5	0.3	25.3	27.2	11.8	5.5	3.2	1.7	0.4	0.0
Jul	6.7	11.4	0.3	19.1	19.5	8.2	3.7	2.1	1.0	0.2	0.0
Aug	12.3	16.0	0.5	20.9	32.2	16.6	9.4	6.4	4.1	1.7	0.3
Sep	77.3	41.1	3.5	21.9	132.5	99.7	80.4	70.1	60.7	47.0	31.4
Oct	152.8	59.5	6.5	23.2	232.4	187.7	160.3	145.2	131.0	109.6	83.2
Nov	238.7	84.4	7.9	29.9	351.4	289.0	250.4	228.8	208.6	177.7	138.9
Dec	309.8	101.3	9.3	33.1	444.7	370.9	324.7	298.9	274.4	236.9	189.1
Annual	2079.9	240.9	74.5	27.9	2394.0	2236.9	2132.0	2070.6	2010.3	1912.7	1777.7

,

Table 5. Values of α, β and probability of rainfall for the municipality of Colniza, state of Mato Grosso.

Colniza Station
Month	$\overline{\mathit{x}}$	S²	α	β	Probability Levels
					90%	75%	60%	50%	40%	25%	10%
Jan	358.6	91.6	15.3	23.3	479.9	415.5	374.3	350.9	328.4	293.4	247.4
Feb	368.2	116.1	10.0	36.6	522.7	438.5	385.7	356.0	327.9	284.6	229.3
Mar	366.1	67.3	29.5	12.3	454.6	409.0	379.2	361.9	345.3	318.7	282.8
Apr	221.9	110.1	4.0	54.6	369.4	283.1	231.6	203.9	178.5	141.2	97.5
May	95.5	62.1	2.3	40.3	178.7	127.1	97.8	82.5	68.9	49.9	29.4
Jun	24.1	22.4	1.1	20.8	53.5	33.3	22.7	17.6	13.3	7.9	3.2
Jul	14.2	19.8	0.5	27.4	38.3	19.0	10.3	6.7	4.1	1.5	0.2
Aug	44.0	36.2	1.4	29.8	92.2	60.4	43.2	34.6	27.2	17.6	8.4
Sep	91.1	44.5	4.1	21.8	150.8	115.9	95.1	83.9	73.6	58.4	40.6
Oct	169.1	57.9	8.5	19.8	246.4	203.8	177.3	162.5	148.6	127.3	100.3
Nov	257.1	91.4	7.9	32.5	379.0	311.5	269.6	246.3	224.4	191.0	149.0
Dec	288.2	90.6	10.1	28.5	408.8	343.1	301.9	278.8	256.9	223.1	179.9
Annual	2298.6	350.4	43.0	53.4	2757.5	2524.2	2370.3	2280.8	2193.6	2053.6	1862.5

,

Table 6. Values of α, β and probability of rainfall for the municipality of Cotriguaçu, state of Mato Grosso.

Cotriguaçu Station
Month	$\overline{\mathit{x}}$	S²	α	β			Probability Levels
					90%	75%	60%	50%	40%	25%	10%
Jan	397.1	94.5	17.6	22.4	522.0	456.2	413.8	389.7	366.5	330.0	281.8
Feb	349.7	97.2	12.9	27.0	478.7	409.5	365.5	340.7	317.0	280.1	232.2
Mar	358.3	94.1	14.4	24.7	483.0	416.5	374.1	350.1	327.1	291.1	244.2
Apr	213.0	88.1	5.8	36.4	330.9	264.1	223.4	201.0	180.2	148.9	110.7
May	60.9	22.5	7.3	8.3	91.0	74.3	63.9	58.2	52.8	44.6	34.4
Jun	11.5	14.4	0.6	18.1	29.7	15.8	9.2	6.3	4.1	1.8	0.4
Jul	3.2	6.7	0.2	14.2	9.6	3.1	1.0	0.4	0.1	0.0	0.0
Aug	24.3	23.2	1.0	22.2	54.8	33.7	22.7	17.4	13.0	7.6	3.0
Sep	76.3	29.4	6.7	11.3	115.6	93.6	80.0	72.5	65.5	54.9	41.8
Oct	161.2	59.1	7.4	21.7	240.1	196.2	169.1	154.0	139.9	118.4	91.5
Nov	277.3	114.9	5.8	47.6	431.0	343.9	290.8	261.6	234.5	193.7	144.0
Dec	345.3	76.0	20.6	16.7	445.7	393.1	359.2	339.7	321.0	291.4	252.1
Annual	2278.6	256.2	79.0	28.8	2612.5	2445.8	2334.4	2269.0	2204.8	2100.9	1957.0

and

Table 7. Values of α, β and probability of rainfall for the municipality of Juína, state of Mato Grosso.

Juína Station
Month	$\overline{\mathit{x}}$	S²	α	β	Probability Levels
					90%	75%	60%	50%	40%	25%	10%
Jan	355.9	100.6	12.5	28.4	489.4	417.7	372.2	346.5	322.0	283.9	234.6
Feb	311.7	83.7	13.8	22.5	422.7	363.4	325.6	304.2	283.8	251.9	210.3
Mar	297.9	103.9	8.21	36.2	436.4	359.9	312.3	285.8	260.9	222.8	174.8
Apr	196.1	91.7	4.5	42.9	319.0	247.9	205.2	182.0	160.6	129.0	91.4
May	55.9	39.7	1.9	28.2	109.0	75.3	56.5	46.8	38.3	26.7	14.7
Jun	10.5	16.6	0.4	26.2	29.8	13.3	6.5	3.8	2.1	0.6	0.0
Jul	6.7	14.9	0.2	33.2	20.3	6.1	1.8	0.7	0.2	0.0	0.0
Aug	25.5	27.6	0.8	29.8	61.1	35.3	22.5	16.5	11.7	6.1	1.9
Sep	80.0	47.6	2.8	28.3	143.9	105.1	82.6	70.8	60.2	45.0	28.2
Oct	146.5	79.8	3.3	43.5	253.6	189.7	152.2	132.2	114.1	87.9	57.9
Nov	214.5	91.1	5.5	38.7	336.5	267.0	224.8	201.7	180.3	148.1	109.1
Dec	324.9	83.5	15.1	21.4	435.5	376.7	339.2	317.8	297.4	265.4	223.5
Annual	2026.6	320.9	39.8	50.8	2447.2	2232.9	2091.7	2009.7	1929.9	1801.9	1627.7

. The percentage found by the authors is similar to what occurs in Cotriguaçu for the month of July, where, with the same percentage, there is a possibility of rainfall amounting to 9.7 mm. The other municipalities exceed this amount for the aforementioned percentage.

3.4. Drought Index (SPI)

The monthly SPI-1 index for the Aripuanã station (Figure 3) reveals that, in 36 years, there were 281 events that were close to normal (67%), 35 were moderately dry (8%), 24 were very dry (6%), and nine were extremely dry (2%).

The SPI-1 index for the Colniza station (Figure 4) shows that, during the 21-year period, there were 152 events that were close to normal (63%), 35 were moderately dry (14%), eight were very dry (3%), and six were extremely dry (3%). We highlight November and December 2002 as extremely dry months, followed by January and February 2003, January 2011, and October 2016.

The SPI-1 index for the Cotriguaçu station (Figure 5) shows that, between 2004 and 2020, there were 132 events close to normal (68%), 16 were moderately dry (8%), seven were very dry (4%), and seven were extremely dry (4%). The extremely dry events happened in the interval from April to October 2005.

For the SPI-1 index, for the Juína station (Figure 6), during the historical series of 36 years, there were 269 events close to normal (64%), 44 were moderately dry (10%), 17 were very dry (4%), and 12 were extremely dry (3%).

4. Discussion

4.1. Annual Distribution of Rainfall

The annual rainfall in the Amazon area of Mato Grosso showed mean variations from 1392.9 to 2110.9 mm in the Juruena river basin, which covers part of the northeastern region. During the interval, 36% of the years (13 years) were above average, while 61% (22 years) were below average. The maximum amount (2524.2 mm) occurred in 2011 and the minimum (1598.4 mm) occurred in the following year, a difference of 925.8 mm.

The amount of rainfall in the state of Mato Grosso varies on average from 1200 to 2200 mm per year, with the northern region receiving the highest volumes [10]. In the studied period, 52% of the years (11 years) were above average, while 48% (10 years) remained below the mean value. The maximum and minimum volumes presented, respectively, were 3017.3 mm in 2013 and 1601.1 mm in 2002, i.e., a variation of 1416.2 mm. This station exhibited the highest maximum rainfall volume that occurred in a year between stations.

The Juína station stands out as having the greatest variation between the minimum and maximum volume of rainfall, as well as presenting the year with the lowest amount of rainfall among the evaluated stations, which was 1376.2 mm in 2002. Aripuanã and Juína had the highest number of years in which the rainfall was lower than their respective averages. Ref. [38] analyzed data from 28 years of records from Juína and reported positive variations in the years 1985, 1994, 1997, and 2010, while 1993, 2004, 2007, and 2008 presented greater negative alterations, thus confirming some of the results found for this station.

4.2. Monthly Distribution of Rainfall

The highest mean accumulation of 44.1 mm and standard deviation of 37.6 mm occurred in August at the Colniza station (Figure 2B). However, this interval cannot be considered as being rainy since the monthly accumulation was less than 100 mm [39]. The reduction in rainfall in the winter period is typical in this particular annual season [40,41,42]. The dry period in the south of the Amazon is due to the influence of the South American Monsoon System (SAMS), which causes a seasonal reduction in evapotranspiration between June and July [43]. As a result, rainfall is very low in these months [44]. SAMS also influences the heavy rains in the region.

Conversely, the months with the highest indices were December, January, February, and March, and they regularly exceeded the average of 288 mm for all stations. Ref. [45] researched the humidity and drought conditions in the Midwest of Brazil and observed that these months were the wettest. Other authors have observed the aforementioned climatic characteristic in these areas of the Amazon biome [40,46,47]. The South American Monsoon System (SAMS), as in dry periods, has a strong influence on the annual rainfall cycle, causing large volumes of precipitation [43]. Some authors argue that a slight increase in rainfall from August to October marks the transition between the dry and rainy periods, while the evapotranspiration increases, which contributes to conditioning the atmosphere to intensify the rainy season [12,48,49]. The months that are considered the wettest (wet season) in the Amazon region are between November and April; therefore, May to October is the period that receives the lowest volume of rainfall [50]. In the four stations studied herein, these ranges accumulated an average total of 1815.9 and 355.0 mm, respectively.

4.3. Monthly Rainfall Probability Levels

The probability of occurrence of rainfall per month can help in the dimensioning of irrigation systems, as an alternative to using the average rainfall, which can lead to the underestimation of the water requirements to values below a 50% probability [51]. However, the authors also stress that the most reliable levels for the implementation of irrigation projects are between a 75% and 80% probability of rainfall, to ensure the efficiency of the system and, in this way, reduce the possibility of a water deficit for the crop.

In the months of September to May, the four meteorological stations showed identical percentages, with a probability exceeding 50% of rainfall being close to the average of the historical series of each month analyzed. In the research conducted by [52], who analyzed the distribution and probability of rainfall in Northern Mato Grosso, the authors argued that the probability of a large volume of rainfall in January, February, and March exceeds 50%. It is considered that this is the period with the greatest water requirements for the main agricultural crops in Mato Grosso [53]; therefore, the rainfall regime of this region was in accordance with the demands of the crops.

The months of June, July, and August fall within the period of greatest water scarcity during the year. However, the probability of rainfall close to the averages for these months was more than 60%. Most of the stations showed a rainfall index close to or equal to zero in June and July, but the probability of a zero index is less than 50%.

Studies on the climate in the municipality of Sinop, located in the Brazilian Amazon, have shown that the probability of rainfall below 10 mm in June, July, and August is 90% [54]. Supplemental irrigation in crops such as corn, beans, and coffee provides adequate soil water conditions during the crop cycle at times when water is needed [55] in dry months or when warm periods occur.

4.4. Drought Index (SPI)

The extremely dry periods occurred in the intervals April 1991, August 1993, March and April 1998, November 2002, February and March 2009, December 2012, and January 2013. Ref. [56] mentions that the El Niño climate phenomenon causes a strong reduction in rainfall in Northern Brazil. The La Ninã effect could also cause reduced rainfall in parts of Central–Western Brazil, including the southern part of the Amazon Rainforest [57]. According to the data [58], the interval between August 1998 and March 2001 was characterized by the predominance of La Ninã. Further reinforcing some of the results found for this station, ref. [38] identified a reduction in rainfall in the years 1998–2001 in Northern Mato Grosso. According to the authors, this was caused by a moderate La Niña event. The moderately humid climate prevailed in 28 situations (7%), 33 events were very humid (8%), and 11 extremely humid events occurred over the 36 years (2%) of this survey. The periods of March, June, July, August, and September 1986; September 1994; February, March, and November 2010; November 2011; and February 2017 stand out as exhibiting extremely humid events.

The annual index, SPI-12, indicates moderately dry conditions in the years of 1990, 1992, 2000, 2007, and 2014; very dry conditions in 1987; and extremely dry conditions in 2012. Regarding humidity, the SPI-12 index presents a moderately humid condition in the years 1994, 1995, 2006, and 2010; very humid in 2011, 2013, and 2016; and extremely humid in the year 1986. In the context of severe drought, there is a reduction in the production of grains such as corn, beans, and coffee and, consequently, this reduction intensifies in the context of extreme drought [59].

Agricultural production tends to be compromised due to reduced growth and lower photosynthesis caused by a lack of water [60]. Moderately humid situations prevailed 22 times (9%), very humid prevailed 16 times (7%), and extremely humid situations occurred twice (1%). The extremely humid climatic conditions occurred in the periods of January 2014 and January 2019.

The SPI-12 indicates moderately dry conditions in 2010 and 2013, very dry conditions 2011 and 2016, and extremely dry conditions in 2002. Ref. [61] analyzed the variability in rainfall in the south of the Amazonas state and showed equivalent results regarding drought events in the years 2010, 2015, and 2016. In relation to the rainfall conditions, events that were considered moderately humid occurred in the years 2014 and 2018, and very humid ones occurred in 2013. However, the records of this station show no extremely humid events between 2000 and 2020.

Ref. [62] observed similar conditions of severe drought in the year 2005 in the southwest of the Amazon; however, the authors relate this event to a delay in the influence of the Pacific and Atlantic Oceans on the rainfall in the Amazon.

The moderately humid climate scenario prevailed in 14 events (7%), 13 were very humid events (7%), and extremely humid events occurred three times in 17 years (1%). April and May 2013 and November 2018 stand out as being extremely humid. The SPI-12 index indicates very dry conditions in 2006 and 2017. Regarding rains, the annual index shows very humid conditions in the years 2013 and 2018. The station did not present significant events in the moderately and extremely dry and wet categories over the 17-year span.

The extremely dry months and years were those of February to July and November and December 2012; May, June, and July 2016; and December 2017. Moderately humid situations occurred in 53 moments (13%), very humid in 20 moments (5%), and extremely humid in three moments (1%). An extremely humid climate occurred in October 1994, November 1996, and January 2007.

The SPI-12 index indicates moderately dry conditions in 2014, very dry conditions in 1993 and 2017, and extremely dry conditions in 2002, 2012, and 2015. In regard to the rains during this period, events that could be considered moderately humid occurred in the years 1994, 2003, and 2018; very humid ones occurred in 1996; and extremely humid ones occurred in 2006 and 2014. Although 2014 was within the normal range according to NOAA data, the period was extremely humid most of the time in the southern part of the Amazon Rainforest. However, at the end of 2014, the warming of the Pacific Ocean occurred, associated with El Niño, which continued throughout 2015 in an intense manner [63].

Ref. [45], using the Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) data between 1985 and 2018 for the municipalities of Cáceres, Cuiabá, Diamantino, Nova Xavantina, and Poxoréo in Mato Grosso, found moderately humid anomalies for SPI-12 that were identical to those found in the Colniza and Juína stations (2018); moderately dry for Aripuanã (2000, 2007) and Colniza (2013); and very dry conditions in Juína (1993). Ref. [64], although using remote sensing data and a monthly analysis for the Brazilian Amazon, identified a lower rainfall volume that was similar to that found in the four stations in the years 2001, 2002, 2009, 2012, and 2015, as well as anomalies of surplus rainfall in the years 2011 and 2014, due to the presence of the El Niño and La Niña phenomena, respectively. SPI-6 drought events, which were studied between 2005 and 2015, were observed by [65] in the south of the state of Pará, which borders the northern portion of Mato Grosso. Also in the state of Pará, and also using the SPI-6 scale, ref. [66] detected the occurrence of 10 drought events and 12 rainy events in 32 years in their historical series. The authors’ work corroborates the present study, in which 30 events were found on the SPI-1 scale in 17 years of observation in Cotriguaçu, the closest station to the south of Pará. The larger number of events in this study is due to the fact that the monthly scale is more sensitive than the six-monthly scale used by the previous authors. However, it is noted that the balance remains in the amount of dry and wet events that occurred in the comparison between the studies.

In the four stations, the monthly SPI-1 index revealed a total of 56 months affected by very dry events and 34 by extremely dry ones. Thus, the annual SPI-12 index indicated seven very dry years and five extremely dry years. The Juína station presented the highest number of annual dry events, with these being two very dry years among the total of five and three extremely dry years among the total of seven years characterized in the studied stations. It is worth noting that Juína has a lower-density forest in its surroundings, which is a consequence of greater agricultural development in relation to the other units.

For any agricultural activity, factors such as drought, excess rainfall, the air temperature, and solar radiation can directly interfere with crop yields [67]. During the early stages of an agricultural drought, there is a reduction in the proto-plasmic layer and the plant cell wall, as well as a photosynthetic decline as a result of the stomata closing, leading to impaired growth [68]. As the drought progresses, tension increases in the tissue that distributes water and solutes (xylem), causing cavitation in the most vulnerable plant channels [69].

Mato Grosso has agribusiness as its economic base, in which excess or insufficient rainfall can compromise economic development [70]. The information provided by these indices is essential for the planning and development of decision support tools that can be applied to state and federal public policies, serving as parameters for the management of the risks of droughts in order to mitigate their impacts [6]. In this way, public authorities can establish policies targeting different sectors of society in order to raise awareness of fire prevention in forests, urban, and agricultural areas [71]. Drought data management can support water management policy measures in balancing urban use and agricultural irrigation in drought-affected regions [72].

5. Conclusions

The annual distribution of rainfall during the periods studied ranged from 1376.2 to 3017.3 mm The monthly distribution and rainfall probability levels indicated a typical water reduction in the months of June, July, and August. The probability of rainfall above the monthly average for all months was over 50%. The monthly SPI-1 index revealed a total of 56 months that were affected by very dry events and 34 extreme dry events. Therefore, the annual SPI-12 index indicated seven very dry and five extremely dry years.

Knowledge about drought events indicates that the location is vulnerable to significant droughts, even in rainy months, which represent a period with higher water demands for the vegetative and reproductive development of the main commercial crops of the state. The information obtained from the results of the study can aid decisions regarding the planning of supplementary irrigation that is applied to the different agricultural crops of the region during a possible water shortage, especially in the stages when the plants have their greatest need for water.

Extreme climatic events such as severe droughts, besides causing environmental damage, can lead to crop failures, affecting the financial balance of farmers and companies in the sector and hence affecting the revenue of the state. Furthermore, these related events tend to intensify the fires in the vast forested areas, particularly between the Cerrado, Amazon Rainforest, and Pantanal.

The study was limited by the small number and distribution of the rainfall stations, implying moderate representativeness over a vast region. For future research, we suggest using the Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) data, whose information is more distributed, in order to obtain a more precise understanding of the weather patterns in the study area.