Advertisement

Differences in precipitation and evapotranspiration between forested and deforested areas in the Amazon rainforest using remote sensing data

  • Juarez Ventura de OliveiraEmail author
  • Douglas Batista da Silva Ferreira
  • Prafulla Kumar Sahoo
  • Giordani Rafael Conceição Sodré
  • Everaldo Barreiros de Souza
  • Joaquim Carlos Barbosa Queiroz
Original Article

Abstract

With the constant deforestation observed in the Amazon region in recent decades, changes in local climates are unavoidable. Therefore, the aim of this study is to understand such climate changes by evaluating differences in precipitation (PRP), evapotranspiration (ET) and potential evapotranspiration (PET) between forested and deforested areas in the Brazilian Amazonia. Eight areas (four forested and four deforested) that maintained their land cover over a period of 15 years (2000–2014) were selected using PRODES (Legal Amazon Deforestation Monitoring Project) Digital’s remote sensing database. The PRP data were retrieved from the Tropical Rainfall Measurement Mission (TRMM, resolution of 0.25°), and the data of ET and PET were obtained from the moderate-resolution imaging spectroradiometer sensor (MODIS, resolution of 0.05°). The differences among PRP, ET and PET in each area were calculated using a regression model. The results show that ET was the only variable which presented a statistically significant difference between forested and deforested areas, mainly due to an intense decrease at the deforested sites during the dry season. Regarding PRP, a modest increase was observed on small areas, probably due to changes in local atmospheric dynamics, but on large areas decrease in PRP was noted. Nevertheless, in both cases there were no statistically significant differences between forested and deforested areas. PET presented a slightly increase value over deforested areas during the dry season, but not statistically significant as well.

Keywords

Precipitation Evapotranspiration Potential evapotranspiration Deforestation Amazonia Climate change 

Introduction

Amazonia is the largest tropical rainforest in the world with an area of approximately 7 million square kilometers spans nine countries in South America and is host to thousands of animal and plant species, many of which are endemic to the region. The largest forested area is present in the Brazilian territory, more precisely in 9 of its 27 states, and is known as the Brazilian Legal Amazon (BLA). There are several factors that can be attributed to deforestation of this Amazon forest. Rubber extraction cycle is one of them, which started occurring from the end of the nineteenth century and beginning of the twentieth century, and through the expansion of the Brazilian roads and farming boarders that started in the 1970s, around 20% of the BLA’s forest area have already been deforested to give place to human activities (Ângelo and Sá 2007; Ferreira and Salati 2005; Margulis 2003; Richards and Van Wey 2015; Weinstein 1983). Several projects such as SIPAM (Amazonia Protection System) and PRODES (Legal Amazon Deforestation Monitoring Project) (Becker 2001, 2005) have been developed over the years aiming at protecting this forest, but the consequences to the atmosphere caused by what have already been deforested cannot be neglected and are discussed by many researchers.

One of the greatest problems caused by the removal of the forest is the release of the carbon stored in this region, which is one of the largest reservoirs on the planet (Fearnside 2016; Nogueira et al. 2017). Exbrayat and Williams (2015) estimated a deficit of 7.3–8 PgC in the total above ground biomass, which corresponds to 1.5% of the recent increase in CO2. This leads to a more intense greenhouse effect and, consequently, it causes higher air temperature, not only locally but also nationally. Other important problem is related to changes in the hydrological cycle, mainly on evapotranspiration (ET), the largest source of water vapor to the atmosphere (approximately 50% of the average annual precipitation (PRP) of 2300 mm returns to the atmosphere via ET). A sharp decrease in ET is related to the increase in deforestation, which will decrease atmospheric humidity and PRP (Costa and Foley 1999; Nobre et al. 1991; Rocha et al. 2004).

The dynamic relationship between forest and climate and how each system is closely linked to the other is well documented. Shukla et al. (1990) used numerical modeling, and they concluded that this relationship for the deforested scenario is highly influenced by the following factors: (a) higher albedo and surface temperature; and (b) less precipitation and evapotranspiration. According to Shukla et al. (1990), deforested soils have lesser capacity to store water, associated with less energy available for evapotranspiration (as more energy is reflected due to higher albedo). This results in reduction of the amount of water that is returned to the atmosphere, thereby reducing precipitation. Increase in surface temperature was not even capable of enhancing the surface convergence of moisture and precipitation. Works such as Werth and Avissar (2002), D’Almeida et al. (2007), Coe et al. (2009) and Bagley et al. (2014) also show that the forest and atmosphere are linked together, that a gradual change in the Amazonian land cover will affect the atmosphere by reducing the amount of water that is returned and the atmospheric dynamic. Gentry and Lopez-Parodi (1980), Lima et al. (2014), Sorribas et al. (2016) and Guimberteau et al. (2017) expand the analyses to rivers discharge. They all agreed that precipitation is reduced due to reduction in evapotranspiration, but river discharge is scale dependent. On small-scale deforestation, the runoff is large enough so that it suppresses the reduction in precipitation and raises river level, but on large-scale deforestation, the reduced evapotranspiration with precipitation prevails and river discharge is reduced.

This study therefore intends to contribute to those researches regarding the impact of deforestation on local climate. The objective of this work is to analyze the differences in ET, PET and PRP observed through remote sensing data in forested and deforested sites at Eastern Amazon and to check if such differences are statistically significant.

Materials and methods

Study area

The study areas were selected by analyzing the data from the Legal Amazon Deforestation Monitoring Project (PRODES) of the Brazilian National Institute for Space Research (INPE). During the first phase of the project, between 1988 and 2002, the land cover calculation was made via an analog analysis of satellite images, and a digital analysis was used in the second phase of the project, between 2003 and 2005. The data used in this study were obtained in the second phase of PRODES. The land cover was classified via the analysis of several images obtained by the TM-LANDSAT satellite throughout the BLA. The images were chosen according to the amount of cloud cover and proximity to the date of calculation of the rate of deforestation, which was the first day of August of each year. The images were acquired at various angles, and their radiometric spectra were analyzed to obtain a cover classification (PRODES 2015). The results of the digital processing of satellite images between 2000 and 2015 are available to the public on the project’s website (PRODES 2015). This information is available in several formats, facilitating its use in several geographic information systems. The study areas were chosen because they maintained only one predominant type of land cover in the 15-year period between 2000 and 2014. Eight regions were selected (Fig. 1): two with areas of approximately 75.000 km2 (F-L1 and D-L1, where ‘F’ stands for forested, ‘D’ for deforested, and ‘L’ for large), two with areas of approximately 18.000 km2 (F-L2 and D-L2, the letters have the same definition as before) and four with areas of approximately 10.000 km2 (F-S1, F-S2, D-S1, and D-S2, where ‘S’ stands for small).
Fig. 1

Land use classification for 2014 and deforested areas for 2000 according to PRODES

The climate condition of this region is homogenous, in a way that of all selected areas, the rainy season starts in November/December and ends in March/April and the dry season, or the less rainy season, starts in May/June and ends in September/October (Fisch et al. 1998; Villar et al. 2009). Although the period of each season is almost the same for all areas, the accumulated value during the year can be different, as the local mechanisms may result on different distribution of PRP.

The soil classifications were obtained from the Brazilian Institute of Geography and Statistics (IBGE 2016) database and are presented in Table 1. Some areas have rivers/lakes or other soil classes that cover an area that is too small in comparison with the ones in Table 1, those classes were not considered in the analysis. In general, the soil of all regions is mainly composed of podzol and latosol, which have moderate to low infiltration rate.
Table 1

Proportion of class of soil on each area

Area

Soil

Proportion (%)

Area

Soil

Proportion (%)

F-L1

Latosol

12.64

D-L1

Latosol

22.54

Litholic

10.99

Sandy

0.94

Podzol

74.63

Litholic

12.18

F-L2

Latosol

21.52

Podzol

62.56

Sandy

8.58

D-L2

Latosol

0.04

Litholic

34.73

Sandy

10.77

Podzol

32.27

Litholic

6.53

Terra Roxa

2.89

Podzol

82.66

F-S1

Latosol

47.02

D-S1

Latosol

86.21

Litholic

10.92

Sandy

1.15

Podzol

42.07

Litholic

9.05

F-S2

Litholic

9.32

Podzol

3.59

Podzol

90.67

Terra Roxa

3.77

  

D-S2

Litholic

9.75

  

Podzol

90.03

D-L1 and D-L2 large deforested area 1 and 2, respectively, D-S1 and D-S2 small deforested area 1 and 2, respectively, F-L1 and F-L2 large forested area 1 and 2, respectively, F-S1 and F-S2 small forested area 1 and 2, respectively

Data description

Several sets of remote sensing data were used for the characterization of the areas of interest and the collection of meteorological data. Furthermore, to include an empirical analysis of the meteorological data, a series of statistical equations were used to evaluate the differences.

TRMM precipitation

The PRP data used in this study were collected during the active period of the Tropical Rainfall Measurement Mission (TRMM). Initiated in 1997 as a joint initiative between the space agencies of the USA (NASA) and Japan (JAXA), the orbital satellite TRMM generated daily PRP estimates along the Earth’s tropical zone until the end of the project in April 2015. Three satellite sensors were used to estimate PRP: Visible and Infrared Scanner (VIRS), Clouds and the Earth’s Radiant Energy System (CERES) and Lightning Imaging Sensor (LIS). While TRMM was active, the algorithms used to estimate PRP underwent several enhancements aimed at improving the generated information. For this study, data from the 3B43_V7 algorithm were used. The code combines TRMM data with PRP data collected by surface rain gauge stations to generate monthly values with a resolution of 0.25°.

Given the remote sensing features present in these data, some errors and uncertainties associated with sensor limitations and the algorithm coding itself may be found in the final product. However, several studies conducted in the Amazon and other regions of the world have been performed to validate TRMM data, and the results indicate that these data have overall validity and reliability despite the observed deviations (Buarque et al. 2011; Huffman et al. 2007).

MODIS evapotranspiration data

The moderate-resolution imaging spectroradiometer (MODIS) is one of the five environmental sensors onboard the Terra and Aqua satellites. This sensor is designed to meet the requirements of research on atmosphere, ocean and land. MODIS is the main sensor developed for satellites used in environmental studies. Its characteristics include large spatial and spectral coverage and assistance in collecting data in spectral regions used by other satellites already in orbit. In addition, it was the first satellite tool used to conduct research on global climate changes and surface and land use, among other environmental changes (Barker et al. 1992; Pereira Filho et al. 2010). In the present study, ET and PET were estimated by MODIS, which is freely available in its online repository (MODIS 2017). The monthly data covered the period between 2000 and 2014 and have a resolution of 0.05°. The algorithm used to calculate these variables uses the Penman–Monteith method and data from the MODIS sensors, including the percentage of vegetation and heat fluxes, to estimate ET and PET (Mu et al. 2007, 2011). The methodology used in the calculation of ET has accumulated errors, because it is derived from other estimates and calculations. However, the errors found in MODIS data are few and, despite their spatial and temporal variability, do not compromise the results of the studies (Kim and Hong 2008; Nadzri and Hasim 2014; Ruhoff et al. 2013).

Water balance

The water balance in each study area was determined according to the method proposed by Thornthwaite and Mather (1957). This methodology allowed the calculation of the water stored in the soil, surplus, deficit, and how much water was used in water recharge and water removal. Using PRP and PET estimated by TRMM and MODIS, respectively, the water balance was calculated using the following equations:
$$ \left( {{\text{PRP}} - {\text{PET}}} \right)\left\{ {\begin{array}{*{20}l} { < 0 \to S_{t} = {\text{AWC}} \cdot {\text{e}}^{{ - \left| {\frac{\text{NAc}}{\text{AWC}}} \right|}} \; {\text{and}}\;{\text{NAc}} = \left( {{\text{PRP}} - {\text{PET}}} \right)} \hfill \\ { \ge 0 \to S_{t} = S_{t - 1} + \left( {{\text{PRP}} + {\text{PET}}} \right)\; {\text{and}}\;{\text{NAc}} = {\text{AWC}} \cdot { \ln }\left( {\frac{S}{\text{AWC}}} \right)} \hfill \\ \end{array} } \right., $$
(1)
$$ \Delta S = S_{t} - S_{t - 1} \left\{ {\begin{array}{*{20}c} {\Delta S > 0 = {\text{Recharge}}} \\ {\Delta S < 0 = {\text{Removal}}} \\ \end{array} } \right., $$
(2)
$$ {\text{Deficit}} = {\text{PET}} - {\text{ET,}} $$
(3)
$$ {\text{Surplus}}\left\{ {\begin{array}{*{20}l} {S < {\text{AWC}} \to {\text{Surplus}} = 0} \hfill \\ {S = {\text{AWC}} \to {\text{Surplus}} = \left( {{\text{PRP}} - {\text{PET}}} \right) - \Delta S} \hfill \\ \end{array} } \right.. $$
(4)
The water balance was calculated starting in the first month in which (PRP − PET) < 0 and after the first sequence of months with (PRP − PET) > 0. This calculation was performed in stages such that the equations above were calculated in the order. In the equations above, ‘S t ’ indicates water storage in the soil at a time ‘t,’ ∆S the variation in storage, and ‘NAc’ is the accumulated negative. Because of the difficulty in obtaining all the necessary parameters (soil water content at field capacity and permanent witting point) to calculate the available water capacity (AWC), this variable was replaced by the capacity of retention of soil water, which is defined by Eq. (5):
$$ S = \frac{25,400}{\text{CN}} - 254. $$
(5)
In Eq. (5), CN is the curve number, which depends on the type of soil and ground cover. CN is a runoff index, where a low value indicates less runoff and a high value indicates more runoff (Balvanshi and Tiwari 2014). With these considerations, a sandy soil is expected to present small CN, as the water infiltrates more easily, while a soil with more clay and low infiltration would present a higher CN. To calculate CN, the soils presented in Table 1 were separated into four groups according to its hydrological characteristics, following the study of Sartori et al. (2005) for the Brazilian soils:
  • A High infiltration rate, sandy soils;

  • B Moderate to high infiltration rate, more clay is found;

  • C Moderate to low infiltration rate, water may be retained in the surface;

  • D Low infiltration rate, water has a higher difficult to percolate into the soil and runoff is constant.

As each area has different proportions of soils kind, a weighted average (Eq. 6) was used to better define single area CN.
$$ {\text{CN}} = \left( {S_{1} \times A_{1} } \right) + \left( {S_{2} \times A_{2} } \right) + \cdots + \left( {S_{n} \times A_{n} } \right). $$
(6)

In Eq. (6), ‘A n ’ is the area of the soil class ‘n’ and ‘S n ’ is the original CN value for each of the categories above as listed in Balvanshi and Tiwari (2014).

Statistical analysis

To jointly evaluate the distribution of variables considering that each variable belongs to a distinct group, regression models written in R (R Core Team 2014) can be used in which each group is represented by a dummy variable (Neter et al. 1990). In the case of two first-order linear functions, the two groups are represented by a single regression function:
$$ Y_{i} = \beta_{0} + \beta_{1} X_{1} + \beta_{2} X_{2} + \varepsilon_{i} , $$
(7)
where ‘Y i ’ is the response variable, ‘Xi1’ is the independent variable (quantitative), ‘Xi2’ is an indicative or dummy variable with two classes, i.e., 0 for group A and 1 for group B (β0, β1 and β2 are the model parameters to be estimated, and ‘ε i ’ is the error).
The response function or expected value E{Y} for this regression model is:
$$ E\left\{ Y \right\} = \beta_{0} + \beta_{1} X_{1} + \beta_{2} X_{2} . $$
(8)
Therefore, when X2 = 0 (group A), Eq. (8) becomes:
$$ E\left\{ Y \right\} = \beta_{0} + \beta_{1} X_{1} :{\text{group}}\;{\text{A}} . $$
(9)
When X2 = 1 (group B), Eq. (9) becomes:
$$ E\left\{ Y \right\} = \beta_{0} + \beta_{1} X_{1} + \beta_{2} X_{2} = \left( {\beta_{0} + \beta_{2} } \right) + \beta_{1} X_{1} :{\text{group}}\;{\text{B}} . $$
(10)
The term (β0 + β2) in Eq. (10) is a constant (that touches the Y-axis), and β1 is the slope. These response functions are shown in Fig. 2.
Fig. 2

Regression equations for groups A and B

Therefore, in the event of β2 = 0 (not significant), Eq. (10) equals to Eq. (9), which indicates the absence of a significant difference in the response Y for groups A and B. Several groups can be compared simultaneously, even in the presence of interactions, i.e., straight lines with different slopes. Models of any order can be developed according to the characteristics of the distributions.

Results

In all regions, PRP presented the expected characteristics of the Amazon region, i.e., the rainy season occurred between December and March, and the dry season occurred between May and September. The highest ET values occurred during the rainy season, and the lowest values occurred during the dry season. For PET, the highest values were observed during the dry season. These characteristics are shown in Fig. 3.
Fig. 3

Monthly mean for the period of 2000–2014 of PRP (blue bar), ET (orange line) and PET (gray line) over the large areas

The PET levels were also similar between the areas. In all the evaluated cases, the PET levels ranged between 150 and 250 mm. However, other variables differed depending on the study area. PRP was higher than 300 mm between January and March in area F-L1 and was higher than the same months in the deforested areas, except January in area D-L2. Similarly, area F-L2 showed higher PRP levels, but closer to those found in deforested areas. The main difference observed in Fig. 3 is the ET values. During the rainy months, ET values in the forested and deforested areas were similar. However, the values decreased to approximately 50 mm during the dry season in both deforested areas (decrease of approximately 90 mm per month). In addition, the forested areas had a slight decrease in the ET values during this period and the values in area F-L2 were higher than those found in other areas in the same months.

In general, smaller areas (Fig. 4) showed a behavior similar to that of larger areas. The rainy and dry periods, as well as the maximum PET and minimum ET, occurred in the same months as before. However, some significant differences were observed. Except in March and April in area F-S2, most months had higher accumulated PRP in deforested areas at the beginning or end of the year.
Fig. 4

Monthly mean for the period of 2000–2014 of PRP (blue bar), ET (orange line) and PET (gray line) over the small areas

ET in area D-S1 was the lowest in all evaluated areas, 24.44 mm in September. In the other deforested area (D-S2), the minimum ET was 65.47 mm, also in September. It is of interest that ET in the forested areas increased during the local dry season and reached to a maximum of 136.25 mm in F-S1 and 145.17 mm in area F-S2 in July and August, respectively. As observed in Fig. 3, PET showed the same pattern, such that the maximum was approximately 250 mm in the two deforested areas. Table 2 presents the value of CN and S calculated from the IBGE soil classification. The table shows that the runoff estimated by the CN is higher in F-S2, followed closely by areas D-L2 and D-S1. Area D-L1 presented the smallest CN, 46.31 mm. The storage parameter, S, is the opposite of CN. As the runoff increases, less water is stored in the soil, so S is high when CN is small. Area D-L1, which has the smallest CN, has the largest value of S, 287.4 mm. These results have direct impact on the water balance of each area, as the amount of water lost or storage on the system depends on CN and S.
Table 2

Curve number (CN) and soil capacity of water retention (S) for each forested (F) and deforested (D) area

 

CN (mm)

S (mm)

F

D

F

D

L1

64.50

46.31

139.79

287.74

L2

62.54

72.98

152.16

94.05

S1

51.96

74.42

234.87

84.12

S2

76.34

64.83

78.71

137.82

F forested, D deforested, L large areas, S small areas

The water balance (Fig. 5) calculated from the PRP and PET indicated that in the forested and deforested areas the duration of the rainy season (green tones) varied between 5 and 6 months, whereas the dry season (yellow tones) varied between 6 and 7 months. The main difference in the water balance was related to the amount of water that could be stored in the soil of each area. In general, the surplus/recharge of water was greater in the forested sites and removal/deficit in deforested sites. Among all forested sites, the input of water on the ground (recharge/surplus) was the smallest for area F-S1, reflecting the mean accumulated PRP observed. The higher deficit (dark yellow area) of all the deforested areas indicates a more intense drought, as the soil gets drier. Despite this condition, in some occasions more water is removed from the deforested soil (D-L1 and D-S2) before it goes into deficit than in the forested soil. Such results are because these deforested soils can store more water (as their S values indicate), so the amount of water removed from the system is bigger than in other soil.
Fig. 5

Soil–water budget for the monthly mean of the period from 2000 to 2014

The analysis above indicated that each variable had distinct variations between the forested and deforested areas. Although some differences are visible, a statistical assessment was necessary to identify significant differences. At this stage, the use of dummy variables allowed the evaluation of whether the differences in PRP, ET and PET between each forested and deforested area were significant. The main descriptive statistics for the variables are presented in Table 3. For all variables, the average and median values were relatively similar. The variances and standard deviations were higher for ET and lower for PET only at D-L1, D-L2 and D-S1, while in the other areas those statistics were higher for PET. The largest and smallest average ET occurred in areas F-L1 and F-L2, respectively. The highest and lowest average ET occurred in areas F-S1 and D-S1, respectively, and the highest and lowest average PET occurred in areas D-S1 and F-S2, respectively.
Table 3

Descriptive statistics for PRP, ET and PET (SD = standard deviation)

PRP

D-L1

D-L2

D-S1

D-S2

F-L1

F-L2

F-S1

F-SE

Mean

189.14

174.02

194.44

180.79

180.09

171.58

172.05

153.03

Median

185.28

169.75

185.56

175.50

175.72

168.76

170.74

151.29

Variance

356.94

316.12

459.51

831.76

356.09

222.39

510.69

400.80

SD

18.89

17.78

21.43

28.84

18.87

14.913

22.59

20.02

ET

D-L1

D-L2

D-S1

D-S2

F-L1

F-L2

F-S1

F-SE

Mean

105.73

102.59

82.98

111.44

115.64

118.39

123.50

119.38

Median

118.15

114.87

96.22

118.69

122.00

120.52

123.13

116.58

Variance

894.13

863.73

1533.31

414.71

200.49

114.54

59.71

160.18

SD

29.90

29.38

39.15

20.36

14.15

10.70

7.72

12.65

PET

D-L1

D-L2

D-S1

D-S2

F-L1

F-L2

F-S1

F-SE

Mean

189.14

174.02

194.44

180.79

180.09

171.58

172.05

153.03

Median

185.28

169.75

185.56

175.50

175.72

168.76

170.74

151.29

Variance

356.94

316.12

459.51

831.76

356.09

222.39

510.69

400.80

SD

18.89

17.78

21.43

28.84

18.87

14.91

22.59

20.02

PRP precipitation, ET evapotranspiration, PET potential evapotranspiration, SD standard deviation, D-L1 and D-L2 large deforested area 1 and 2, respectively, D-S1 and D-S2 small deforested area 1 and 2, respectively, F-L1 and F-L2 large forested area 1 and 2, respectively, F-S1 and F-S2 small forested area 1 and 2, respectively

For the evaluation or comparison of the distributions of the variables in each study area, a regression model with dummy variables was used. Figure 3 suggests that models of order 3 for all variables are best suited to fit the data. Initially, the deforested areas D-S1, D-S2, D-L1 and D-L2 were compared with the forested areas F-S1, F-S2, F-L1 and F-L2. The code for defining areas using dummy variables in the deforested areas is shown in Table 4.
Table 4

Deforested area codification

Area

d 1

d 2

d 3

D-S1

0

0

0

D-S2

1

0

0

D-L1

0

1

0

D-L2

0

0

1

D-L1 and D-L2 large deforested area 1 and 2, respectively, D-S1 and D-S2 small deforested area 1 and 2, respectively, F-L1 and F-L2 large forested area 1 and 2, respectively, F-S1 and F-S2 small forested area 1 and 2, respectively

The following regression model of order 3 was used:
$$ \begin{aligned} Y_{i} & = \beta_{0} + \beta_{1} X_{i} + \beta_{2} X_{i}^{2} + \beta_{3} X_{i}^{3} + \beta_{4} d_{1} + \beta_{5} d_{2} + \beta_{6} d_{3} + \beta_{7} X_{i} d_{1} + \beta_{8} X_{i} d_{2} + \beta_{9} X_{i} d_{3} \\ & \quad + \beta_{10} X_{i}^{2} d_{1} + \beta_{11} X_{i}^{2} d_{2} + \beta_{12} X_{i}^{2} d_{3} + \beta_{13} X_{i}^{3} d_{1} + \beta_{14} X_{i}^{3} d_{2} + \beta_{15} X_{i}^{3} d_{3} + \varepsilon_{i} , \\ \end{aligned} $$
(10)
where ‘Y i ’ is the response variable (PRP, ET or PET); ‘d1,’ ‘d2’ and ‘d3’ are dummy variables used to represent the distributions in the study area; ‘X i ’ are independent variables (month); ‘ε i ’ is the error; and ‘β i ’ (for ‘i’ varying between 0 and 15) are the model parameters to be estimated. When d1 = d2 = d3 = 0, Eq. (10) becomes:
$$ Y_{i} = \beta_{0} + \beta_{1} X_{i} + \beta_{2} X_{i}^{2} + \beta_{3} X_{i}^{3} + \varepsilon_{i} . $$
(11)
When d1 = 1, d2 = 1 and d3 = 1, Eq. (11) becomes:
$$ Y_{i} = \left( {\beta_{0} + \beta_{4} + \beta_{5} + \beta_{6} } \right) + \left( {\beta_{1} + \beta_{7} + \beta_{8} + \beta_{9} } \right)X_{i} + \left( {\beta_{2} + \beta_{10} + \beta_{11} + \beta_{12} } \right)X_{i}^{2} + \left( {\beta_{3} + \beta_{13} + \beta_{14} + \beta_{15} } \right)X_{i}^{3} + \varepsilon_{i} . $$
(12)

If β4 = β5 = β6 = 0, β7 = β4 = β8 = β9 = 0, β10 = β11 = β12 = 0 and β13 = β14 = β15 = 0, i.e., they are not significant, and Eq. (12) equals to Eq. (11), which indicates the absence of significant differences between the evaluated distributions. The significance of these parameters will indicate a difference in the respective distribution in relation to the distribution in the order of the significance.

Initially, the forested areas were compared with each other, and the deforested areas were compared with each other. The results indicate that the differences in the three variables (PRP, ET and PET) in deforested areas were not significant. However, significant differences were observed between the larger and the smaller forested areas. Subsequently, the distributions of each forested area were compared with those of the deforested areas. As expected, given in the previous results (Fig. 3, 4 and 5), the distributions of PRP and PET were not significantly different between the studied areas, and therefore, the figures for these variables are not shown in the main article, only as supplementary content (Table S1, S2 and S3). However, the distribution of ET was significantly different in most of the evaluated areas (Fig. 6) and only the difference between areas F-L1, D-L2 and D-S2 were not significant. The model adjusted by using only the significant estimates is shown in Eqs. 13.a (F-L2), 13.b (F-S1) and 13.c (F-S2). The descriptive levels p for significance levels below 10% are shown below the estimates.
Fig. 6

Observed ET and the adjusted model for this variable in areas F-L1 (upper left), F-L2 (upper right), F-S1 (lower left) and F-S2 (bottom right)

$$ \begin{aligned} Y_{i} & = 128.87_{{p\, < \,{0.0001}}} - 57.86_{p\, < \,0.067} DL2 + 41.68_{p\, < \,0.04} X_{i} DS1 + 35.38_{p\, < \,0.08} X_{i} DS2 \\ \quad & + \,52.59_{p\, < \,0.011} X_{i} DL1 - 10.56_{p\, < \,0.003} X_{i}^{2} DS1 - 6.36_{p\, < \,0.072} X_{i}^{2} DS2 - 10.451_{p\, < \,0.004} X_{i}^{2} DL1 \\ \quad & - \,7.50_{{p\, < \,{0.012}}} X_{i}^{2} DL2 + 0.609_{p\, < \,0.001} X_{i}^{3} DS1 + 0.304_{p\, < \,0.092} X_{i}^{3} DS2 + 0.531_{p\, < \,0.004} X_{i}^{3} DL1 \\ \quad & + \,0.424_{p\, < \,0.0198} X_{i}^{3} DL2, \\ \end{aligned} $$
(13.a)
$$ \begin{aligned} Y_{i} & = 85.975_{p\, < \,0.0003} + 41.371_{p\, < \,0.0048} X_{i} - 11.525_{p\, < \,0.0001} X_{i}^{2} + 0.659_{p\, < \,0.0001} X_{i}^{3} \\ & \quad - 41.199_{p\, < \,0.042} X_{i} FS1 + 12.027_{p\, < \,0.0011} X_{i}^{2} FS1 + 0.734_{p\, < \,0.0014} X_{i}^{3} FS1, \\ \end{aligned} $$
(13.b)
$$ \begin{aligned} Y_{i} & = 85.975_{p\, < \,0.0003} + 41.371_{p\, < \,0.0048} X_{i} - 11.525_{p\, < \,0.0001} X_{i}^{2} + 0.659_{p\, < \,0.0001} X_{i}^{3} \\ & \quad - \,56.566_{p\, < \,0.0063} X_{i} FS2 + 15.323_{p\, < \,0.0001} X_{i}^{2} FS2 - 0.913_{p\, < \,0.0001} X_{i}^{3} FS2. \\ \end{aligned} $$
(13.c)

Discussion

Precipitation in Amazonia is controlled by large-scale weather systems. The Intertropical Convergence Zone (ITCZ), South Atlantic Convergence Zone (SACZ), El Niño/Southern Oscillation (ENSO) and the Madden–Julian Oscillation modulate PRP in most of the Amazon region, resulting in a similar annual variability between the different evaluated areas (Figs. 3 and 4) (Ferreira et al. 2015; Foley et al. 2002; Oliveira et al. 2015; Souza et al. 2004). In addition to these general parameters, the local characteristics of each region, including proximity to water bodies, topography, land cover and geographical position, can induce the formation of mesoscale convective systems and allowing even the influence of frontal systems in certain areas, yielding local differences (Figs. 3 and 4) (Fitzjarrald et al. 2008; Cohen et al. 2014; Silva Dias et al. 2004). Among the factors that may cause spatial variability in the weather of each region, this study focused on areas with and without forest cover (for water balance, the deforested areas were considered to be rangelands). These changes, regardless of the size of the area, significantly influence local weather conditions.

Parameters related to the radiation balance, including albedo, short and long wave radiation, and sensible and latent heat, are modulated directly by land cover. A previous study on the difference between forested areas and rangelands in the Amazon (von Randow et al. 2004) found an average reduction of 13.3% in the radiation balance by altering the forest cover for grazing, an increase of 45% (28%) in the sensitive heat and a decrease in evaporation of 17 and 24% in the dry and rainy seasons, respectively. These results indicate a warmer and drier atmosphere, with a lower probability of precipitation.

The reduced evaporation in deforested areas compared with the forested areas found in this study is in accordance with previous results by von Randow et al. (2004) at Rondônia, Brazil. However, the PRP levels in smaller deforested areas during the study period were higher than in forested areas. The deforested area heats up faster and higher than the surrounding forest creating a convective circulation over the area, which advects moisture from the forest and lift it up on the deforested site. This new circulation can enhance the development of clouds and consequently enhances PRP, if enough moisture is advected (Avissar et al. 2002; Sampaio et al. 2007; Saad et al. 2010). However, as deforestation grows to a larger area, the main supply of moisture, the evapotranspiration process, becomes less efficient and less PRP occurs, even though such atmospheric circulation changes happened (D’Almeida et al. 2007).

As mentioned earlier, the minimum ET in the deforested areas was lower than that in forested areas, and the amplitude between maximum and minimum was higher in the deforested areas. Several factors control ET in the Amazon, including atmospheric relative humidity, radiation, PRP, soil type, vegetation type and depth of the roots, and each has its contribution to total ET (Hasler and Avissar 2007; Werth and Avissar 2004). PET was initially calculated to identify these contributions. This parameter is calculated taking into account a soil without water restrictions and a low and uniform grass cover (Lu et al. 2005). Therefore, PET increases with increase in atmospheric demand for water vapor; in the other words, PET increases as the atmosphere gets drier and when there is more energy available to evaporate water. Figures 3 and 4 indicate a similar variability of PET in all studied areas, regardless of land cover, and the maximum always occurred in the dry season and was slightly higher in the deforested areas. These results indicate that the atmospheric factors contribute equally to the temporal variability of ET in the evaluated regions. The small variability of PET between the areas is explained by variations in the local energy balance. As sensible heat increases and latent heat decreases in deforested areas, the atmosphere becomes hotter and drier demanding more moisture than over the forested.

In addition to climatic factors, elements of vegetation and soil can explain the variability in ET between the areas. The water from the rain is stored in soils and collected by plant roots during the dry season back into the atmosphere through the process of evapotranspiration. However, the plants in deforested areas have shallow roots, so their access is limited to the water available in the upper layers of the soil. By contrast, forest trees can access water table in the deeper regions of the soil, maintaining an optimum water balance and thereby avoiding a decrease in ET in the drier months or even resulting an increase in ET in this period because of ideal atmospheric conditions (Fan and Miguez-Macho 2010; Harper et al. 2014; Juárez et al. 2007; Miguez-Macho and Fan 2012; Williams et al. 1998). However, for the water to be stored in the soil and consequently be available to be captured by plant roots, the soils’ texture plays an important role, as this characteristic can drastically alter infiltration rate and storage capability (Saxton et al. 1986; Saxton and Rawls 2006). On a sandy soil, water can infiltrate faster than on a clayey soil (El Maayar and Chen 2006; Oldham 2014). In addition, as a clayey soil can become saturated more easily than a sandy soil, the surface runoff will be greater on the clayey one.

The soil characteristics analyzed with CN, S and soil–water budget showed that the capability of storing water in two deforested sites (D-L1 and D-S2) is higher than two forested sites (F-L2 and F-S2), but ET doesn’t corresponds to this result. This negative correlation indicates that, although the water may be available on the soil, it is not been transferred to the atmosphere. If precipitation did not show any significant changes in comparison with the forested sites, it is plausible to assume one of the two explanations, either (1) rangeland’s vegetation cannot capture the soil water, as its roots are not long enough, or (2) the less dense and diverse vegetation of rangelands cannot evapotranspirate as effective as the forested one.

The land use also changes the soil structure and the availability of water for plants (Fig. 5). In soils whose original forest cover was exchanged for rangeland, the amount of macropores and the water storage capacity decrease, whereas the surface and subsurface flow increases (Moraes et al. 2006). This structure allows less water to be stored and to become available for ET, contributing to the water balance results. In fact, the compaction of rangeland soil due to animal and machinery trapping can produce a considerable change. Martinez and Zinck (2002) analyzed the soil over forest and rangeland; they concluded that the porosity was reduced from 58–62% (forest) to 46–49% (rangeland), density increased in 42% and infiltration rate reduced from 15 to 1 cm/h. With such changes, the surface runoff increases over the rangelands and less water is stored on the soil, as it becomes more compact due to animal stomping and less water can percolate through its structure (Gholzom and Gholami 2012; Gholami 2013; Chen et al. 2014; Chyba et al. 2014; Santos and Augustin 2015).

Conclusions

The present study using remote sensing information of PRP via TRMM and ET and PET data via MODIS compared forested and deforested (rangeland) areas in the 15-year period between 2000 and 2014, and made the following conclusion:
  1. (1)

    Evapotranspiration (ET) was the only variable with a statistically significant difference between forested and deforested areas. The sites contrasted mainly during the dry season (May/June to October/November), with higher ET occurring in the forested. Although soil characteristics may play an important role, the vegetation type and root system depth are the primarily responsible for these results.

     
  2. (2)

    Potential evapotranspiration (PET) showed no statistically significant difference or any significant difference on the observed data. This variable depends only on atmospheric conditions, and such conditions does not present large discrepancies between each site, corroborating the idea that ET is controlled mainly by factors other than atmospheric ones (temperature, net radiation, humidity), such as those presented in (1).

     
  3. (3)

    Precipitation (PRP) also did not showed statistically significant differences. The large scale of the main precipitation systems in the Amazon tends to spread PRP almost equally over a large area. Although the smaller-scale systems contribute differently on each site to the accumulated PRP, they are not sufficient to make the significant difference between forested and deforested areas.

     

The impact of deforestation on local climate is clear, but it is important to remember that weather systems of other regions can use the humidity provided by the Amazonian forest and the reduction in evapotranspiration due to deforestation can have a great impact on those systems. Therefore, studying the impact of deforestation should be encouraged to understand how changes in land cover will affect local dynamics and to understand how these small changes can influence the weather of vast regions.

Notes

Acknowledgements

The authors would like to thank Instituto Tecnológico Vale (ITV) for providing financial and structural support in this study.

Supplementary material

12665_2018_7411_MOESM1_ESM.docx (17 kb)
Supplementary material 1 (DOCX 17 kb)

References

  1. Ângelo H, Sá SPP (2007) O desflorestamento na Amazônia Brasileira. Ciência Florestal 17:217–227CrossRefGoogle Scholar
  2. Avissar R, Silva Dias PL, Silva Dias MAF, Nobre C (2002) The Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA): insights and future research needs. J Geophys Res 107:8086CrossRefGoogle Scholar
  3. Bagley JE, Desai AR, Harding KJ, Snyder PK, Foley JÁ (2014) Drought and deforestation: Has land cover change influenced recent precipitation extremes in the Amazon? J Clim 27:345–361CrossRefGoogle Scholar
  4. Balvanshi A, Tiwari HL (2014) A comprehensive review of runoff estimation by the Curve Number method. Int J Innov Res Sci Eng Technol 3:17480–17485Google Scholar
  5. Barker JL, Harden MK, Anuta EA, Smid J, Hougt D (1992) MODIS spectral sensivity study: requirements and characterization, 1st edn. NASA, Washington, p 1992Google Scholar
  6. Becker BK (2001) Modelos e cenários para a Amazônia: o papel da ciência. Revisão das políticas de ocupação da Amazônia: é possível identificar modelos para projetar cenários? Parcerias Estratégicas 12:135–159Google Scholar
  7. Becker BK (2005) Geopolítica da Amazônia. Estud Avançados 19:71–86CrossRefGoogle Scholar
  8. Brazilian Institute of Geography and Statistics. IBGE [Internet]. http://downloads.ibge.gov.br/downloads_geociencias.htm. Accessed 18 Feb 2016
  9. Buarque DC, Paiva RCD, Clarke RT, Mendes CAB (2011) A comparison of Amazon rainfall characteristics derived from TRMM, CMORPH and the Brazilian rain gauge network. J Geophys Res 116:D19105-1–D19105-12CrossRefGoogle Scholar
  10. Chen G, Weil RR, Hill RL (2014) Effects of compaction and cover crops on soil least limiting water range and air permeability. Soil Till Res 136:61–69CrossRefGoogle Scholar
  11. Chyba J, Kroulík M, Kristof K, Misiewicz PA, Chaney K (2014) Influence of soil compaction by farm machinery and livestock on water infiltration rate on grassland. Agron Res 1:59–64Google Scholar
  12. Coe MT, Costa MG, Soares-Filho BS (2009) The influence of historical and potential future deforestation on the stream flow of the Amazon River–Land surface processes and atmospheric feedbacks. J Hydrol 369:165–174CrossRefGoogle Scholar
  13. Cohen JCP, Fitzjarrald DR, D’Oliveira FAF, Saraiva I, Barbosa IRS, Gandu AW, Kuhn PA (2014) Radar-observed spatial and temporal rainfall variability near the Tapajos-Amazon confluence. Rev Bras Meteorol 29:23–30CrossRefGoogle Scholar
  14. Costa MH, Foley JA (1999) Trends in the hydrologic cycle of the Amazon Basin. J Geophys Res 104:14189–14198CrossRefGoogle Scholar
  15. D’Almeida C, Vörösmarty CJ, Hurtt GC, Marengo JA, Dingman SL, Keim BD (2007) The effects of deforestation on the hydrological cycle in Amazonia: a review on scale and resolution. Int J Climatol 27:633–647CrossRefGoogle Scholar
  16. El Maayar M, Chen JM (2006) Spatial scaling of evapotranspiration as affected by heterogeneities in vegetation, topography, and soil texture. Remote Sens Environ 102:33–51CrossRefGoogle Scholar
  17. Exbrayat J, Williams M (2015) Quantifying the net contribution of the historical Amazonian deforestation to climate change. Geophys Res Lett 42:2968–2976CrossRefGoogle Scholar
  18. Fan Y, Miguez-Macho G (2010) Potential groundwater contribution to Amazon evapotranspiration. Hydrol Earth Syst Sci 14:2039–2056CrossRefGoogle Scholar
  19. Fearnside P (2016) Brazil’s Amazonian forest carbon: the key to Southern Amazonia’s significance for global climate. Res Environ Change.  https://doi.org/10.1007/s10113-016-1007-2 Google Scholar
  20. Ferreira AMM, Salati E (2005) Forças de transformação do ecossistema amazônico. Estudos Avançados 19(54):25–43CrossRefGoogle Scholar
  21. Ferreira DBS, De Souza EB, De Moraes BC, Meira Filho LGM (2015) Spatial and temporal variability of rainfall in Eastern Amazon during the rainy season. Sci World J.  https://doi.org/10.1155/2015/209783 Google Scholar
  22. Fisch G, Marengo JA, Nobre CA (1998) Uma revisão geral sobre o clima da Amazônia. Acta Amazônica 28:101–126CrossRefGoogle Scholar
  23. Fitzjarrald DR, Sakai RK, Moraes OLL et al (2008) Spatial and temporal rainfall variability near the Amazon-Tapajós confluence. J Geophys Res Biogeosci 113:G00B11-1–G00B11-17CrossRefGoogle Scholar
  24. Foley JA, Botta A, Coe MT, Costa MH (2002) El Niño-Southern oscillation and the climate, ecosystems and rivers of Amazonia. Glob Biogeochem Cycles 16:79-1–79-20CrossRefGoogle Scholar
  25. Gentry AH, Lopez-Parodi J (1980) Deforestation and increased flooding of the upper Amazon. Science 210:1354–1356.  https://doi.org/10.1126/science.210.4476.1354 CrossRefGoogle Scholar
  26. Gholami V (2013) The influence of deforestation on runoff generation and soil erosion (Case study: Kasilian Watershed). J For Sci 59:272–278CrossRefGoogle Scholar
  27. Gholzom EH, Gholami V (2012) A Comparison between natural forests and reforested lands in terms of runoff generation potential and hydrologic response (case study: Kasilian Watershed). Soil Water Res 7:166–173CrossRefGoogle Scholar
  28. Guimerteau M, Ciais P, Ducharne A, Boisier JP, Aguiar APD, Biemans H, Deurwaerder H, Galbraith D, Kruijt B, Langerwisch F, Poveda G, Ramming A, Rodriguez DA, Tejada G, Thonicke K, von Randow C, von Randow RCS, Zhang K, Verbeeck H (2017) Impacts of future deforestation and climate change on the hydrology of the Amazon Basin: a multi-model analysis with a new set of land-cover change scenarios. Hydrol Earth Syst Sci 21:1455–1475CrossRefGoogle Scholar
  29. Harper A, Baker IT, Denning AS, Randal DA, Dazlich D, Branson M (2014) Impact of evapotranspiration on dry season climate in the Amazon forest. J Clim 27:574–591CrossRefGoogle Scholar
  30. Hasler N, Avissar R (2007) What controls evapotranspiration in the Amazon Basin? J Hydrometeorol 8:380–395CrossRefGoogle Scholar
  31. Huffman GJ, Adler RF, Bolvin DT, Gu G, Nelkin EJ, Bowman KP, Hong Y, Stocker EF, Wolff DB (2007) The TRMM multisatellite precipitation analysis (TMPA): quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J Hydrometeorol 8:38–55CrossRefGoogle Scholar
  32. Juárez RIN, Hodnett MG, Fu R, Golden ML, von Randow C (2007) Control of dry season evapotranspiration over the Amazonian forest as inferred from observations at a southern Amazon forest site. J Clim 20:2827–2839CrossRefGoogle Scholar
  33. Kim J, Hogue TS (2008) Evaluation of a MODIS-based potential evapotranspiration product at the point scale. J Hydrometeorol 9:444–460CrossRefGoogle Scholar
  34. Lima LS, Coe MT, Filho BSS, Cuadra SV, Dias LCP, Costa MH, Lima LS, Rodrigues HO (2014) Feedbacks between deforestation, climate and hydrology in Southwestern Amazon: implications for the provision of ecosystem services. Landsc Ecol 29:261–274CrossRefGoogle Scholar
  35. Lu J, McNulty SG, Amatya DM (2005) A comparison of six potential evapotranspiration methods for regional use in the Southeastern United States. J Am Water Resour Assoc 41:621–633CrossRefGoogle Scholar
  36. Margulis S (2003) Causes of deforestation of the Brazilian Amazon. World Bank Working Papers (22):100Google Scholar
  37. Martínez LJ, Zinck JA (2002) Temporal variation of soil compaction and deterioration of soil quality in pasture areas of Colombian Amazonia. Soil Till Res 75:3–18CrossRefGoogle Scholar
  38. Miguez-Macho G, Fan Y (2012) The role of groundwater in the Amazon water cycle: 2. Influence on seasonal soil moisture and evapotranspiration. J Geophys Res Atmos 117:D15114-1–D15114-27Google Scholar
  39. MODIS (2017) MODIS global evapotranspiration project (MOD16) [Internet]. http://www.ntsg.umt.edu/project/mod16. Accessed 01 Feb 2017. Soil Till Res 75:3–17
  40. Moraes JM, Schuler AE, Dunne T, Figueiredo RO, Victoria RL (2006) Water storage and runoff processes in plinthic soils under forest and pasture in Eastern Amazonia. Hydrol Process 20:2509–2526CrossRefGoogle Scholar
  41. Mu Q, Heinsch FA, Zhao M, Running SW (2007) Development of a global evapotranspiration algorithm based on MODIS and global meteorology data. Remote Sens Environ 111:519–536CrossRefGoogle Scholar
  42. Mu Q, Zhao M, Running SW (2011) Improvements to a MODIS global terrestrial evapotranspiration algorithm. Remote Sens Environ 115:1781–1800CrossRefGoogle Scholar
  43. Nadzri MI, Hashim M (2014) Validation of MODIS Data for localized spatio-temporal evapotranspiration mapping. IOP Conf Ser Earth Environ Sci.  https://doi.org/10.1088/1755-1315/18/1/012183 Google Scholar
  44. Neter J, Wasserman W, Kutner MH (1990) Applied linear statistical models. McGraw-Hill/Irwin, BostonGoogle Scholar
  45. Nobre CA, Sellers PJ, Shukla J (1991) Amazonian deforestation and regional climate change. J Clim 4:957–988CrossRefGoogle Scholar
  46. Nogueira EM, Yanai AM, Vasconcelos SS, Graça PMLA, Fearnside PM (2017) Carbon stocks and losses to deforestation in protected areas in Brazilian Amazonia. Reg Environ Change.  https://doi.org/10.1007/s10113-017-1198-1 Google Scholar
  47. Oldham CD (2014) A soilmoisture accounting-procedure with a Richards’ equation-based soil texture-dependent parameterization. Water Resour Res 44:506–523Google Scholar
  48. Oliveira JV, Vitorino MI, de Abreu Sá LD (2015) Implication of Madden–Julian Oscillation phase on the Eastern Amazon climate. Atmos Sci Lett 16:318–323CrossRefGoogle Scholar
  49. Pereira Filho AJ, Carbone RE, Janowiak JE, Arkin P, Joyce R, Hallak R, Ramos CGM (2010) Satellite estimates over South America—possible applicability to the water management of large watersheds. J Am Water Resour Assoc 46:344–360CrossRefGoogle Scholar
  50. PRODES (2015) Projeto PRODES—Monitoramento da floresta amazônica brasileira por imagem de satélite [Internet]. http://www.obt.inpe.br/prodes/index.php. Accessed 16 Dec 2015
  51. Richards P, Van Wey L (2015) Where deforestation leads to urbanization: how resource extraction is leading to urban growth in the Brazilian Amazon. Nat Soc 105:803–823Google Scholar
  52. Rocha HR, Goulden ML, Miller SD, Menton MC, Pinto LDVO, Freitas HC, Figueira AMS (2004) Seasonality of water and heat fluxes over a tropical forest in eastern Amazonia. Ecol Appl 14:22–32CrossRefGoogle Scholar
  53. Ruhoff AL, Paz AR, Aragao LEOC, Mu Q, Malhi Y, Collishonn W, Rocha HR, Running SW (2013) Assessment of the MODIS global evapotranspiration algorithm using eddy covariance measurements and hydrological modelling in the Rio Grande basin. Hydrol Sci J 58:1658–1676CrossRefGoogle Scholar
  54. Saad I, Rocha HR, Silva Dias MAF, Rosolem R (2010) Can the deforestation breeze change the rainfall in Amazonia? A case study for the BR-163 Highway region. Earth Interact 14:1–25CrossRefGoogle Scholar
  55. Sampaio G, Nobre C, Costa MH, Satyamurty P, Soares-Filho BS, Cardoso M (2007) Regional climate change over eastern Amazonia caused by pasture and soybean cropland expansion. Geophys Res Lett.  https://doi.org/10.1029/2007GL030612 Google Scholar
  56. Santos WLS, Augustin HRR (2015) Water and sediment loss through runoff in areas of forest and pasture cover in southwestern Amazonia–Acre–Brasil. Z Geomorphol 59:023–039CrossRefGoogle Scholar
  57. Sartori A, Neto FL, Genovez AM (2005) Classificação hidrológica de solos brasileiros para estimativa da chuva excedente com o método do Serviço de Conservação do Solo dos Estados Unidos parte 1: Classificação. Rev Bras Recur Hídricos 10:05–18Google Scholar
  58. Saxton K, Rawls W (2006) Soil water characteristic estimates by texture and organic matter for hydrologic solutions. Soil Sci Soc Am J 70:1569–1578CrossRefGoogle Scholar
  59. Saxton KE, Rawls WJ, Romberger JS, Papendick RI (1986) Estimating generalized soil-water characteristics from texture. Soil Sci Soc Am J 50:1031–1036CrossRefGoogle Scholar
  60. Shukla J, Nobre C, Sellers P (1990) Amazon deforestation and climate change. Science 247:1322–1325CrossRefGoogle Scholar
  61. Silva Dias MAF, Silva Dias PL, Longo M, Fitzjarrald DR, Denning AS (2004) River breeze circulation in eastern Amazonia: observations and modelling results. Theor Appl Climatol 78:111–121CrossRefGoogle Scholar
  62. Sorribas MV, Paiva RCD, Melack JM, Braco JM, Jones C, Carvalho L, Beighley E, Forsberg B, Costa MH (2016) Projection of climate change effects on discharge and inundation in the Amazon basin. Clim Change 136:555–570CrossRefGoogle Scholar
  63. Souza EB, Kayano MT, Ambrizzi T (2004) The regional precipitation over the eastern Amazon/northeast Brazil modulated by tropical Pacific and Atlantic SST anomalies on weekly timescale. Rev Bras Meteorol 19:113–122Google Scholar
  64. R Core Team (2014) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.r-project.org
  65. Thornthwaite CW, Mather JR (1957) Instructions and tables for computing potential evapotranspiration and the water balance. Pub Climatol 10:183–243Google Scholar
  66. Villar JCE, Ronchail J, Guyot JL, Cochonneau G, Naziano F, Lavado W, Oliveira E, Pombosa R (2009) Spatio-temporal rainfall variability in the Amazon Basin countries (Brazil, Peru, Bolivia, Colombia and Ecuador). Int J Climatol 29:1574–1594CrossRefGoogle Scholar
  67. von Randow C, Manzi AO, Kruijt B et al (2004) Comparative measurements and seasonal variations in energy and carbon exchange over forest and pasture in South West Amazonia. Theor Appl Climatol 78:4–26Google Scholar
  68. Weinstein B (1983) The Amazon Rubber Boom 1850–1920, 1st edn. Stanford California Press, Stanford, p 356Google Scholar
  69. Werth D, Avissar R (2002) The local and global effects of Amazon deforestation. J Geophys Res 107:LBA 55-1–LBA 55-8CrossRefGoogle Scholar
  70. Werth D, Avissar R (2004) The regional evapotranspiration of the Amazon. Bull Am Meteorol Soc.  https://doi.org/10.1175/JHM-393.1 Google Scholar
  71. Williams M, Malhi Y, Nobre AD et al (1998) Seasonal variation in net carbon exchange and evapotranspiration in a Brazilian rain forest: a modelling analysis. Plant Cell Environ 21:953–968CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Juarez Ventura de Oliveira
    • 1
    Email author
  • Douglas Batista da Silva Ferreira
    • 1
  • Prafulla Kumar Sahoo
    • 1
  • Giordani Rafael Conceição Sodré
    • 1
  • Everaldo Barreiros de Souza
    • 1
    • 2
  • Joaquim Carlos Barbosa Queiroz
    • 2
  1. 1.Instituto Tecnológico Vale (ITV)BelémBrazil
  2. 2.Universidade Federal do Pará (UFPa)BelémBrazil

Personalised recommendations