Journal of Soils and Sediments

, Volume 14, Issue 3, pp 525–537

Use of geophysical methods for the study of sandy soils under Campinarana at the National Park of Viruá, Roraima state, Brazilian Amazonia

Authors

    • Soil DepartmentFederal University of Viçosa
  • Elpídio Inácio Fernandes Filho
    • Soil DepartmentFederal University of Viçosa
  • Carlos Ernesto Gonçalves Reynaud Schaefer
    • Soil DepartmentFederal University of Viçosa
  • Anôr Fiorini de Carvalho
    • Soil DepartmentFederal University of Viçosa
  • José Frutuoso do ValeJr.
    • Soil and Agricultural Engineering DepartmentFederal University of Roraima
  • Guilherme Resende Corrêa
    • Biodiversity and Forest InstituteFederal University of West of Pará
SOILS, SEC 2 • GLOBAL CHANGE, ENVIRON RISK ASSESS, SUSTAINABLE LAND USE • RESEARCH ARTICLE

DOI: 10.1007/s11368-013-0811-2

Cite this article as:
de Mendonça, B.A.F., Filho, E.I.F., Schaefer, C.E.G.R. et al. J Soils Sediments (2014) 14: 525. doi:10.1007/s11368-013-0811-2

Abstract

Purpose

The vegetation of the Campinaranas occurs in humid areas with hydromorphic sandy soils at the Amazon region. Thus, the determination and in situ monitoring of moisture content in Campinarana soils, besides the detection of subsurface layers are key measures for studying these soil–vegetation systems. Also, the application of ground penetrating radar (GPR) in deep sandy sedimentary sequence of Amazonia is a promising tool to enhance the knowledge on depositional and soil formation features.

Materials and methods

We studied representative soils of the Campinaranas at the National Park of Viruá, state of Roraima (Brazilian Amazonia), through the use of geophysical methods (soil moisture sensors and GPR). The study was applied in four sandy soils. Besides chemical and physical analysis of soils, soil moisture sensors were installed for monitoring during an entire hydrological year (2010/2011), and performed the calibration of sensors , coupled with imaging of the soil along transects with GPR.

Results and discussion

As a result of calibration of the soil moisture sensors we obtained a general equation with an R2 greater than 0.9. There is an influence of soil moisture content and soil temperature in the distribution of vegetation types in Campinaranas. The use of GPR identified some determinants characteristics in these soils for the differentiating the Campinaranas, represented by spodic and C horizons.

Conclusions

The spodic horizons in soils under Forest Campinarana provided potential errors in the determination of soil moisture, requiring calibration data for the precise use of this device. The investigation of the soil through the GPR showed interesting results, which allowed continuous visualization of the main soil horizons along transects in the phytophysiognomies of Campinaranas.

Keywords

CampinaranaGPRMoisture sensor calibrationSandy soilsSoil moisture monitoring

1 Introduction

The vegetation of the Campinaranas, also known as Campinas or Amazon Caatinga, occurs in humid areas and in hydromorphic sandy soils of the Amazon region, called the white-sand soils (Anderson 1981). This vegetation is characterized by a striking contrast with the surrounding rainforest landscape on clayey soils. Vegetation types in Campinaranas are strongly influenced by annual cycles and by consequent changes in water table levels, so that physiognomic gradients occurs associated with different soil hydrological conditions (Mendonça et al. 2013). Similarly to the floodplains and flooded forests of the Amazon region, the flora distribution in Campinaranas is also associated with the dynamics of water in periods with and without flooding (Prance 1996; Ferreira 1997).

However, the literature on Amazon Campinaranas is sparse, with much discussion and controversy about the origin of this vegetation (Ducke and Black 1954; Anderson et al. 1975; Anderson 1978; Prance and Schubart 1978; Anderson 1981; Ferreira 1997). Even though some studies suggest a close soil–vegetation relationship in white-sand areas, there is no definite conclusion on which soil properties are most associated with Campinaranas vegetation (Richards 1952; Rodrigues 1961). The predominant soils are Spodosols and Entisols, usually hydromorphic and associated with extensive flat surfaces (Brasil 1975a, b, 1976, 1977a, b, 1978). Despite its wide spatial distribution, difficult access, deep sandy horizons, frequent high water table level, and cemented horizons are aspects that contribute to the lack of studies of the genesis of these soils and their ecological functions (Dubroeucq and Blancaneaux 1987; Dubroeuq and Volkoff 1998).

In this regard, the determination and in situ monitoring of moisture content in Campinarana soils, besides the detection of subsurface layers with different physical properties (spodic horizons; orsteins) are key in soil–vegetation systems studies. Such measurements can be performed using geophysical methods, such as the TDRs (time domain reflectometry) and GPR (ground penetrating radar). Both methods use electromagnetic principles, with the emission of radio waves at very high frequencies (10 to 1,000 MHz) (Annan et al. 1991; Campbell Scientific 2006) and are able to measure quickly, accurately and continuously over long periods and are not harmful to operators during use and non-destructive material analyzed (Alfaro Soto et al. 2007).

Many studies use such methods (TDR and GPR) for determining and mapping the moisture content in the soils (Huisman et al. 2002; Lunt et al. 2005; Van Overmeeren et al. 1997; Weiler et al. 1998), even with good agreement between both methods (Huisman et al. 2001). However, the basic idea of these techniques consists in measuring the travel time (in microseconds) in a sequence of microwave pulses and is, therefore, an indirect measurement which has a direct relation with the dielectric constant of water in the substrate and with the minerals involved (Tommaselli and Bacchi 2001).

Thus, the accurate use of this data requires calibration equations for each specific type of soil and equipment used (Weiler et al. 1998; Evett and Heng 2008; Annan 2009). For TDR, many authors have developed calibration equations for specific soils and different types of sensors, often by different methods (Noborio 2001; Gong et al. 2003; Lukanu and Savage 2006; Souza et al. 2006; Alfaro Soto et al. 2007; Yeoh et al. 2008; Stangl et al. 2009). According to Stangl et al. (2009) when such equations are derived from multiple regression models with porosity and temperature as additional variables, they attain better results. For studies of the GPR, the measured depths are corrected by physical and morphological characteristics of the soil, since they have properties influencing the electromagnetic waves (Doolittle and Butnor 2009).

We studied soils representative of the Campinaranas physiognomies of the National Park of Viruá where datalogger systems of sensors were installed to monitor the soil moisture content (CS616, Campbell) and soil temperature (T105, Campbell), whereas GPR transects were conducted across the different vegetation types. The key point in the present study was the use of geophysical methods to identify evidences of soil–vegetation relationships at the Campinaranas, with emphasis on soil hydrological and physical properties. In order to have an efficient/effective monitoring of soil moisture, we developed a calibrate equation for the studied soils, since the adjustment equations provided by the manufacturer are not applicable for all soils. We report the data obtained for a hydrological year (2010/2011), coupled with calibration data and GPR images. This study is part of the ongoing in situ monitoring of moisture content and temperature in soils of the Campinaranas of the National Park of Viruá, to provide a better understanding of the pedo-ecological dynamics in these environments.

2 Materials and methods

2.1 Study area

The National Park of Viruá is located in south-central state of Roraima (60°58′30″W and 1°18′7″N), in the municipality of Caracaraí, distant about 190 km from the capital Boa Vista. This region encompasses a strongly regional transitional regime in Amazonia (Brasil 1975a). The Köppen classification of climate is defined as Amw' (monsoon rain type) (Peel et al. 2007). According to data from the National Water Agency (ANA 2011), obtained from the Caracaraí meteorological station, a series of 30-years shows an annual rainfall variation of 1,300–2,350 mm, with an average of 1,794 mm.

The geology is mainly characterized by a Quaternary sedimentary cover of fluvial and eolian origin and deeply weathered (Brasil 1975a; CPRM 2000). The mean altitude of the National Park is 46 m, under flat relief. The soils are sandy and hydromorphic, mostly Spodosols and Entisol. In the National Park of Viruá the main vegetation type is the Campinaranas, with all its phytophysiognomies. According to Veloso et al. (1991) classification, Campinaranas is divided into three subgroups: Forest Campinarana, Arboreous Campinarana and Grassy–woody Campinarana.

2.2 Soil sampling and installations of sensors

Samples were collected from four soil profiles according to the Campinaranas phytophysiognomies: two profiles in the Forest Campinarana in the interior (FC) and at the edge (EC); one in the Arboreous Campinarana (AC), and one in Grassy-woody Campinarana (GC) (Table 1 and Fig. 1). Morphological description and sampling of soils were performed according to Santos et al. (2005). The soils were classified according to the Soil Taxonomy (Soil Survey Staff 1999).
Table 1

Sites of monitoring, classification of soils, altitudinal (h) obtained with GPS navigation, vegetation type, soil layers and textural class with moisture sensors

Sites

Soil class

h

Vegetation

Layers

Horizons

Depths (cm)

Textural Classes

1

Oxyaquic Quartzipsamments

65 m

Grassy-woody Campinarana (GC)

GCSup.

EC

3

Loamy sand

GCInter.

C1

54

Loamy sand

GCSubsup.

Cr

109

Loamy sand

2

Oxyaquic Alorthods

65 m

Arboreous Campinarana (AC)

ACSup.

A1

4

Loamy sand

ACInter.

E

42

Loamy sand

ACSubsup.

Bh2

82

Loamy sand

3

Typic Haplorthods

66 m

Forest Campinarana—edge (EC)

ECSup.

A

2

Loamy sand

ECInter.

Bh

38

Loamy sand

ECSubsup.

C

100

Loamy sand

4

Typic Haplorthods

66 m

Forest Campinarana—interior (FC)

FCSup.

A

5

Sandy loam

FCInter.

Bh

41

Sandy loam

FCSubsup.

Bhs

103

Sandy loam

https://static-content.springer.com/image/art%3A10.1007%2Fs11368-013-0811-2/MediaObjects/11368_2013_811_Fig1_HTML.gif
Fig. 1

Monitoring sites and GPR transects in the phytophysiognomies of Campinaranas at the National Park of Viruá, state of Roraima, Brazil

A CR1000 datalogger system (Campbell Scientific) connected to soil moisture sensors (CS616, Campbell Scientific), soil temperature (T107, Campbell Scientific) and air temperature (type E, Campbell Scientific) was installed in the FC and EC and another similar systems for AC and GC (see Fig. 1). Each system is powered by a 12-V sealed battery (type SBS C11, Hawker) and is placed in a plastic barrel of 60 L. The loggers were programmed to get temperature and moisture values at an hour interval.

The soil moisture sensors were installed on March/2010 at three depths for each monitoring site: superficial, intermediate and subsurperficial (see Table 1). For each layer in which the sensor has been installed, approximately 3–4 kg of soil was collected for subsequent calibration of the sensors CS616, and for determinations of soil bulk density and texture. The soil temperature sensors (T107) were installed only in the superficial layers of each site and the air temperature sensor (type E) was installed 1.5 m above the ground for each system.

2.3 Soil analysis

Soil samples were air-dried and passed through a 2-mm sieve, to obtain air-dried fine earth (ADFE). The assessment of particle size was based on wet sieving, dispersion, and sedimentation, followed by siphoning of the <0.002-mm fraction (Ruiz 2005). The particle size was determined from a dispersion of 10 g of NaOH TFSA 0.1 mol L−1 and horizontal agitation at 50 rpm for 16 h. Then, the coarse sand and fine fractions are separated by wet sieving in sieves with meshes of 0.2- and 0.053-mm aperture, respectively. The clay fraction was determined by the pipette method, and the silt fraction calculated by difference (Ruiz 2005). Bulk density was determined from the volumetric ring method and the density of the particle by the method of the flask in ethanol (Embrapa 1997).

The analytical chemical and physical measurements were obtained using the following procedures: water pH and 1 mol L−1 KCl, using a potentiometer, both in the ratio soil/solution of 1:2.5 with an hour of contact and shaking of the suspension at reading; available P, exchangeable Na and K by Mehlich-1, P being determined spectrophotometrically, Na and K by flame emission photometry; Ca and Mg by atomic absorption spectroscopy and exchangeable Al by titration after extraction with 1 mol L−1 KCl in the ratio 1:10; and potential acidity (H + Al) by titration after extraction with 0.5 mol L−1 Ca acetate, pH 7.0. The P-rem is a quantite of phosphate in equilibrium solution with answers to a phosphate concentration adds in soil (Donagemma et al. 2008), and was determined by CaCl2 10 mmol L−1 solution, with P 60 mg L−1 (Novais and Smyth 1999). The total organic carbon (TOC) of ADFE was determined by titration with K2Cr2O7 remaining 0.2 mol L−1 Fe (NH4)2(SO4)2.6H2O after wet oxidation treatment (Yeomans and Bremner 1988).

2.4 Calibration of the sensors in the laboratory CS616

According to the manufacturer of the soil moisture sensor, model CS616 (Campbell Scientific 2006), the accuracy of this sensor can reach ±2.5% in volumetric moisture content, when using the standard calibration in a soil with an electric conductivity ≤0.5 dS m−1, a density of ≤1.55 g cm−3, and clay content less than 30% by measures ranging from 0 to 50% moisture content volume. The calibration given by the manufacturer comprises the following quadratic equation:
$$ \theta v=-0.0663-0.0063\times \mathrm{time}+0.007\times {\mathrm{time}}^2, $$
where time is provided in microseconds.
For the calibration of laboratory we simulated known gravimetric moisture content in soil samples, packed in PVC tubes, and compared it with readings obtained by the sensor in microseconds (Fig. 2). This methodology was based on several studies (Gong et al. 2003; Campbell Scientific 2006; Alfaro Soto et al. 2007; Evett and Heng 2008; Stangl et al. 2009). For this calibration, we used disturbed soil samples, since these Spodosols are structureless (single grain).
https://static-content.springer.com/image/art%3A10.1007%2Fs11368-013-0811-2/MediaObjects/11368_2013_811_Fig2_HTML.gif
Fig. 2

Laboratory calibration curves for each layer in the four sites monitored, with the calibration curve of the standard manufacturer represented in each chart. The solid lines represent the calibrated curves and the dashed lines the standard curve of the manufacturer

The collected samples were placed in PVC tubes with 10 cm of diameter and 35 cm of height, in a sufficient volume to fill the tube at about 2,700 cm3. The PVC tubes were sealed at the bottom with a cloth and trapped with elastic on the side. These samples were slightly compacted by gentle pressing of the PVC tubes on the table surface, to reach a similar bulk density of the undisturbed soil. At this stage, the moisture sensors are installed and programmed to get the moisture value at an hour interval. We collect the first signs, with the dry soil, and record information which is the time (in microseconds) and the total weight of soil–PVC–filter set.

The PVC tubes are placed in buckets with distilled water for wetting the soil sample. After complete wetting of the top of the soil for one night, water was removed by siphoning and 1 day expected to homogenize the soil moisture. Then we repeated the process of collecting information, as the soil loses moisture naturally. We weighed the soil–PVC–filter set (wet) and made the read data. After collection of five points, the soil was removed from the PVC tube and dried at 105 °C, to quantify the residual water by difference. Only one point was obtained by measuring the saturated soil in situ, which consisted of the maximum of the measurements obtained in microseconds for a year's monitoring (2010/2011). This hydrological year was extremely wet and with periods during which the soils were completely saturated, which ensured the direct correlation of soil moisture content with the total porosity of the soils. From the relation between the bulk density obtained by the method of volumetric ring, and the granulometry by the method of the flask with ethanol, we obtained the total porosity (Embrapa 1997).

2.5 Statistical data

From the gravimetric moisture content and the bulk density of soils and particles in each layer, we obtained the volumetric moisture content and built to quadratic equations that best fit the data set. The regressions analyses were performed using Microsoft Excel 2010. From the sets of data, basic statistics were applied and a Student t test at a significance level p < 0.05, comparing the averages of the calibrated equations and by the manufacturer equation. The Student t test checks the null hypothesis that the mean of sets of values calculated in each equation is statistically equal to the average soil moisture data obtained in the laboratory. The statistical package used was Statistica 6 (StatSoft Inc. 2003).

2.6 Use of GPR (ground penetration radar)

We used the GPR, GSSI model SIR-3000 with 400-MHz antenna, being processed through the software Radan 6.6. Higher-frequency (400–500 MHz) antennas often provide more satisfactory results for soil investigations (Doolittle and Butnor 2009). Two GPR transects were undertaken in Oct 2008 following the three Campinaranas physiognomies FC-AC e GC-AC (Fig. 1). Each transect was carried out in two replicates, corresponding to a 100-m-long extension. In the same area investigated by GPR, four representative soils were collected for each phytophisiognomy (Tables 1 and 2), and four auger drilling along the transects for verifying the soil layers down to 400 cm depth. For the calibration of the GPR penetration depth, we used an introduced metal bar buried at 50 cm, and detected by the GPR signal.
Table 2

Chemical and physical properties of a sequence of soils in the National Park of Viruá, south-central Roraima

Horizon (cm)

pH H2O

P

K

Na

Ca

Mg

Al

H + Al

BS

CEC

PB

m

OM

Prem

CS

FS

Silt

Clay

mg dm−3

cmolc kg−3

%

dag kg−1

mg L−1

dag kg−1

P1 Oxyaquic Quartzipsamments—Grassy–woody Campinarana

 EC (0–15)

5.23

1.2

1

0

0.01

0.01

0.41

1.7

0.02

1.72

1.2

95.3

1.85

56.6

23

57

19

1

 C1 (15–85)

4.97

1.1

0

0

0

0

0.1

0.7

0

0.7

0

100

0.84

55.4

37

50

12

1

 C2 (85–100)

5.36

1.2

0

0

0

0

0.1

1.9

0

1.9

0

100

0.41

27.3

23

50

24

3

 Cr (100+)

5.24

0.5

0

0

0

0

0.1

0.7

0

0.7

0

100

0.25

48.4

28

50

16

6

P2 Oxyaquic Alorthods—Arboreous Campinarana

 A1 (0–8)

4.86

1.2

8

12.7

0

0.05

0.92

4.1

0.13

4.23

3.1

87.6

2.75

56.4

31

52

13

4

 A2 (8–20)

5.06

0.7

2

1.8

0

0.01

0.31

2

0.03

2.03

1.5

91.2

1.41

56.6

33

48

17

2

 E (20–50)

5.49

0.5

0

0

0

0

0.1

0.8

0

0.8

0

100

0.43

57.1

39

47

12

2

 EC (50–70)

5.65

0.8

0

0

0

0

0.21

1

0

1

0

100

0.57

55.1

39

49

11

1

 Bh1 (70–80)

5.58

0.5

0

0

0

0

0.41

1.4

0

1.4

0

100

0.75

45.1

39

46

14

1

 Bh2 (80–90)

5.54

0.7

0

0

0

0

0.41

2.4

0

2.4

0

100

0.95

36.5

33

53

12

2

 Bhs (90-105+)

5.47

1.4

0

0

0

0

0.51

3.2

0

3.2

0

100

1.10

22.7

28

52

18

2

P3 Typic Haplorthods—Forest Campinarana—edge

 A/E (0–12)

4.60

3.2

10

5.1

0

0.01

1.95

11.3

0.06

11.36

0.5

97

4.16

26

24

52

19

5

 Bh (12–60)

5.15

1.5

9

0

0

0.01

0.92

8.3

0.03

8.33

0.4

96.8

2.56

14.3

25

50

20

5

 BC (60–70)

5.21

1.2

7

0

0

0

0.41

5.1

0.02

5.12

0.4

95.3

2.22

18.6

27

48

21

4

 C (70-100+)

5.23

1.7

3

0

0

0

0.1

2.5

0.01

2.51

0.4

90.9

0.52

31.4

25

50

22

3

P4 Typic Haplorthods—Forest Campinarana—interior

 A (0–5)

4.52

2.2

15

11.9

0

0.45

3.24

21.1

0.54

21.64

2.5

85.7

8.8

24.8

23

40

26

11

 E (5–13)

4.95

2.1

3

0.9

0

0.42

1.43

11.9

0.43

12.33

3.5

76.9

3.4

18.6

18

51

23

8

 Bh (13–70)

5.29

0.6

0

0

0

0.43

0.67

9.3

0.43

9.73

4.4

60.9

3.43

12.1

16

47

27

10

 Bhs (70–110)

4.98

0.6

0

0

0

0.42

0.29

3.8

0.42

4.22

10

40.8

3.6

22.5

17

47

27

9

 C (110-170+)

4.75

0.4

0

0

0

0.41

0.19

1.6

0.41

2.01

20.4

31.7

0.6

46.6

20

45

20

15

Legend: BS base sum, CEC cation exchange capacity at pH 7, PB percentage of bases, m (Al3+/Na++K++Ca2++Mg2++Al3+) × 100; OM organic matter, CS coarse sandy, FS fine sandy

The basic processing of the GPR data consisted of the following steps: (a) marking at time zero, we identifies the time reference with respect to the ground surface, in order words, identifying the start time of arrival of direct wave; (2) temporal gain, which consists of equalizing the amplitudes of the wave emitted and choose a time window in which the amplitudes signal are normalized with respect to the maximum amplitude (AGC—Automactic Gain Control), (3) filtering DC (Dewow) is a filtering for removal of low frequency components present in the data, (4) migration is a processing technique which is to apply a mathematical operator along the section to reposition the events that appear in the section of radar, in the right place in time or depth (Jol 2009).

3 Results and discussion

3.1 General characteristics of soils

The Spodosols are the main soils in the Campinarana landscapes of the National Park of Viruá (Mendonça et al. 2013). These soils occur in all three vegetation types: Forest Campinarana, Arboreous Campinarana and Grassy-woody Campinarana. However, they have thicker spodic horizons, ranging between 40 and100 cm thick, and more organic matter (OM) in the FC (Table 2). In the AC phytophysiognomies the E horizon is thicker (greater than 50 cm), whitish with increasing organic matter, increased Cation Exchange Capacity (CEC) (see Table 2). In GC, the Oxyaquic Quartzipsamments are the dominant soils, with lower organic matter contribution of this vegetation type, insufficient to allow accumulation in surface and subsurface (see Table 2). However, we notice areas, especially near small shrub island groupings, where Typic Haplorthods occur as inclusions, with notable evidence of lateral podzolization in lamellae features (Mendonça et al. 2013). The process of the lateral podzolization in the Campinaranas vegetation was reported by Reichardt et al. (1975), who defined the magnitude of the flow of water in the water table and indicated the existence of a horizontal flow of saturated water in the order of 0.0018 cm−1 m, which means 0.0018 cm3 of water cm2 at a crossing perpendicular to the flow in 1 h.

The variations of these soils in the same order, especially regarding the increase of OM in surface and subsurface (see Table 2), is regulated by subtle terrain variations with contrasting subsurface drainage. Periodically, high water table level form a general flooded environment, with varying conditions to support vegetation, and incorporation and decomposition of organic matter. During the soil survey, we also detected a close soil–vegetation relationship, especially by the OM contents in subsurface.

Soils are generally loamy sand, with a predominance of fine sand, followed by coarse sand, silt, and clay (see Table 2). Such characteristics suggest an aeolian origin of the upper sediments (Içá formation), originated from the reworking of weathered sandstone (Brasil 1975a; Santos and Nelson 1995; CPRM 2000; Carneiro Filho et al. 2003). These soils are chemically acid, dystrophic and nutrient depleted, with a pH of 4.5 to 5.6, very low bases' sum (BS) (mean less than 0.5 cmolc dm−3) (see Table 2).

In FC, the water table level does not reach the surface allowing a greater vegetation with a forest canopy 15 m tall, higher than AC (Mendonça et al. 2013). The vegetation size of AC is shorter and more stunted, but have common species with FC. The longer flooding period of GC hinders the establishment of larger trees.

3.2 Calibration of the moisture content sensor

For all studied sites, the calibration in each layer was a quadratic equations adjusted with a R2 greater than 0.96 (Table 3). Figure 2 shows the calibration curves obtained in the laboratory (solid lines) for each layer and the manufacturer's standard curve (dashed line). The calibrated data have higher volumetric moisture content at all monitoring sites in relation to data adjusted by the equation of the manufacturer (see Fig. 2 and Table 4). In general, for each site, the relative proportions of soil moisture content are kept in depth with the calibration in the laboratory; however, averages, maximums and minimums are underestimated when using the manufacturer's equation (see Table 4).
Table 3

Equations adjusted from the calibration laboratory, with the layers at each site monitored, adjusted equations and their coefficients of determination (R2)

Layers

Adjusted equations

R2

1—Grassy-woody Campinarana (GC)

 GCSup.

y = −0.0014x2 + 0.0933x − 1.0899

0.9838

 GCInter.

y = −0.0004x2 + 0.0523x − 0.7213

0.9716

 GCSubsup.

y = −0.0004x2 + 0.0501x − 0.6879

0.9770

2—Arboreous Campinarana (AC)

 ACSup.

y = −0.0013x2 + 0.0926x − 1.0979

0.9916

 ACInter.

y = 0.0001x2 + 0.0329x − 0.5540

0.9890

 ACSubsup.

y = −0.0013x2 + 0.0904x − 1.0854

0.9896

3—Forest Campinarana − edge (EC)

 ECSup.

y = −0.0009x2 + 0.0728x − 0.9166

0.9658

 ECInter.

y = −0.0017x2 + 0.1075x − 1.2287

0.9894

 ECSubsup.

y = −0.0024x2 + 0.1353x − 1.5439

0.9788

4—Forest Campinarana—interior (FC)

 FCSup.

y = −0.0007x2 + 0.0646x − 0.8384

0.9820

 FCInter.

y = −0.0006x2 + 0.0622x − 0.8112

0.9812

 FCSubsup.

y = −0.0010x2 + 0.0782x − 0.9848

0.9854

Table 4

Maximum, mean, minimum and standard deviation of soil moisture content at each layer studied in the four studied site in 1 year of monitoring (2010–2011), using (cubic meter per cubic meter) the equation obtained from the manufacturer (a) and calibrated the equation (b). Density of soil (Ds) and porosity in the respective layers studied

Layer

Maximum

Mean

Minimum

Standard Deviation

Ds

Porosity

 

m3 m−3

 

g cm−3

%

1—Grassy-woody Campinarana (GC)

 GC Sup. a

0.50

0.35

0.08

0.10

 GC Inter. a

0.30

0.27

0.06

0.03

 GC Subsup. a

0.24

0.23

0.17

0.01

 GC Sup. b

0.46

0.43

0.20

0.05

1.48

0.48

 GC Inter. b

0.42

0.40

0.12

0.04

1.58

0.42

 GC Subsup. b

0.34

0.33

0.27

0.01

1.86

0.36

2—Arboreous Campinarana (AC)

 AC Sup. a

0.40

0.35

0.12

0.05

 AC Inter. a

0.28

0.26

0.06

0.03

 AC Subsup. a

0.26

0.25

0.10

0.02

 AC Sup. b

0.52

0.50

0.29

0.03

1.32

0.52

 AC Inter. b

0.41

0.39

0.10

0.04

1.60

0.44

 AC Subsup. b

0.40

0.39

0.22

0.02

1.70

0.40

3—Forest Campinarana—edge (EC)

 EC Sup. a

0.81

0.07

0.01

0.06

 EC Inter. a

0.37

0.30

0.10

0.06

 ECSubsup. a

0.31

0.29

0.15

0.02

 EC Sup. b

0.56

0.14

0.02

0.06

1.29

0.51

 EC Inter. b

0.46

0.44

0.26

0.03

1.45

0.47

 EC Subsup. b

0.36

0.36

0.29

0.01

1.76

0.38

4—Forest Campinarana—interior (FC)

 FC Sup. a

0.55

0.14

0.05

0.05

 FC Inter. a

0.37

0.32

0.14

0.04

 FC Subsup. a

0.34

0.32

0.14

0.03

 FCSup. b

0.56

0.25

0.10

0.06

1.04

0.61

 FC Inter. b

0.51

0.47

0.27

0.04

1.33

0.52

 FC Subsup. b

0.44

0.43

0.25

0.03

1.62

0.42

Despite the data underestimate, the t-student test to evaluate the data and equations calibrated by the manufacturer showed significant differences only for the intermediate layers of FC and EC, unpaired data with the significance level of p <0.05. These layers correspond to the spodic horizons, where high amounts of OM are found (see Table 2). Also, they have the lowest Prem, with 14.3 and 12.1 mg L−1 (see Table 2), due to the presence of low crystallinity mineral linked to Al and/or Fe. The organic matter content, the concentration of ions in solution, clay content, and soil compaction are all factors which interfere with the dielectric properties of the soil, which may provide potential errors in determining the volumetric moisture content (Stangl et al. 2009; Gong et al. 2003; Siddiqui and Drnevich 1995).

Figure 2 shows that the results of the dry soils are coincident with the standard equation of the manufacturer and calibrated equations. However, with increasing soil moisture the error increases. According to Ruelle and Laurent (2008), the precision of the equations calibrated by the manufacturer decreases with increasing moisture content. Thus, the equation of the manufacturer seems to be adequate for sandy loam or sandier soils (Table 1) in dry conditions, but other factors such as high OM content (Table 2) or soil density (Table 4) can attenuate the signal, hence recommending further calibration. Figure 2 also highlights the low moisture content retention capacity of soils when air-dried, close to zero, due to the sandy texture.

The maximun values of volumetric moisture, greater than the total porosity for each layer (Table 4) are explained by intrinsic errors of equipment accuracy (±2.5%), and approximations of some coefficient used for each equation. However, the values obtained at the ECsup. layer based on the suggested equipment equation (0.81 m3 m−3) pointed out the need for calibration of sensors, in view of high OM at the surface, and abundant roots.

From the total data set obtained in the calibration laboratory, we built a general equation for the soils under Campinaranas (Eq. (1)). These results indicate significant differences between the equation of the manufacturer and the observed data, while data obtained with calibrated general equation did not differ significantly. Figure 3 shows the best-fit equation calibrated (solid line) compared to the standard equation of the manufacturer (dashed line) the data distribution
https://static-content.springer.com/image/art%3A10.1007%2Fs11368-013-0811-2/MediaObjects/11368_2013_811_Fig3_HTML.gif
Fig. 3

Calibration curve for general soils (solid line) and calibrated standard manufacturer's equation (dotted line)

$$ \theta v=-0.0010\times {\mathrm{time}}^2+0.0805\times \mathrm{time}-0.9969;\kern0.5em {R}^2=0.9313 $$
(1)

3.3 Pedoclimatic aspects of the soil–vegetation Campinaranas

The different soil bulk densities obtained in the respective layers at each site can be directly related to the phytophysiognomies gradient of the Campinaranas. The layers of GC (site 1) have the highest density of soils in relation to other sites, which difficult the infiltration of rainwater and favor the flooding, often observed under this vegetation type. Despite, the GC occurs in the bottom of the landscape. Site 4 (FC) has the lowest densities in all layers, which can be associated with high amounts of soil organic matter and many fine and medium roots in FCSup. and FCInterm. layers. These conditions favor the infiltration and drainage of rainwater for the AC and GC and may be related to the establishment of trees. In all sites, the soil density (Ds) increases and porosity decreases with increasing depth (see Table 4).

The data calibrated in the laboratory show that the intermediate layers of FC and EC have the highest moisture content (see Table 4 and Fig. 4), which indicates a possible drainage barrier on the lower parts of the intermediate layers (spodic horizons). Such “hard pans” are often associated with Spodosols found in the Amazon region (Sioli and Klinge 1962; Sombroek 1966). The upper layers with the lowest moisture content possibly associated with higher porosity (see Table 4), and high undecomposed organic matter. Both conditions seem to favor the occurrence of FC vegetation, which serves as a reservoir of water during the dry periods, in the intermediate layers. On the other hand, at the surface, they allow better drainage during the wet periods.
https://static-content.springer.com/image/art%3A10.1007%2Fs11368-013-0811-2/MediaObjects/11368_2013_811_Fig4_HTML.gif
Fig. 4

Graphs of volumetric soil moisture content in a hydrological year collection of CS616 sensors in the National Park of Viruá, state of Roraima (Brasil), at different depths and using the equation provided by the manufacturer (a) and the equation calibrated in the laboratory (b)

In AC and GC, surface layers showed the highest moisture content during the monitoring period, with moisture content at the GC site decreasing with depth (see Fig. 4). These results can be related to soil conditions of these vegetation types, which are often flooded in the rainy season, whereas severe water deficit can occur in the dry season (Mendonça et al. 2013).

The surface layers of GC and AC and intermediate layers of FC and EC showed similar trends in moisture content, for the period between April to mid-August, which coincides with the period of greatest precipitation in this region (ANA 2011) (see Figs. 4 and 5). In the period from September to March, which corresponds to the driest months, the soil moisture content shows fluctuations with a few small events, like the wettest months (see Figs. 4 and 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs11368-013-0811-2/MediaObjects/11368_2013_811_Fig5_HTML.gif
Fig 5

Rainfall stations in the surrounding of the National Park of Viruá: a station “Fazenda Paraná” (01°07'35"; −60°23'58"), with time series from 1979 to 2011, b station “Agropecuária Boa Vista” (01°27'39"; −60°46'30"), data from 1985 to 2010 (Source: ANA, 2011)

The surface layers of FC and EC, which have good drainage and the lowest densities of the soils, have peaks with maximum moisture content, possibly related to the time of rainfall in line with the time of the data acquisition. After these events, we can see the gradual reduction of soil moisture content (see Fig. 4), due to more efficient drainage of these soils.

According to data from the monitoring period (2010/2011), the soil temperature at the surface is also matched by the different phytophysiognomies of Campinaranas. The Fig. 6 show the monthly average temperature of the surface layers of each site, and air temperature. The mean temperature was 26.5, 26.4, 28.1, and 29.1 °C, with a standard deviation of 1.1, 1.3, 3.1, and 3.7, respectively for FC, EC, AC and GC. The GC and AC presented at the monitoring of this year, 34 and 4 records, with soil temperature above 40 °C, respectively. This may be related to higher exposure to direct radiation and low organic matter and nutrient contents of these environments. This further indicates a thermal limitation for establishment and higher fire risk. Cooler temperature of the Forest Campinarana and little variation becomes, over time, a benefitial trade off for trees colonization. The average air temperature showed a difference of 0.3 °C between the Forested Campinarana (FC and EC) and the grassy, open areas (AC and GC).
https://static-content.springer.com/image/art%3A10.1007%2Fs11368-013-0811-2/MediaObjects/11368_2013_811_Fig6_HTML.gif
Fig. 6

Average monthly temperature in the surface layers at each site and mean air temperature

3.4 Study with Georadar

During the GPR survey, the soils were very wet due to rainy weather period. According to Smith and Jol (1995), sandy-textured, quartz-rich soils saturated with water cause only an additional attenuation of the GPR signal. The attenuation of the signal is generated by increasing the electrical conductivity of the soil in the presence of water, which also occurs in soils rich in soluble salts and clay (McNeill 1980; Smith and Jol 1995). However, in general, the electrical conductivity of soils is governed by the amount and types of clay minerals (McNeill 1980). The studied soils shows low clay and soluble salts contents (Table 2), which reveals a low attenuation of the GPR signal that can be appropriate for enhanced knowledge of these soils properties.

The resolution and depth investigated by GPR is limited by the frequency of the antennas used and the electrical properties of soils, which were lower in soils under saturated conditions (Doolittle and Butnor 2009). Thus, in the studied soils even in high humidity condition and with an antenna of 400 MHz, the GPR prospection reached 4.5 meters, but with much noise below 2 m depth (Fig. 7). However, most soil properties at the soil–vegetation interface occur within 200 cm, allowing inferences on soil–vegetation relationships.
https://static-content.springer.com/image/art%3A10.1007%2Fs11368-013-0811-2/MediaObjects/11368_2013_811_Fig7_HTML.gif
Fig. 7

GPR images obtained with 400-MHz antenna in different phytophysiognomies of Campinaranas in the National Park of Viruá, state of Roraima

The GPR has been used to estimate the depth and thickness of spodic horizons (Collins and Doolittle 1987; Doolittle 1987; Burgoa et al. 1991; Ucha et al. 2002; Doolittle and Butnor 2009) also may indicate the presence of albic horizons (Doolittle et al. 2005), ortstein (Mokma et al. 1990) and distinguish changes of colors associated with abrupt changes and contrasting the organic carbon content (Collins and Doolittle 1987). In Fig. 7, it can be seen the reduction of the spodic horizon (Bh and Bhs) and increases the E horizon in the FC phytophysiognomies toward AC. In GC we observed the upper C horizon (see Fig. 7), which are more dense (see Table 4) and with the presence of yellowish mottling, which are the conditions associated with a drainage deficiency, which facilitate reflection of GPR waves. The spodic horizon is observed only in FC and AC and does not occur in GC (see Fig. 7). The presence of continuous C horizon and higher soils densities seem to correlate with the difficulty of internal drainage in these environments. Furthermore, subtle altitudinal differences (see Table 1) are essential for understanding the hydrological dynamics of the Campinaranas, not illustrated in Fig. 7.

Knowledge of the soil characteristics and properties is critical to effective use of GPR, and it is possible to determine levels of fitness for its use in certain regions, using the criteria of some taxonomic classes (Doolittle and Butnor 2009). For the studied soils, the use of GPR identified some determinants characteristics of the soils for the differentiation of the phytophysiognomies of the Campinaranas (see Fig. 7).

Some studies use data from the GPR and correlated with soil moisture sensors (TDR) for the determination and mapping of moisture content in soils (Huisman et al. 2002; Lunt et al. 2005; Van Overmeeren et al. 1997; Weiler et al. 1998). Contrary to the results of Weiler et al. (1998), Huisman et al. (2001) found no structural differences between the calibration equations of GPR and TDR, based on the wave speed on the ground. Huisman et al. (2002) compared the TDR and GPR in an experimental system for irrigation and concluded that the GPR is suitable to capture signals in the soil moisture content. For the studied soils, the correlation analysis was not performed because the methods were not applied simultaneously.

4 Conclusions

  1. 1.

    There is an influence of soil moisture content and soil temperature in the distribution of vegetation types in the Campinaranas ecosystem.

     
  2. 2.

    Although the Campinarana soils have rather similar physical characteristics, they have varying amounts of organic matter, and different soil densities between the different layers and between vegetation types.

     
  3. 3.

    The conspicuous presence of high OM and low crystalline mineral forms in the spodic horizons of Forest Campinarana sites, cause potential errors in the determination of soil moisture content, requiring calibration of these data, using the CS616 sensor.

     
  4. 4.

    The equation of the sensor CS616 provided by manufacturer underestimated the soil moisture. The results obtained with the equation of the manufacturer showed significant differences compared with the total data obtained in the laboratory, while the general equation calibrated for the studied soils have no significant differences.

     
  5. 5.

    The use of GPR in these sandy soils showed interesting results, which allowed continuous visualization of the main soil horizons along transects in the phytophysiognomies of Campinaranas. It is particularly appropriate for enhanced knowledge of soil formation, with cost/time effectiveness for the analysis of soil gradients.

     
  6. 6.

    The GPR method identified determinants characteristics of soils for differentiating the Campinaranas. The main differences observed in the GPR images are related to the presence or absence of strong reflectors, represented by spodic and C horizons.

     

Acknowledgements

We thank the Protected Areas of Amazon Program (ARPA), Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio) of Roraima, and the National Park of Viruá team, for financing and the logistics supports during the expedition. Also thanks Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for the scholarships.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013