Original research articles

Vegetation index cartography as a methodology complement to the terroir zoning for its use in precision viticulture

Vol 51, No 3 (2017): OENO One

DOI: http://dx.doi.org/10.20870/oeno-one.2017.51.4.1589

Abstract

Aim: Precision Viticulture (PV) is a form of vineyard management based on tools that offer winegrowers georeferenced information of each vineyard, mainly sector mapping (sub-areas) differentiated by characteristics capable of influencing vineyard usage. This provides knowledge of the variations in these sectors and PV treats each one of them in an independent and optimised manner. This allows, amongst many other possibilities, to monitor fruit ripening with the objective of performing site-specific harvest based on the characteristics of each given sector. Local variations in soil features and natural environmental factors, such as climate, lithology, geomorphology and soil, determine the units that drive or limit PV.

Methods and results: In this paper, multispectral images are used. These have been obtained between veraison and harvest in three different years in order to calculate four vegetation indexes (VI) that have been used since the end of the last century to delimit homogenous sectors in vineyards: the Normalized Difference Vegetation Index (NDVI), the Improved Soil Adjusted Vegetation Index (MSAVI), the Simple Ratio Index (SR) and the Modified Simple Ratio Index (MSR). Mapping of these VI has allowed to relate their distribution with natural environmental factors with the objective of valuing their use in the discrimination of homogenous sectors as a complement and/or alternative to traditional methodologies to terroir zoning. Results show that, in the area studied, the vineyards planted in alluvial soil and conglomerated zones, over dominant fine-loamy, mixed, mesic, Calcixerollic Xerochrept soil series, at elevations between 519 and 604 m, oriented east and on slopes less than 5º present higher values for all four indexes throughout the three years of study.

Conclusions: It is precisely these environmental elements (lithology, soil, elevation, orientation and slope) and many soil features that must be relatively uniform in order to make an efficient use of the studied VI.

Significance and impact of the study: The study addresses the use of VI as a companion tool to viticultural zoning, which has not been much explored at such scale level. In addition, the results obtained may lead to changes in the use of VI, which are usually used without taking into account soil and/or terrain features.

Introduction

The fundamental objective of Precision Viticulture (PV) is to use detailed information about the biophysical characteristics and performance of a vineyard, at high spatial resolution, as the basis for viticultural management and decision making (Bramley, 2010). Its practical implementation is dependent on various technological developments: crop sensors and yield monitors, local and remote sensors, Global Positioning Systems (GPS), Variable-Rate Application (VRA) equipment and machinery, Geographical Information Systems (GIS) and data analysis and interpretation systems (Arno et al., 2009). Therefore, PV allows maximizing yield and quality while minimizing environmental impacts and risk (Proffitt et al., 2006).    

Before the advent of PV, variability within vineyards was normally managed as "noise" (background level) and often went unnoticed. Therefore, the group of parameters involved in this variability could not be explained nor linked to each other (Cook and Bramley, 1998; Bramley and Hamilton, 2004).

Local variations in climate, lithology, geomorphology and soil factors determine the existence of sectors, which define the units that drive or limit PV (Gómez-Miguel, 2011). More precisely, some elements, mainly climate and soil, allow for sector discrimination based on grapevine development and grape composition, which can be explained by their influence on the water status of the plant (van Leeuwen et al., 2004).

The graphic representation of the different sectors, defined by the environmental characteristics, is carried through cartography and the mapping technique provided by the use of GIS. Climate maps have been produced (Jones et al., 2010), as well as lithological, geological and geomorphological maps (Morlat, 1989; Vaudour, 2010), soil maps (Morlat and Bodin, 2006), altitude, slope and orientation maps and/or a combination of various elements integrated in one unique map (Gómez-Miguel and Sotés, 1997). The completion of these maps, specially the soil map, is a complex and expensive task, but they are the basis of PV because not only do they allow the differentiation of homogenous sectors (sub-areas), they also help explain the effects of this variability in the vineyard.

When working at a regional level, medium and small scales are used (≤1:25.000/50.000), while when studying variability within or between vineyards, the used scales are larger (≥1:5.000) (Gómez-Miguel, 2011). Thus, at the field level, PV technology highlights micro-scale variability which is mainly due to variations in soil depth and physicochemical properties (Bramley and Hamilton, 2007), whereas when the area considered is larger, variability sources linked, for instance, to elevation, slope or aspect, are more likely to appear (Santesteban et al., 2013).

One of the ways to discriminate the existing differences in the behaviour of plants located within the same vineyard or in different vineyards is through vegetation indexes (VI) calculated from multispectral images, used in PV since the 1990s, which allow us to identify, at different scales, different sub-areas in the vineyards. There are studies associating the Normalized Difference Vegetation Index (NDVI) with foliage density or vigour (Hall et al., 2003; Johnson et al., 2003), plant water status (Acevedo-Opazo et al., 2008), leaf area index (LAI) (Johnson et al., 2003) or grape yield and composition (Hall et al., 2011), amongst others. On top of this, vine vigour could be used as surrogate indicators of vine yield and grape quality (Hall et al., 2002).

Algorithms have been proposed that relate different wavelengths with leaf chlorophyll content (Steele et al., 2008), leaf anthocyanin content (Steele et al., 2009) or net photosynthesis (Zarco-Tejada et al., 2013).

In this paper, four VI are calculated, analysed and mapped. More specifically, the distributions of NDVI (Rouse et al., 1973), Improved Soil Adjusted Vegetation Index (MSAVI) (Qi et al., 1994), Simple Ratio Index (SR) (Rouse et al., 1973) and Modified Simple Ratio Index (MSR) (Chen, 1996) are associated with natural environmental factors. The objective is to evaluate the use of these indexes in the discrimination of homogeneous sectors (sub-areas) as a methodological complement to terroir zoning for use in PV.

Materials and Methods

1. Study area

The study area is located in the Council of Oyón (Álava), within the Qualified Denomination of Origin Rioja (DOCa Rioja), in north-central Spain (Figure 1). From the vineyard register (MAGRAMA, 1999) and a number of field works, 300 vineyard plots are selected, with a total area of 972.46 ha. In the region, 96.41% of vineyards are planted with the Tempranillo variety (CRDOCa Rioja, 2016) and the plots selected for this study grow this variety. All features related to vineyard management of the plots (training system, soil maintenance, fertilization, green pruning, etc.) can be considered sufficiently similar. Some common characteristics of the vineyards are: tillage for weed control, plant density of around 2800 plants/ha, with 12 buds/plant distributed in 6 thumbs, managed in Cordon Royat. Thus, vineyard management should have no influence on VI distribution through the statistical analysis of randomly disposed elements.

Figure 1. Distribution of the vineyard plots in the Council of Oyón within the Qualified Denomination of Origin Rioja (DOCa Rioja, Spain).

In prior studies (Gómez-Miguel and Sotés, 1997; Fernández, 1999), zoning was done on the terroir of the DOCa Rioja and six climatic classes were identified. The area studied in this paper belongs to the fourth class, which is characterised by a Winkler Index (WI) of 1606ºC, a heliothermal Huglin Index (HI) of 2131ºC and an average annual rainfall of 430.63 mm. The analysis of climate data from the last 30 years at the nearest weather station (Viana) to the studied plots corresponded to an WI of 1702ºC, a HI of 2208ºC and an average annual rainfall of 451.6 mm. Moreover, according to the Multicriteria Climatic Classification System (MCCS) (Tonietto and Carbonneau, 2004), climate in the area is temperate warm (HI+1), with cool nights (CI+1) and moderately dry (DI+1) (Table 1), with 18.7ºC and 217 mm during the growing season (April to September). As the study area shows limited extension in space, it can be considered that most of the climate variability is induced by topographical features.

Table 1. Multicriteria Climatic Classification System (MCCS) of the studied years and 30-year average (Viana St., Navarra).


2005

2010

2013

MEAN 82-13

HI

2271

2076

1987

2208

HI CLASS

HI+1

HI-1

HI-1

HI+1

CI

13.4

12.9

13.3

13.2

CI CLASS

CI+1

CI+1

CI+1

CI+1

DI

-6

40

81

47

DI CLASS

DI+1

DI+1

DI-1

DI+1

HI: Heliothermal Huglin Index, ºC; CI: Cold Night Index, ºC; DI: Dryness Index, mm.

2. Terrain and soil features

Figure 2 shows maps of the main terrain characteristics:

Elevation (Figure 2A) of Oyón is between 350 and 950 m in the north of the Council and the slope increases from south to north (Figure 2B).

Regarding geology and land use (Figure 2D), Quaternary fluvial deposits and conglomerates are found in the drainage system. These cross the municipality from north-west to south-east and their lithology corresponds to the river/torrent and slope modelling in Haro’s facies. The same geographical distribution is shown by the soil map (Figure 2E).

Figure 2. Elevation (A), Slope (B), Aspect (C), Geology and Land Use (D) and Soil Map (E) in Oyón.

Some epipedon soil features of the main Soil Map Units (SMU) are presented in Table 2. These characteristics are: coarse elements (EG1), sand content (AT1), silt content (LM1), clay content (AC1), electrical conductivity (CE1), total limestone (CT1), active lime (CA1), organic matter (MO1), Olsen phosphorus (PO1), potassium (KCl), cation exchange capacity (CC1) and exchangeable cation calcium (CI1). The water reserve (RMM) of the main soils is also presented.

Table 2. Epipedon soil features of Soil Map Units.


Main soil of each soil map unit

EG1

%

AT1

%

LM1

%

AC1

%

CE1

dS/m

CT1

%

CA1

%

MO1

%

PO1

mg/kg

KC1

mg/kg

RMM

mm

CC1

meq/100g

CI1

meq/100g

Loamy-skeletal, mixed, mesic, Calcixerollic Xerochrept

15.6

50.1

33.1

16.8

0.18

15.7

6.3

0.97

6.3

109

79.9

11.2

10.5

Fine-loamy, mixed, mesic, Calcixerollic Xerochrept

2.4

23.8

51.8

24.3

0.21

29.8

7.2

1.31

6.3

229

104.3

12.7

11.1

3. Vegetation Indexes

For the comparative analysis, the above cited VI were used (Table 3). The calculation of the various VI is done from the multispectral images SPOT 5, geometrically corrected by measuring control points and block adjustment. Subsequently, pansharpening is made from the georeferenced panchromatic image (2.5 m spatial resolution) and georeferenced multispectral image (10 m spatial resolution) to obtain a fused image of 4 bands and 2.5 m spatial resolution. The spectral region contained in each of the bands is B1 (500-590 nm, Green), B2 (610-680 nm, Red), B3 (780-890 nm, Near-Infrared) and B4 (1580-1750 nm, Medium-Infrared). Images were taken on dates close to the beginning of ripening (stage 35) and before complete ripening (stage 38) according to Eichhorn and Lorenz scale (Coombe, 1995), in particular, the 13th of August 2005, the 14th of August 2010 and the 3rd of September 2013.

The NDVI, MSAVI and MSR values have been transformed by multiplying by 100 and adding 100, while the SR values have been multiplied by 100, which can easily be rendered with a specific colour ramp or colour map.

Table 3. Vegetation Indexes and the equations that define them.


Vegetation index

Equation

Reference

NDVI



NIR-RNIR+R

(Rouse et al., 1973)

MSAVI



2*NIR+1-2*NIR+1-8*NIR-R2

(Qi et al., 1994)

SR



NIRR

(Rouse et al., 1973)

MSR



NIRR-1NIRR0.5+1

(Chen, 1996)

Figure 3 shows a map of each studied VI in the year 2010. These VI maps, along with those of the years 2005 and 2013, will be compared to main terrain and soil characteristics.

Figure 3. Distribution by classes of SR, MSR, MSAVI and NDVI for 2010 in Oyón.

4. Data analysis

The ArcGIS 10.1 software (Environmental Systems Research Institute, ESRI) has been used for the gathering of the VI, operating with B2 (Red, R) and B3 (Near-Infrared, NIR) bands, following the equations in Table 3. This software is also used for (i) spatial auto-correlation study following the Moran´s I method (Moran, 1950; Li et al., 2007), (ii) the modelling of semivariograms and calculations of their featured parameters (nugget, sill, range) and (iii) the calculations and mapping of altitude, orientation and slope from the Digital Elevation Model (DEM) with a 5 m grid spacing (IGN, 2012). Geological data are from the national geological map (IGME, 1972-2003).

Moran´s I spatial auto-correlation analysis determines if the pattern follows a scattered (Z-score<-2.58), random (-2.582.58) distribution, with a statistical confidence level of 99% (p<0.01), where Z-scores are standard deviations of the VI spatial distributions (ArcGIS, 2017).

The Cambardella Index (Ic) is also used to define the VI distributions of different levels of spatial dependence, given their semivariogram features:

Ic=CoCo+C1*100

where Co is the nugget variance and C1 is the estimate of the spatial structural variance (sill). If the ratio is less than or equal to 25%, the distribution is considered strongly spatially dependent; if the ratio is between 25 and 75%, the distribution is considered moderately spatially dependent; and if the ratio is greater than 75%, the distribution is considered weakly spatially dependent (Cambardella et al., 1994; Cambardella and Karlen, 1999).

The average value of the four VI is also calculated for each of the 300 plots and a Principal Component Analysis (PCA) is performed relating 12 VI (3 years x 4 VI per year) with aspect, slope, elevation, RMM and the soil features presented in Table 2.

In order to graphically represent each index (Figure 3), the obtained absolute values are reclassified into five types (very high, high, medium, low, very low), each of them related to a quantile. In order to know if there is a common tendency to all indexes, each type is intersected separately, obtaining a new distribution we call Global Intersect (GI), which has the surface area included in each type for all indexes and for every year (Figure 4). Before creating this general intersection, similar interactions take place for each of the four indexes separately. All these calculations are run with the previously mentioned software.

Figure 4. Schema of the methodology for the calculation of the Very High Global Intersect; the green area contains plots classified as very high for the four indexes. The calculation of the High, Medium, Low and Very Low classes is carried out in an analogous way.

The soil classification used throughout the text is the 1994 version of the Soil Survey Staff (1994), given that it is the one followed in the zoning tasks used for the proposed soil profiling (Gómez-Miguel and Sotés, 1997; Fernández, 1999).

Results and discussion

1. Vegetation Indexes intercomparison

Table 4 shows different statistics on the annual VI distribution. Looking at the average values achieved by these indexes, per index itself and per year of study, 2013, defined by a lower average temperature and more rainfall, presents the lowest values in all four indexes. On the contrary, 2005, the only year characterised as temperate warm, presents the highest average VI values. Intermediate values are observed in 2010.

Therefore, the highest absolute values of the four VI have been obtained when HI is greater, associated with the temperature regime during the growing season. Even though 2013 showed to be the most humid (DI=81), collecting twice the rainfall than the average series (data not shown) during the growing season (April to September) and characterised as humid, one can see that it has the lowest absolute values due to HI being the lowest in the years studied (Table 1).

According to the Ic, the spatial variability for all the indexes is reasonably dependent. On the other hand, calculating the Moran´s Index and taking into account its critical value (Z-score), it is concluded that all indexes are distributed through a group pattern, which is not driven by chance (Table 4).

Table 4. Annual index distribution descriptors.


VI

Year

Min

Max

Mean

SD

Co

C1

Ic

Z-score

NDVI

2013

76

109

90.75

5.08

35.96

14.39

71.4

93.67

MSAVI

2013

38

116

78.12

11.99

97.81

179.18

35.3

92.44

MSR

2013

100

100

99.94

0.04

34.77

19.26

64.4

111.12

SR

2013

62

123

84.06

9.06

0.014

0.009

60.9

92.72

NDVI

2010

92

125

104.40

5.50

37.89

18.45

67.3

78.83

MSAVI

2010

83

139

107.33

9.67

102.78

67.77

60.3

83.93

MSR

2010

100

100

100.05

0.07

65.74

47.21

58.2

97.06

SR

2010

86

172

110.81

13.36

0.012

0.011

53.8

77.51

NDVI

2005

98

131

111.17

5.13

21.99

15.65

58.4

75.83

MSAVI

2005

96

146

119.34

8.21

44.69

50.66

46.9

79.74

MSR

2005

100

101

100.17

0.11

106.52

81.1

56.8

108.26

SR

2005

96

193

126.65

13.81

0.009

0.006

60.0

79.42

VI: Vegetation Index; Min: minimum; Max: maximum; parameters associated with the semivariogram: Co (Nugget), C1 (Sill); Ic: Cambardella Index; Z-score: Moran´s Index critical value.

2. Relationship between Vegetation Indexes and soil/terrain features

There are terrain (slope) and soil (PO1) features that do not show correlation with the VI studied. Other characteristics, such as aspect and CI1, show correlation with VI in isolated years: aspect in 2005 and CI1 in 2010. All other features are correlated with all indexes and for the three years, either positively (elevation, EG1 and AT1) or negatively (the others) (Table 5).

The inverse correlation found between MO1 and RMM with the VI is remarkable. It would be interesting to carry out studies, such as the one presented here, at different scales to confirm this relationship and understand its causes.

There are works that, mainly looking for a fast and low-cost method for the identification of soil map units, (i) study the relationship between some soil features and resistivity (Andrenelli et al., 2013) or between NDVI and soil apparent electric conductivity (ECa) map obtained by electromagnetic induction (Andre et al., 2012; Martini et al., 2013) and (ii) create remote sensing soil maps using multispectral images (Lagacherie et al., 2012) or generate terrons (continuous soil-landscape units) from algorithms (Malone et al., 2014).

Table 5. Pearson´s correlation matrix for the calculated index, terrain and soil features. Values in bold are different from zero, with an alpha significant level = 0.01.

Table 6. Surface area (ha) included in the different classes for the 3-year intersection of the indexes (NDVI Intersect, MSAVI Intersect, MSR Intersect and SR Intersect) and its general intersection (Global Intersect), elevation, geology, aspect and slope.


Variables

NDVI13

NDVI10

NDVI05

MSAVI13

MSAVI10

MSAVI05

MSR13

MSR10

MSR05

SR13

SR10

SR05

ASPECT

SLOPE

ELEVATION

EG1

AT1

LM1

AC1

CE1

CT1

CA1

MO1

PO1

KC1

RMM

CC1

NDVI10

0.795

NDVI05

0.661

0.782

MSAVI13

0.996

0.784

0.657

MSAVI10

0.798

0.997

0.784

0.791

MSAVI05

0.645

0.766

0.997

0.643

0.771

MSR13

0.954

0.781

0.634

0.930

0.774

0.610

MSR10

0.746

0.961

0.729

0.722

0.937

0.706

0.765

MSR05

0.680

0.782

0.952

0.664

0.773

0.930

0.684

0.772

SR13

0.996

0.797

0.660

0.985

0.796

0.641

0.971

0.761

0.690

SR10

0.782

0.994

0.770

0.766

0.982

0.751

0.782

0.984

0.786

0.789

SR05

0.675

0.793

0.994

0.667

0.791

0.984

0.659

0.751

0.975

0.677

0.786

ASPECT

-0.143

-0.145

-0.172

-0.145

-0.143

-0.172

-0.162

-0.142

-0.139

-0.139

-0.145

-0.170

SLOPE

0.045

-0.031

0.105

0.059

-0.023

0.116

0.007

-0.060

0.043

0.031

-0.041

0.089

-0.078

ELEVATION

0.256

0.271

0.293

0.266

0.284

0.294

0.232

0.212

0.244

0.243

0.251

0.288

0.018

0.623

EG1

0.422

0.502

0.430

0.407

0.499

0.420

0.414

0.469

0.437

0.431

0.497

0.437

-0.025

-0.229

-0.129

AT1

0.424

0.500

0.431

0.409

0.497

0.422

0.415

0.467

0.438

0.433

0.495

0.438

-0.024

-0.229

-0.128

0.996

LM1

-0.380

-0.467

-0.390

-0.366

-0.466

-0.381

-0.373

-0.434

-0.397

-0.389

-0.462

-0.397

0.020

0.223

0.142

-0.934

-0.899

AC1

-0.369

-0.459

-0.379

-0.355

-0.457

-0.371

-0.365

-0.427

-0.386

-0.378

-0.454

-0.387

0.026

0.208

0.122

-0.914

-0.873

0.995

CE1

-0.235

-0.320

-0.245

-0.222

-0.317

-0.238

-0.248

-0.302

-0.249

-0.245

-0.319

-0.253

0.052

0.075

-0.019

-0.616

-0.554

0.764

0.824

CT1

-0.388

-0.470

-0.397

-0.375

-0.469

-0.389

-0.376

-0.436

-0.404

-0.396

-0.465

-0.404

0.011

0.243

0.171

-0.946

-0.919

0.988

0.968

0.656

CA1

-0.250

-0.328

-0.261

-0.241

-0.329

-0.256

-0.243

-0.300

-0.267

-0.257

-0.322

-0.266

0.003

0.184

0.151

-0.664

-0.598

0.884

0.892

0.747

0.858

MO1

-0.353

-0.444

-0.364

-0.339

-0.443

-0.356

-0.351

-0.413

-0.370

-0.363

-0.439

-0.372

0.028

0.195

0.111

-0.883

-0.836

0.985

0.997

0.861

0.948

0.907

PO1

0.014

-0.020

0.010

0.020

-0.016

0.014

-0.016

-0.029

0.012

0.009

-0.024

0.005

0.069

-0.130

-0.210

0.003

0.040

0.057

0.151

0.663

-0.091

0.031

0.206

KC1

-0.395

-0.484

-0.405

-0.380

-0.482

-0.395

-0.391

-0.452

-0.411

-0.405

-0.480

-0.412

0.031

0.210

0.111

-0.962

-0.933

0.986

0.987

0.794

0.964

0.809

0.976

0.160

RMM

-0.343

-0.434

-0.354

-0.330

-0.432

-0.346

-0.341

-0.403

-0.360

-0.353

-0.429

-0.362

0.026

0.193

0.112

-0.863

-0.813

0.981

0.994

0.864

0.942

0.924

0.999

0.205

0.965

CC1

-0.237

-0.322

-0.248

-0.225

-0.321

-0.242

-0.242

-0.300

-0.253

-0.246

-0.319

-0.255

0.032

0.121

0.055

-0.634

-0.562

0.844

0.887

0.947

0.763

0.917

0.920

0.425

0.822

0.933

CI1

-0.102

-0.170

-0.112

-0.094

-0.170

-0.108

-0.112

-0.158

-0.114

-0.108

-0.168

-0.117

0.029

0.043

0.005

-0.329

-0.243

0.612

0.675

0.888

0.505

0.816

0.729

0.537

0.576

0.753

0.939

3. VI-based zoning

Table 6 shows the surface area included in each of the five defined classes when creating the intersection of the three years of study for each of the four indexes. The common surface to the same class during all years and for all indexes (GI) is also shown in this table and Figure 5A.

Very high and very low classes are studied in greater depth than the others because they represent a larger surface area (84% of the total GI area) and because all others can be considered as transitional.

NDVI is the most discriminant index in relation to the very high class. That is, the vineyard surface area occupying the very high class in the NDVI Intersect (67.49 ha) is similar to that occupying this class in the intersection of the four indexes (GI) (61.71 ha). Therefore, the vineyards included in the very high class of the NDVI can be considered to be included in the very high class of the other calculated VI (Table 6).

In a similar manner, plots included in the very low class for the four indexes (GI) (46.77 ha) are basically defined by the plots of this class in the MSR (54.15 ha) (Table 6).


Very High

High

Medium

Low

Very Low

Total

INDEX

Global Intersect

61.71

9.24

4.66

7.43

46.77

129.81

NDVI Intersect

67.49

29.67

24.92

38.78

93.54

254.4

MSAVI Intersect

89.03

34.38

26.11

28.75

79.98

258.25

MSR Intersect

71.74

31.76

26.8

35.52

54.15

219.97

SR Intersect

80.3

37.05

25.31

35.3

80.07

258.03

ELEVATION (m)

690-968

0.05

0.00

0.00

0.00

0.00

0.05

605-689

11.75

1.50

0.50

0.40

0.90

15.05

519-604

32.09

3.93

1.44

1.71

4.78

43.95

454-518

11.09

2.48

1.50

2.63

19.00

36.70

381-453

6.73

1.33

1.24

2.69

22.07

34.06

GEOLOGY

Fluvial and Torrential (Quaternary)

0.63

0.07

0.04

0.03

0.24

1.01

Natural Forest

12.48

0.50

0.18

0.26

4.94

18.36

Haro Facies (Quaternary)

23.88

1.72

0.59

0.55

5.24

31.98

Haro Facies (Tertiary)

24.98

4.92

2.70

4.63

40.64

77.87

ASPECT

North

3.20

0.46

0.31

0.68

5.65

10.30

East

20.10

2.54

1.04

1.58

8.46

33.72

South

24.11

3.90

1.86

3.32

20.82

54.01

West

14.29

2.34

1.45

1.85

11.83

31.76

SLOPE (º)

0.00-5.00

29.98

3.99

1.70

2.63

15.98

54.28

5.00-10.00

21.6

3.84

2.21

3.46

20.68

51.79

10.00-15.00

6.87

1.08

0.55

0.94

6.00

15.44

>15.00

3.26

0.33

0.21

0.40

4.10

8.30

The influence of latitudinal variation (between 42º 19´29´´ N and 42º 36´34´´ N) in the distribution of classes seems remarkable, because many soil and terrain features exhibit a south-to-north gradient.

The greater part of the studied vineyards is grown on sandstone and lutites of Haro’s facies (Middle-Upper Miocene; Gómez-Miguel and Sotés, 1997), occupying the very low class by 52.20% and the very high class by 32.08% (Table 6 and Figure 2D). Four multi-taxa SMU have been outlined about this tertiary formation. The main series common to all is fine-loamy, mixed, mesic, Calcixerollic Xerochrept, including different secondary soil series (Gómez-Miguel and Sotés, 1997) (Figure 2E). The vineyards included in three of these SMU (44C, 45C and 47C) present a tendency towards the very low class, while in SMU 46C, vineyard surface area is similarly distributed (approximately 20 ha) in the very high and very low classes. The reason for the different behaviour of the vineyards grown over SMU 46C is probably the presence of the secondary series mesic, Calcic Haploxeralf, while SMU 44C, 45C and 47C are occupied mainly by the secondary series fine-loamy, mixed, mesic, Calcixerollic Xerochrept and fine-loamy, mixed, mesic, Lithic Xerorthent (Gómez-Miguel and Sotés, 1997).

Figure 5. Global Intersect Map (A) and occupied surface (B) by GI classes included in the main mapping units: 17C (main series loamy-skeletal, mixed, mesic, Calcixerollic Xerochrept); 44C, 45C, 46C, 47C (main series fine-loamy, mixed, mesic, Calcixerollic Xerochrept); and MNF (Miscellaneous Natural Forest).

Over fluvial of the Haro´s facies (Figure 2D) there are vineyards which show a clear tendency to a very high class (Table 6). In fact, 74.75% of the vineyards grown over alluvial materials of Haro’s facies are included in the very high class. The main soil series found on this lithology are loamy-skeletal, mixed, mesic, Calcixerollic Xerochrept (Figure 2E), which also presents the greatest surface area of the very high class (Figure 5B).

Vineyard area over the fluvial modelling and the river/torrent modelling of the Quaternary is practically null, while 15% of the studied vineyards are grown in a land cover class, called Miscellaneous areas (Natural Forest; MNF), which are areas that have been cleared and grown with vineyards since the soil map was made (1997). This area tends to a very high class (68%), with 27% in the very low class (Table 5).

Almost all of the vineyards are planted between 380 and 690 m (Table 6 and Figure 2A), with the largest concentration of plants between 519 and 604 m. The vineyards included in the very high class are located in areas of greater altitude. More than 50% of this class can be found at heights between 519 and 604 m. The rest of this class is found mainly at heights between 605-689 m and 454-518 m. On the other hand, in lower elevation areas (381-518 m), 88% of total vineyards are included in the very low class.

The spatial distribution of the very high class in the higher areas is oriented in accordance with the Quaternary coating of Haro’s facies (Figure 2D), with this feature, together with the type of soil developed over this lithology, most likely accounting for the high VI values. Furthermore, altitude indirectly affects soil salinity (areas at a lower altitude have higher salinity levels), which could negatively affect the vegetative growth and the VI values.

42% of GI vineyards are oriented south and only 8% are oriented north (Table 6 and Figure 2C). The rest of the vineyards are evenly distributed, with 50% east and 50% west. Regarding the two class types of interest, 60% of the vineyards oriented east are included in the very high class, while 55% of those oriented north are included in the very low class. The south and west orientations do not show a clear trend towards either class.

82% of vineyards can be found on slopes between 0 and 10º (Table 6). Almost 50% of all the plots included in the very high class have a slope between 0 and 5º, while for those on slopes between 5 and 10º, the very low class is greater (44%).

Conclusion

In this paper, the existing relationship between the spatial distribution of VI and some natural environmental factors has been studied. The vineyards that reach the higher VI (in relative terms) are as follows:

- The vineyards planted on soil developed over alluvial deposits of gravel and conglomerates of the Quaternary cover of Haro´s facies, probably due to better fertility. This soil corresponds to fine-loamy, mixed, mesic, Calcixerollic Xerochrept as the main taxonomic unit, for which values of certain features such as thick elements, thick gravel and fine sand are high.

There are epipedon features correlated with VI during the three years with a 99% confidence level: (i) directly: coarse elements and sand content and (ii) inversely: silt content, clay content, electrical conductivity, total limestone, active limestone, organic matter, potassium and cation exchange capacity.

- The vineyards planted at higher altitude are oriented according to alluvial geology; this feature, together with the soil type, is probably responsible for the elevated values obtained in the VI. Furthermore, altitude indirectly affects soil salinity (areas at a lower altitude have higher salinity levels), which could negatively affect vegetative growth.

- The vineyards planted on slopes between 0 and 5º and oriented east or west, probably due to a greater capability to intercept solar radiation.

One can conclude that, in the study area, it is possible to use the VI to differentiate homogenous areas in grapevine behaviour, as long as certain elements of the physical environment are controlled and maintained homogenous in each of the sectors (sub-areas). These elements are the ones that have created variations in VI values, that is, altitude, geology, soils, slope and orientation.

If the elements mentioned in the above paragraph are maintained constant, the entities in charge of the control and protection of winegrowing could use the VI to determine the relative growth stage of the vineyards, choose strategic areas for sampling or propose selective harvesting (by sectors). That is, VI can be a useful tool, provided there is a prior zoning that includes an in-depth study of soil, terrain and climate.

Finally, the four indexes studied (NDVI, MSAVI, SR, MSR) have shown a similar spatial distribution and a high correlation between each other, possibly because the same spectral bands (Red and Near-Infrared) are operated in order to calculate them. The similarities in this behaviour lead us to think that their use for PV analysis related to vigour, water status, foliar area index and performance is the same order of magnitude as that of NDVI which, as already stated, has been used more frequently (Vaudour et al., 2015).

The study has been carried out on a district but could be done over a larger area (region), provided there is a previous zoning study at the appropriate scale.

Download article

References

  • Acevedo-Opazo C., Tisseyre B., Guillaume S. and Ojeda H., 2008. The potential of high spatial resolution information to define within-vineyard zones related to vine water status. Precision Agriculture, 9(5), 285-302. doi:10.1007/s11119-008-9073-1
  • Andre F., van Leeuwen C., Saussez S., Van Durmen R., Bogaert P., Moghadas D., de Rességuier L., Delvaux B., Vereecken H. and Lambot S., 2012. High-resolution imaging of a vineyard in south of France using ground-penetrating radar, electromagnetic induction and electrical resistivity tomography. Journal of Applied Geophysics, 78, 113-122. doi:10.1016/j.jappgeo.2011.08.002
  • Andrenelli M.C., Magini S., Pellegrini S., Perria R., Vignozzi N. and Costantini E.A.C., 2013. The use of the ARP© system to reduce the costs of soil survey for precision viticulture. Journal of Applied Geophysics, 99, 24-34. doi:10.1016/j.jappgeo.2013.09.012
  • ArcGIS, 2017. What is a z-score? what is a p-value?-ArcGIS pro | ArcGIS desktop. Retrieved from https://pro.arcgis.com/en/pro-app/tool-reference/spatial-statistics/what-is-a-z-score-what-is-a-p-value.htm#
  • Arno J., Martinez-Casasnovas J. A., Ribes-Dasi M. and Rosell J. R., 2009. Review. Precision viticulture. Research topics, challenges and opportunities in site-specific vineyard management. Spanish Journal of Agricultural Research, 7(4), 779-790. doi:10.5424/sjar/2009074-1092
  • Bramley R.G.V., 2010. Precision viticulture: managing vineyard variability for improved quality outcomes. In A. G. Reynolds (Ed.), Managing wine quality (pp. 445-479). Woodhead Publishing.
  • Bramley R.G.V. and Hamilton R.P., 2007. Terroir and precision viticulture: are they compatible? Journal International des Sciences de la Vigne et du Vin, 41(1), 1-8. doi:10.20870/oeno-one.2007.41.1.855
  • Bramley R.G.V. and Hamilton R.P., 2004. Understanding variability in winegrape production systems. 1. Within vineyard variation in yield over several vintages. Australian Journal of Grape and Wine Research, 10(1), 32-45. doi:10.1111/j.1755-0238.2004.tb00006.x
  • Cambardella C.A. and Karlen D.L., 1999. Spatial analysis of soil fertility parameters. Precision Agriculture, 1(1), 5-14. doi:10.1023/A:1009925919134
  • Cambardella C.A., Moorman T.B., Novak J.M., Parkin T. B., Karlen D.L., Turco R.F. and Konopka A.E., 1994. Field-scale variability of soil properties in central Iowa soils. Soil Science Society of America Journal, 58(5), 1501-1511. doi:10.2136/sssaj1994.03615995005800050033x
  • Chen J.M., 1996. Evaluation of vegetation indices and a modified simple ratio for boreal applications. Canadian Journal of Remote Sensing, 22(3), 229-242. doi:10.1080/07038992.1996.10855178
  • Cook S.E. and Bramley R.G.V., 1998. Precision agriculture - opportunities, benefits and pitfalls of site-specific crop management in Australia. Australian Journal of Experimental Agriculture, 38(7), 753-763. doi:10.1071/EA97156
  • Coombe B.G., 1995). Growth stages of the grapevine: adoption of a system for identifying grapevine growth stages. Australian Journal of Grape and Wine Research, 1(2), 104-110. doi:10.1111/j.1755-0238.1995.tb00086.x
  • CRDOCa Rioja, 2016. Memoria 2015.
  • Fernández L., 1999. Caracterización, integración y análisis espacial de variables edafoclimáticas. Metodología y aplicación al cultivo del viñedo en la D.O.ca.Rioja. PhD thesis, Universidad Politécnica de Madrid, Spain.
  • Gómez-Miguel V., 2011. Terroir. In J. Böhm (Ed.), Atlas das castas da península ibérica: história, terroir, ampelografia (pp. 104-153). Lisboa, Portugal: Dinalivro.
  • Gómez-Miguel V. and Sotés V., 1997. Delimitación de zonas vitícolas de la DOCa Rioja. Madrid: Universidad Politécnica de Madrid.
  • Hall A., Lamb D. W., Holzapfel B.P. and Louis J.P., 2011. Within-season temporal variation in correlations between vineyard canopy and winegrape composition and yield. Precision Agriculture, 12(1), 103-117. doi:10.1007/s11119-010-9159-4
  • Hall A., Louis J. and Lamb D., 2003. Characterising and mapping vineyard canopy using high-spatial-resolution aerial multispectral images. Computers and Geosciences, 29(7), 813-822. doi:10.1016/S0098-3004(03)00082-7
  • Hall A., Lamb D. W., Holzapfel B. and Louis J., 2002. Optical remote sensing applications in viticulture - a review. Australian Journal of Grape and Wine Research, 8(1), 36-47. doi:10.1111/j.1755-0238.2002.tb00209.x
  • IGME, 1972-2003. Mapa geológico de España (1:50.000). Madrid: Instituto Geológico y Minero de España.
  • IGN, 2012. Modelo digital del terreno (MDT05).
  • Johnson L.F., Roczen D. E., Youkhana S.K., Nemani R.R. and Bosch D.F., 2003. Mapping vineyard leaf area with multispectral satellite imagery. Computers and Electronics in Agriculture, 38(1), 33-44. doi:10.1016/S0168-1699(02)00106-0
  • Jones G.V., Duff A.A., Hall A. and Myers J. W., 2010. Spatial analysis of climate in winegrape growing regions in the western United states. American Journal of Enology and Viticulture, 61(3), 313-326.
  • Lagacherie P., Bailly J.S., Monestiez P. and Gomez C., 2012. Using scattered hyperspectral imagery data to map the soil properties of a region. European Journal of Soil Science, 63(1), 110-119. doi:10.1111/j.1365-2389.2011.01409.x
  • Li H., Calder C.A. and Cressie N., 2007. Beyond Moran's I: testing for spatial dependence based on the spatial autoregressive model. Geographical Analysis, 39(4), 357-375. doi:10.1111/j.1538-4632.2007.00708.x
  • MAGRAMA, 1999. Registros vitícolas. Madrid: Ministerio de Agricultura, Pesca y Alimentación. Recurso eléctronico.
  • Malone B.P., Hughes P., McBratney A.B. and Minasny B., 2014. A model for the identification of terrons in the Lower Hunter Valley, Australia. Geoderma Regional, 1, 31-47. doi:10.1016/j.geodrs.2014.08.001
  • Martini E., Comina C., Priori S. and Costantini E.A.C., 2013. A combined geophysical-pedological approach for precision viticulture in the Chianti hills. Bollettino di Geofisica Teorica e Applicata, 54(2), 165-181. doi:10.4430/bgta0079
  • Moran P.A.P., 1950. Notes on continuous stochastic phenomena. Biometrika, 37(1-2), 17-23. doi:10.2307/2332142
  • Morlat R., 1989. Le terroir viticole : contribution à l'étude de sa caractérisation et de son influence sur les vins : application aux vignobles rouges de moyenne vallée de la Loire. PhD thesis, Université de Bordeaux II, Institut d'œnologie.
  • Morlat R. and Bodin F., 2006. Characterization of viticultural terroirs using a simple field model based on soil depth - II. Validation of the grape yield and berry quality in the Anjou vineyard (France). Plant and Soil, 281(1-2), 55-69. doi:10.1007/s11104-005-3769-z
  • Proffitt T., Bramley R., Lamb D. and Winter E., 2006. Precision viticulture: a new era in vineyard management and wine production. Ashford (South Australia): Winetitles.
  • Qi J., Chehbouni A., Huete A.R., Kerr Y.H. and Sorooshian S., 1994. A modified soil adjusted vegetation index. Remote Sensing of Environment, 48(2), 119-126. doi:10.1016/0034-4257(94)90134-1
  • Rouse J.W.Jr., Haas R.H., Schell J. A. and Deering D. W., 1973. Monitoring vegetation systems in the Great Plains with ERTS. In Proceedings of the 3rd ERTS Symposium, NASA SP-351 1, pp. 309-317.
  • Santesteban L.G., Guillaume S., Royo J. B. and Tisseyre B., 2013. Are precision agriculture tools and methods relevant at the whole-vineyard scale? Precision Agriculture, 14(1), 2-17. doi:10.1007/s11119-012-9268-3
  • Soil Survey Staff, 1994. Keys to soil taxonomy, 6th ed. Washington D.C.: USDA. Natural Resources Conservation Service.
  • Steele M.R., Gitelson A.A., Rundquist D.C. and Merzlyak M.N., 2009. Nondestructive estimation of anthocyanin content in grapevine leaves. American Journal of Enology and Viticulture, 60(1), 87-92.
  • Steele M., Gitelson A. A. and Rundquist D., 2008. Nondestructive estimation of leaf chlorophyll content in grapes. American Journal of Enology and Viticulture, 59(3), 299-305.
  • Tonietto J. and Carbonneau A., 2004. A multicriteria climatic classification system for grape-growing regions worldwide. Agricultural and Forest Meteorology, 124(1-2), 81-97. doi:10.1016/j.agrformet.2003.06.001
  • van Leeuwen C., Friant P., Chone X., Tregoat O., Koundouras S. and Dubourdieu D., 2004. Influence of climate, soil and cultivar on terroir. American Journal of Enology and Viticulture, 55(3), 207-217.
  • Vaudour E., 2010. Los terroirs vitícolas: definiciones, caracterización y protección. Zaragoza: Acribia D.L.
  • Vaudour E., Costantini E., Jones G.V. and Mocali S., 2015. An overview of the recent approaches to terroir functional modelling, footprinting and zoning. Soil, 1, 287-312. doi:10.5194/soil-1-287-2015
  • Zarco-Tejada P.J., Catalina A., Gonzalez M.R. and Martin P., 2013. Relationships between net photosynthesis and steady-state chlorophyll fluorescence retrieved from airborne hyperspectral imagery. Remote Sensing of Environment, 136, 247-258. doi:10.1016/j.rse.2013.05.011

Authors


Alvaro Martínez

Affiliation : Universidad Politécnica de Madrid; Puerta de Hierro, 2; 28040-Madrid
Country : Spain

alvaro.martinezh@alumnos.upm.es

Vicente D Gomez-Miguel

Affiliation : Universidad Politécnica de Madrid; Puerta de Hierro, 2; 28040-Madrid
Country : Spain

Attachments

No supporting information for this article