Advertisement

Land suitability assessment for cassava production in Indonesia using GIS, remote sensing and multi-criteria analysis

  • Riska Ayu Purnamasari
  • Tofael Ahamed
  • Ryozo Noguchi
Article

Abstract

Sustainable land use is essential for increasing the production of cassava as a diversified crop for ensuring food security in Indonesia. Understanding spatial factors and criteria is required for locating suitable production areas to increase cassava production. In this study, a spatial model was developed to assess the suitability of land for supporting sustainable cassava production. The model was divided into three stages considering different criteria. First, satellite digital images were processed from Landsat-4 Thematic Mapper (TM), Landsat 8 Operational Land Imager (OLI), and Sentinel-2 satellites to create vector data layers and a normalized difference vegetation index (NDVI) database. Second, a spatial analysis was performed to identify highly suitable areas for cassava production using a geographical information system (GIS) and the multi-criteria analysis. Third, a sustainability evaluation was conducted based on land suitability information for a study period of 5 years. Land suitability assessment was performed to increase cassava production. We found that 43.11% (11094 ha) of the study area was highly suitable for cassava production, whereas 30.87% (8233 ha) was moderately suitable and 9.83% (2623 ha) was marginally suitable with incorporating AHP analysis. Moreover, 17.69% (4718 ha) of the land was occupied by residents and settlements. On the other hand, ANP analysis also conducted to confirm the AHP results. We have found approximate similar results with no significant differences in any of the suitability classes. This research recommends that the integrated approach of GIS-based multi-criteria can be extended with satellite remote sensing vegetation datasets to assess regional production and site-specific management of cassava crops.

Keywords

AHP Cassava GIS Suitability Sustainability 

1 Introduction

Land suitability assessments are important for sustainable land use and for the selection of potential crops in the changing climates of Indonesia. Indonesia is a developing country with the fifth largest population in the world. The large population increases dependence on rice as a staple food, which could create the threat of food insecurity (Khush 2005). To mitigate this dependency, diversification through the consumption of local foods, such as cassava, is desirable. Cassava is a good alternative that poses fewer risks as a root crop and plays an important role in Indonesia, which is one of Asian countries to support sustainable local food production (Campo et al. 2011; Noerwijati and Budiono 2015; Feenstra 1997; Ariningsih 2018). Cassava can be easily grown, cultivated, and distributed to local communities (Kolawole et al. 2010). The benefits of cassava as a local food could strengthen the food security of developing countries. In the future, cassava has the potential to become a promising crop that can adapt to changing climatic patterns due to its low water and soil acidity requirement compared to rice (FAO 2013; Khumaida et al. 2016). Therefore, sustainable cassava production in Indonesia must ensure maximum benefits for growers. While considering the sustainability of cassava production, criteria related to environmental, ecological, economic, and social indicators must be addressed (Sydorovych and Wossink 2008; Tiwari et al. 1991). Furthermore, food security is one of the major concerns in the context of agricultural sustainability and the sustainable supply of food for the increasing population (Ahamed et al. 2015). Sustainable land use for cassava production significantly drives maximizing the production of cassava to contribute to the food security of Indonesia.

To increase cassava production, suitable areas and ecological conditions must be identified (Heumann et al. 2011). Such important tasks associated with increasing the production of cassava can be addressed through spatial analyses of land use suitability. Suitability classification reflects the suitability of each land unit for cassava production. In the Food and Agriculture Organization’s (FAO) 1976 framework for land evaluation, land was divided into four classes: highly suitable (S1), moderately suitable (S2), marginally suitable (S3), and not suitable (N). Spatial assessments of land suitable for cassava production could serve as a starting point for sustainability evaluations. Additionally, interactions between suitability and sustainability have been reported in the FAO’s (1976) international framework for evaluating sustainable land management. Environmental factors deemed suitable can reflect the level of sustainability for the same land use over a period of time.

As a spatial tool, geographic information systems (GIS) have been used to conduct spatial analyses of suitability for various purposes, especially land suitability (Ferretti and Pomarico 2013; Malczewski 2006; Smyth and Dumanski 1993). In addition, applications of remote sensing in agriculture include several aspects such as plant phenology, economic features, and land use management (Ceballos-Silva and Lopez-Blanco 2003). These applications have played an important role and suggest that remote sensing technology is suitable for monitoring agricultural activities. In regional scales of land suitability assessment, satellite remote sensing provides the opportunity to include phonological information of vegetation. The vegetation information can help determine the growth information of cassava plantations and can help inform the decision-making process of land suitability (Vrieling et al. 2011).

Therefore, investigating land suitability depends on multiple criteria and factors in the decision-making process that can largely be assessed using geospatial datasets (Ceballos-Silva and Lopez-Blanco 2003). A key step of land suitability assessment for cassava production is to determine the weight of each factor that influences land suitability. The presence of various and multiple criteria makes land suitability assessment complicated because factors that influence land suitability have unequal levels of significance (Elsheikh et al. 2013). This inequality of weight also varies by location, land use and productivity. The criteria for evaluation are largely dependent on geographical aspects and the socio-economic status of the country. A common rule for choosing a weight is very challenging, as growers have perceptions of weight that match their experiences.

A number of multi-criteria decision rules have been implemented to solve the land-use suitability problems. The decision rules can be classified into multi-objective and multi-attribute decision-making methods (Malczewski 1999, 2004). The multi-objective approaches are mathematical programming model-oriented methods such as linear programming. The single-objective multi-criteria evaluation has a ‘goal’ and computed using multi-attribute analysis. The methodology has several ways to weight the criteria such as, ordered weighted averaging (OWA) using weighted linear combination (WLC), analytic hierarchy process (AHP), and analytic network process (ANP). AHP method introduced by Satty in 1980 has incorporated into the GIS for land-use suitability analysis. As an extension of the criterion importance weighting in WLC, the OWA allows the decision-maker to specify a degree of risk in their approach to decision-making (Rinner and Voss 2013; Feizizadeh and Blaschke 2014). AHP method uses pairwise comparison of each criterion, while WLC directly assigns the weights of relative importance to each attribute map layer and OWA involves two-step weighting (criterion and order weights) (Ahmed 2015).

The analytical hierarchy process (AHP) is a multiple criteria decision-making process that uses analytical hierarchies to determine the importance of criteria and their associated relationships in complex problems (Brandt et al. 2015; Qureshi et al. 2017; Saaty 1980). The analytical hierarchy process has the advantage of assigning weights based on the preferences of experts for the regional concepts. For this reason, the AHP-modelling framework is widely accepted and has been extensively applied for multi-criteria decision analysis (MCDA) purposes and utilized in many decision-making problems regarding land suitability evaluation at a regional level (Zabihi et al. 2015; Akıncı et al. 2013; Zolekar and Bhagat 2015; Malczewski 2004).

Furthermore, GIS and AHP tools have recently been used for land suitability assessment and planning for suitable sites of agricultural land use, major crops and local foods (Pramanik 2016; Akinci et al. 2013; Bunruamkaew and Murayama 2011; Elsheikh et al. 2013; Zolekar and Bhagat 2015; Zabihi et al. 2015; and Widiatmaka 2016). In land suitability analysis, criteria associated with topographic features, vegetation and weather parameters are included. The extension and evaluation of suitability analysis methods can help to assess and improve the sustainability of crop production over time. Selecting the most appropriate model for land suitability assessment is important for current and future land use planning. Several approaches have been used to conduct land suitability assessments. The FAO land evaluation framework (1976) was the first procedure to assess local, regional, and national land-use planning. In recent years, computing technologies combined with GIS have included geospatial criteria to help find solutions for land suitability at the regional scale. Therefore, GIS, remote sensing and AHP can be used in land suitability analysis for various criteria related to ecological conditions or maximizing cassava production at the regional scale in Indonesia. Thus, the aim of this study was to develop a spatial model to assess land suitability levels for cassava production by integrating GIS, remote sensing and AHP.

2 Methodology

The model was built in three stages. First, Landsat-4 Thematic Mapper (TM), Landsat-8 Operational Land Imager (OLI) and Sentinel-2 Multispectral Instrument (MSI) satellites digital images and vector data layers were processed to establish criteria for the suitability analysis. Such criteria included land cover type, topographical features and the normalized difference vegetation index (NDVI). Second, we obtained highly suitable sites for increasing cassava production using GIS and AHP techniques. Third, we evaluated the sustainability levels of cassava production using four categories and images from the satellite database (Fig. 1). Primary data were collected through fieldwork involving questionnaires, interviews, and surveys. Additionally, secondary data from Statistics Indonesia and the Indonesian Geospatial Agency were used. A global positioning system (GPS) receiver was used in our field survey to determine the locations of cassava fields in the city of Serang and to provide ground truth information (Table 1).
Fig. 1

The framework of site suitability for cassava production

Table 1

List of data and original data sources used for land suitability assessment for cassava production

No.

Data

Description

Source

1

Land Use Map

Scale at 1:50.000

2011, Ministry of Environment and Forestry

2

NDVI Map

Extracted from 10-m resolution

2016, Sentinel-2 MSI

3

Slope Map

Derived from 30-m resolution

2015, DEM STRM

4

Elevation Map

Derived from 30-m resolution

2015, DEM STRM

5

Road Map

Scale 1:50.000

2005, Indonesia Geospatial Agency

6

River Map

Scale 1:50.000

2005, Indonesia Geospatial Agency

7

Rainfall Map

Scale 1:50.000

2010, Indonesia Geospatial Agency

8

Location of Market

GPS Data

2014, Survey

9

Cassava Field Location

GPS Data

2014, Survey

10

11

12

Cassava Production

Land Use/Cover Map 2010

Land Use/Cover Map 2016

Statistics Data

Derived from 30-m resolution

Derived from 30-m resolution

2014, Indonesian Statistics

2016, Landsat 4 TM

2016, Landsat 8 OLI

2.1 Study area

Geographically, the city of Serang is located at 5°99′–6°22′ south and 106°07′–106°25′ east. The city is bordered by the Java Sea to the north and is surrounded by the Serang Regency to the east, south, and west. The city of Serang holds a position as the central government of the Banten Province and is an alternative area for Indonesia’s state capital, Jakarta, which is located approximately 70 km away. The city includes six districts and 46 villages and covers a total area of 266.7 km2. Most of the area is flat land with an elevation of less than 500 meters and is characterized by a tropical climate (Fig. 2a–c). The city includes coastal land to the north, rural areas to the south and north and an urban area in the middle of the region. The urban area includes infrastructural facilities that support socio-economic development. Residences are also concentrated in the central part of the region. Rice cultivation constitutes the main land use in the northern area, whereas fields and dry land are found in the southern area. Cassava is an important alternative source of food, especially for traditional cuisine that is prepared for traditional events. In the city of Serang, cassava has historically been grown by poor farmers with minimal input on poorly managed land. When land is managed poorly, cassava can cause severe erosion on steep slopes (Howeler 1991).
Fig. 2

Geographical extent of study area. a Serang City, b Banten Province and c Indonesia

2.2 Criteria for suitability analysis

The criteria for the suitability analysis were land cover, slope angles, elevation levels, soil types, rainfall, distance from rivers, distance from roads and the vegetation index (Fig. 3a–h). The details of criteria’s characteristics focusing on the Serang city are given in the following sections.
Fig. 3

ah Criteria for land suitability analysis for cassava production

2.2.1 Land use/land cover

Land use and land cover (LULC) data files describe the vegetation, water, natural surfaces, and cultural features of a land surface (Akıncı et al. 2013). Most land in the city of Serang is covered by rice fields. Other areas include fields, settlements, forests, plantations, and water bodies. The LULC database was divided into four classes. Class I referred to fields with fertile soils that were easily cultivated for cassava. Class II land was used for rice cultivation with cassava intercropping. Class III referred to plantation and forested land on steep slopes, and class IV land was unsuitable for cassava cultivation due to the presence of settlements, residents, water bodies or mangrove forests.

2.2.2 Slope

In the city of Serang, most topography was classified as slopes between 0% and 45% in steepness. On slopes between 0% and 15%, most crops were easily cultivated. For cassava cultivation, slope angles were considered when determining cassava land management. Steep-sloped areas generally undergo soil erosion (Heumann et al. 2011), and soil steepness levels can affect soil formation. Additionally, a slope of 15% is optimal for livestock production and crop planting (FAO 2000). Land variety, in terms of slope angles, constitutes an important factor in determining the suitability of cassava production areas.

2.2.3 Distance from rivers

The Ci Banten River, the main river in Serang, supplies irrigation water. Other rivers in the area include the Cilandak, Cikaduan, Cikarang, Cipari, and Pelamunan Rivers. The physical factors associated with water supply, such as the distance from water bodies, streams, rivers, and irrigation zones, were used to determine suitability levels for cassava production. Rice fields were found in plains located close to major water resources, such as large rivers and water bodies, whereas cassava can be planted on sloped areas located farther from water resources.

2.2.4 Rainfall

Serang is characterized by a tropical climate, and significant periods of rainfall occur throughout the year. The average temperature and rainfall levels are 27.4 °C and 1500–2000 mm/year, respectively (BPS 2014). Cassava can also be intercropped with maize, legumes or rain-fed crops in areas of high and well-distributed rainfall (Devendra and Thomas 2002). Cassava can grow in areas that receive as little as 400 mm of average annual rainfall. However, higher yields have been obtained in the presence of greater water supplies (FAO 2013). Moisture stress on cassava roots can result in low yields, especially in years characterized by low rainfall. Therefore, irrigation management should be practiced effectively.

2.2.5 Soil

The major soil types found in Serang are alluvial, red regosol, red yellow podzolic, and latosol soils. Alluvial soils are mostly used in rice-based cropping systems, and regosol soils are used for upland rice and dry land crop cultivation. Regosol soils are found in hilly areas and in the center of mountain slopes. In Java, cassava-growing areas are generally located where soils classified as Mediterranean, alluvial, podzolic, latosols or regosols are found. According to Wargiono (2000), latosol areas are optimal for cultivating cassava. Latosol soils have good physical properties and are deep and tolerant to erosion. However, podzols include low levels of organic matter and tend to erode easily. Wargiono (2000) divided soil types for cassava cultivation into four classes. Class I includes latosol, gray hydromorphic, and planosol soils. Class II includes yellow podzolic soils. Class III refers to yellow regosol and red podzolic soils. Class IV refers to unsuitable soils that consist of gray alluvial hydromorphic soils with high water contents.

2.2.6 Elevation

In Asia, practically no cassava is grown at an elevation of 1000 meters above sea level. In Indonesia, most cassava-growing areas are located in the lowland humid and sub-humid tropics (Heumann et al. 2011). In some areas, cassava can be grown in hilly or mountainous areas, but the sustainability of these systems is compromised when sustained inputs are introduced for maintaining soil fertility and reducing erosion. Additionally, elevation has a strong effect on temperatures in some areas. In the city of Serang, elevation ranges from 12.5 to 375 m. Most of the area is suitable for cassava production, although the optimal elevation for cassava production is approximately 62.5–137.5 m.

2.2.7 Distance from roads

The number of vehicles in the city has increased due to economic growth, but road networks have not been expanded at the same rate. Therefore, traffic congestion in the city has increased. Regarding socio-economic factors, main roads are needed to sell fresh cassava at any distance from areas of cultivation. In selecting areas suitable for cassava production, the distance from roads must be considered because such distances affect transportation costs for supply processes. Shorter distances between fields and roads facilitate access to the transportation infrastructure and link farmers and farming activities to marketing channels.

2.2.8 Normalized difference vegetation index (NDVI)

To avoid soil erosion during cassava production, land covered by low vegetation can reduce the rate of surface runoff. Vegetation index variations were assessed using a satellite-based measure: the normalized difference vegetation index (NDVI). The NDVI is a vegetation index that is correlated with several important biophysical properties and that generates different crop indices (Ahamed et al. 2013; Elhag 2014). The proportion of vegetative biomass in the area being sensed or captured in satellite data is important for crop monitoring. Additionally, crop stages can be determined from NDVI data. In Indonesia, cassava production begins with planting at various times, but most field harvests occur during June or July. In this study, the NDVI was calculated for each cassava field using temporal information from Sentinel-2 MSI images acquired at the end of the growing period and before the harvest in May.

2.3 Digital image processing

We used image data for each criterion. A 1:50,000 map of land cover types, rainfall levels, distances from rivers, soil types, elevations, distances from roads, and NDVI data were used for the analysis. Basic vector data layers were collected from the Geospatial Information Agency of Indonesia. Landsat-4 TM, Landsat-8 OLI, and Sentinel-2 MSI vegetation index (VI) datasets were used for field-level crop monitoring in conjunction with NDVI data.

2.3.1 NDVI computation technique

The NDVI was proposed by Rouse et al. 1973, and it has become the most popular indicator for studying vegetation health and crop production. The NDVI is developed from two important wave bands: the red and near infrared (NIR) bands. It has been widely used for agricultural mapping and yield monitoring. The NDVI is calculated as follows:
$$ {\text{NDVI}} = \frac{{\text{RNIR}} - {\text{Rred}}}{{\text{RNIR}} + {\text{Rred}}} $$
(1)

We acquired all available cloud-free Sentinel-2 scenes and calculated the NDVI from band combinations corresponding to the red and NIR reflections using Band 4 and Band 8. The Sentinel-2 mission combines two satellites—Sentinel-2A and Sentinel-2B—equipped with identical multispectral instruments capable of acquiring data in 13 bands at different spatial resolutions (between 10 m and 60 m). These satellites provide continuity for the Satellite pour l’Observation de la Terre (SPOT) missions of the European Space Agency (ESA).

2.4 Reclassification of criteria

Reclassification was used to simplify or change our interpretation of raster data by changing a single value to a new value or by grouping ranges of values into single values. Each criteria source map was reclassified into four classifications. The classification used the following suitability classes: highly suitable (S1), moderately suitable (S2), marginally suitable (S3), and not suitable (N) (Fig. 4a–h). Spatial data were converted into raster layers and then processed in ArcGIS® (ESRI, USA). They were then classified into four classes as integer rasters that represented different suitability levels based on assigned threshold values (Table 2) (Tienwong et al. 2009).
Fig. 4

ah Reclassification of criteria

Table 2

Reclassification of criteria of land suitability assessment for cassava production

Criteria

Suitability class

Sub-criteria

Percentage area (%)

Area (ha)

LULC

S1

S2

S3

Class I

Class II

Class III

11.38

43.27

27.61

3059

11,631

7422

N

Class IV

17.74

4767

Slope (%)

S1

S2

S3

0–8%

8–15%

15–25%

83.81

10.25

3.07

22.352

2.734

818

N

> 25%

2.87

765

Rainfall (mm.)

S1

1000–1500

89.22%

23.794

S2

1500–2000

10.78%

2.875

Distance from roads (m)

S1

S2

S3

< 1000

1000–2000

2000–3000

88.31

10.51

1.11

23.794

2.803

296

N

> 3000

0.07

18

Distance from rivers (m)

S1

S2

S3

< 500

500–1000

1000–1500

72.4

20.76

4.66

19.309

5536

1.242

N

> 1500

2.18

581

Elevation (meters)

S1

S2

S3

12.5–62.5

62.5–137.5

137.5–212.5

76.93

17.14

4.14

20,517

4571

1104

N

212.5-337.5

1.79

477

Soil type

S1

S2

S3

Latosol

Podzolic

Regosol

37.93

21.36

20.48

10,115

5698

5462

N

Alluvial hydromorphic

20.23

5395

NDVI

S1

S2

S3

Vegetation

Rice field

Forest

10.06

13.83

43.94

1829

2514

7986

N

Waterbody, settlements

32.16

5845

For each of the suitability levels, we chose a suitability score. The suitability score is a way of computing values across the source layers so that there is a common standard. All source layer values are placed on the same scale with the same units. The same scale is used for all individual suitability layers and for the final overall suitability layer. In this study, we used a score of 9 for highly suitable areas, a score of 6 for moderately suitable areas, a score of 3 for marginally suitable areas, and a score of 1 and a restricted value for unsuitable areas.

2.5 Land suitability assessments

The land suitability assessment for the cassava production model was developed using the classification categories of land suitability proposed by the FAO (FAO 1976). The suitability classification is designed to determine the suitability of each land unit for a particular use. In the FAO’s framework for land evaluation, land, the first class, is designated as suitable (S) or not suitable (N). These suitability classes can then be further sub-divided as needed. In practice, three classes (S1, S2 and S3) are often used to identify land that is highly suitable, moderately suitable, or marginally suitable for cassava production. The AHP application was used to support our weighted overlay calculations in the GIS environment. The AHP results were obtained from experts in related fields and from literature reviews. Through this process, the consistency ratio (CR) was calculated and was used in the land suitability analysis. The AHP method was applied to determine the relative importance of all of the selected criteria and factors (Ahamed et al. 2013).

A set of questionnaires within the AHP framework was developed. In the questionnaire, respondents can determine the relative importance of each criterion with respect to others, for example, the importance of soil with respect to land use, water, roads and markets, and vice versa. Sets of questionnaires were disseminated to five experts with relevant backgrounds (cassava experts, agriculture experts, and agriculture planners) during the field trip. The AHP is widely used by decision-makers and researchers. Calculation of criteria weights is central in the AHP method and depends on experts’ opinions and determination for each criterion.

The study results are fully dependent on the applied AHP evaluation, how the criteria were defined and how the criteria were measured. The structured interviews were performed with relevant experts who were working for the cassava production in Indonesia for more than 10 years. Through this process, the consistency ratio (CR) was calculated and used in the land suitability analysis. The AHP method was applied to determine the relative importance of all of the selected criteria and factors (Ahamed et al. 2013). The total suitability score (Si) of each land unit was calculated using the following expression:
$$S_{i} = \mathop \sum \limits_{i = 1}^{n} W_{i} \times R_{i} .$$
(2)

2.5.1 Analytical hierarchy process (AHP)

Weights were used to determine the priorities of criteria (land cover, distance from rivers, rainfall levels, distance from roads, slope angles, elevation levels, soil types and vegetation index data) and to identify the suitability of different land uses for cassava production. The resultant AHP weights were used to determine the priority of each criterion for weighted overlay applications using GIS.

In the first stage of the analysis, we organized elements of the decision model into a hierarchy that included first level (goal), second level (criteria), and third level (alternative) elements. The first level involved selecting the goal. The second level of the hierarchy considered rules or criteria associated with the goal. The lowest level considered alternative decisions (Fig. 5).
Fig. 5

The AHP framework to select suitable areas for cassava production

The second phase involved scoring the criteria via pairwise comparisons and scoring scales of relative importance (Table 3). Questionnaires were used to gather expert opinions on the relative importance of the considered criteria and factors. Comparative results (for each factor pair) were described as integer values of 1 (equal value) to 9 (extremely different), where a higher number denotes that the chosen factor was considered to be more important than other factors to which it was compared. For example, when comparing land cover and slope angle criteria, a score of 1 indicates that both were equally relevant to evaluating suitability, and a score of 9 indicates that land cover is more important than the slope angle. All scores were assembled in a pairwise comparison matrix with diagonal and reciprocal scores located in the lower left-hand triangle. Reciprocal values (1/3, 1/5, 1/7, and 1/9) were used where the row criterion was found to be less important than the column criterion (Table 4).
Table 3

Preference scale for AHP pairwise comparison (Saaty 1989)

Scale

Degree of preference

Explanation

1

Equal importance

Two activities contribute equally to the objective

3

Moderate importance

Experience and judgments slightly favor one activity over another

5

Strong or essential importance

Experience and judgments strongly favor one activity over another

7

Very strong importance

An activity is favored very strongly over another

9

Extreme importance

The evidence favoring one activity over another is the highest possible order of affirmation

2,4,6,8

Intermediate values between two adjacent judgments

When compromise is needed

Reciprocals

Opposites

Used for inverse comparisons

Table 4

Pairwise comparison for the AHP model among the criteria selected for cassava production

 

Soil

Land Cover

Elevation

Slope

Rainfall

Distances from roads

River

NDVI

Soil

1

3

5

5

7

9

9

3

Land cover

0.33

1

3

3

7

7

9

1

Elevation

0.2

0.3

1

1

3

5

7

0.3

Slope

0.2

0.3

1

1

3

3

5

0.3

Rainfall

0.14

0.14

0.33

0.33

1

3

3

0.14

Distance from roads

0.11

0.14

0.2

0.33

0.33

1

1

0.14

Distance from rivers

0.11

0.11

0.14

0.2

0.33

1

1

0.11

NDVI

0.33

1

3

3

7

7

9

1

Third, we calculated the matrix and ensured the consistency of the pairwise comparison criteria. The AHP also provided measurements for calculating normalized values of each criterion and alternatives and for determining the normalized principal Eigen factors and priority vectors. The pairwise matrix was calculated and is given by the following expression:
$$\left[ {\begin{array}{*{20}c} {C_{11} } & {C_{12} } & {. . .C_{1n} } \\ {C_{21} } & {C_{22} } & {. . .C_{2n} } \\ {\begin{array}{*{20}c} . \\ . \\ C \\ \end{array}_{n1} } & {\begin{array}{*{20}c} . \\ . \\ C \\ \end{array}_{n2} } & { \begin{array}{*{20}c} . \\ . \\ . \\ \end{array} \begin{array}{*{20}c} . \\ . \\ C \\ \end{array}_{nn} } \\ \end{array} } \right].$$
(3)
The sum of each column of the pairwise matrix was denoted as follows:
$$C_{ij} = \mathop \sum \limits_{i = 1}^{n} C_{ij} .$$
(4)
We then divided each element of the matrix by its column total to generate a normalized pairwise matrix:
$$X_{ij} = \frac{{C_{ij} }}{{\mathop \sum \nolimits_{i = 1}^{n} C_{ij} }} = \left[ {\begin{array}{*{20}c} {X_{11} } & {X_{12} } & {. . .X_{1n} } \\ {X_{21} } & {X_{22} } & {. . .X_{2n} } \\ {\begin{array}{*{20}c} . \\ . \\ X_{n1} \\ \end{array} } & {\begin{array}{*{20}l} . \\ . \\ X_{n2} \end{array} } & { \begin{array}{*{20}c} . \\ . \\ . \\ \end{array} \begin{array}{*{20}c} . \\ . \\ X_{nn} \\ \end{array} } \\ \end{array} } \right].$$
(5)
Finally, we divided the sum of the normalized matrix column by the number of criteria used (n) to generate the weighted matrix of priority criteria:
$$W_{ij} = \frac{{\mathop \sum \nolimits_{j = 1}^{n} X_{ij} }}{n} = \left[ {\begin{array}{*{20}c} {W_{11} } \\ {W_{12} } \\ {\begin{array}{*{20}c} . \\ . \\ W_{1n} \\ \end{array} } \\ \end{array} } \right].$$
(6)
The initial consistency vectors were derived by multiplying the pairwise matrix by the vector of weights:
$$\left[ {\begin{array}{*{20}c} {C_{11} } & {C_{12} } & {. . .C_{1n} } \\ {C_{21} } & {C_{22} } & {. . .C_{2n} } \\ {\begin{array}{*{20}c} . \\ . \\ C_{n1} \\ \end{array} } & {\begin{array}{*{20}c} . \\ . \\ C_{n2} \\ \end{array} } & { \begin{array}{*{20}c} . \\ . \\ . \\ \end{array} \begin{array}{*{20}c} . \\ . \\ C_{nn} \\ \end{array} } \\ \end{array} } \right] \; \times \; \left[ {\begin{array}{*{20}c} {W_{11} } \\ {W_{12} } \\ {\begin{array}{*{20}c} . \\ . \\ W \\ \end{array}_{1n} } \\ \end{array} } \right] \; = \; \left[ {\begin{array}{*{20}c} {C_{11} W_{11} + } & {C_{12} W_{11} + } & {. . . + C_{13} W_{11} } \\ {C_{21} W_{12} + } & {C_{22} W_{12} + } & {. . . + C_{23} W_{12} } \\ {\begin{array}{*{20}c} . \\ . \\ {C_{n1} W_{1n} } \\ \end{array} } & {\begin{array}{*{20}c} . \\ . \\ {C_{n1} W_{1n} } \\ \end{array} } & { \begin{array}{*{20}c} . \\ . \\ . \\ \end{array} \begin{array}{*{20}c} . \\ . \\ {C_{n1} W_{1n} } \\ \end{array} } \\ \end{array} } \right]\; = \;\left[ {\begin{array}{*{20}c} {V_{11} } \\ {V_{12} } \\ {\begin{array}{*{20}c} . \\ . \\ V \\ \end{array}_{1n} } \\ \end{array} } \right].$$
(7)
The principal eigenvector (λmax) was then calculated by averaging the values of the consistency vector:
$$\lambda_{\hbox{max} } = \sum\limits_{i}^{n} {{\text{CV}}_{ij} } .$$
(8)
Eigen values were calculated by averaging the rows of each matrix. Eigen values were also referred to as relative weights. The largest Eigen value was equal to the number of criteria, and when λmax = n, judgments were consistent. Normalized Eigen values were generated as weights of priority criteria. The principle value suggests that eight criteria were consistent, as the calculation results reveal a maximum value of 8.34 (Table 5). The judgments were also checked to determine the consistency index (CI), which was calculated as:
$${\text{CI}} = \frac{{\lambda_{ \hbox{max} } - n}}{n - 1}$$
(9)
Table 5

Normalized matrix for the criteria selected for cassava production

 

Soil

Land Cover

Elevation

Slope

Rainfall

Distance from roads

Distance from rivers

NDVI

Total

Average

Consistency measure

Soil

0.413

0.500

0.365

0.360

0.244

0.25

0.204

0.500

2.840

0.355

8.649

Land cover

0.136

0.166

0.219

0.216

0.244

0.194

0.204

0.166

1.549

0.193

8.621

Elevation

0.082

0.050

0.073

0.072

0.104

0.138

0.159

0.050

0.730

0.091

8.238

Slope

0.082

0.050

0.073

0.072

0.104

0.083

0.113

0.050

0.629

0.078

8.411

Rainfall

0.057

0.023

0.024

0.023

0.034

0.083

0.068

0.023

0.338

0.042

7.977

Distance from roads

0.045

0.023

0.014

0.023

0.011

0.027

0.022

0.023

0.192

0.024

8.167

Distance from rivers

0.045

0.018

0.010

0.014

0.011

0.027

0.022

0.018

0.168

0.021

8.023

NDVI

0.136

0.166

0.219

0.216

0.244

0.194

0.204

0.166

1.549

0.193

8.621

Here n is the total number of criteria. Saaty (1989) also introduced the consistency ratio (CR) and compared it to the consistency index and the random index (RI) value, which is the calculated value for matrices of different sizes (Table 6). The consistency ratio was calculated as:
$$ {\text{CR}} = \frac{{\text{CI}}}{{\text{RI}}} . $$
(10)
Table 6

Random consistency index (RI) to determine consistency ratio (CR) (Saaty 1989)

N

1

2

3

4

5

6

7

8

9

10

RI

0

0

0.58

0.9

1.12

1.24

1.32

1.41

1.45

1.49

A lower CR ratio indicates a higher degree of consistency. Further confirmation and understanding about weight and influences among the criteria, ANP also employed in this research.

2.5.2 Analytical network process (ANP)

ANP is an extension of the AHP and proposed by Satty 1980 and 2001. ANP is a nonlinear structure with bilateral relationships (Azizi et al. 2014). In this research, ANP was used to obtain the weight of the criteria to compare with the weight from AHP. In the ANP analysis, first, the construction of a conceptual model was developed to determine relationships among the criteria and alternatives. If no relationship exists among the criteria, then there is influence among the criteria and alternatives. The, criteria were compared pairwise by Super Decisions Software® to form an un-weighted super-matrix. Then, the priorities derived from pairwise comparison matrices were entered as parts of the columns referred as the evaluation matrix U for criteria (C1, C1, C3, C4, C5, C6, C7, C8) and alternatives (A1, A2, A3, A4). The evaluation matrix for the criteria can be expressed as follows:
$$U = \left[ {\begin{array}{*{20}c} {U_{11} } & {U_{12} } & {. . .U_{18} } \\ {U_{21} } & {U_{22} } & {. . .U_{28} } \\ {\begin{array}{*{20}c} . \\ . \\ U \\ \end{array}_{41} } & {\begin{array}{*{20}c} . \\ . \\ U \\ \end{array}_{42} } & { \begin{array}{*{20}c} . \\ . \\ . \\ \end{array} \begin{array}{*{20}c} . \\ . \\ U \\ \end{array}_{48} } \\ \end{array} } \right].$$
(11)
In contrast, the evaluation matrix V in which alternatives (A1, A2, A3, A4) are evaluating according to the criteria (C1, C1, C3, C4, C5, C6, C7, C8) can be expressed as follows:
$$V = \left[ {\begin{array}{*{20}c} {V_{11} } & {V_{12} } & {V_{13} V_{14} } \\ {V_{21} } & {V_{22} } & {V_{23} V_{24} } \\ {\begin{array}{*{20}c} . \\ . \\ V \\ \end{array}_{81} } & {\begin{array}{*{20}c} . \\ . \\ V \\ \end{array}_{82} } & { \begin{array}{*{20}c} . \\ . \\ { V_{83} } \\ \end{array} \begin{array}{*{20}c} . \\ . \\ V \\ \end{array}_{84} } \\ \end{array} } \right].$$
(12)
Then, the weighted supermatrix is expressed as a function of the evaluation matrices U and V. The supermatrix S should be a probability matrix and irreducible. The weighted supermatrix can be expressed follows:
After that, limit supermatrix was obtained by raising the weighted supermatrix to powers by multiplying the matrix itself (Table 8). The limit supermatrix can be expressed as follows:
$$S_{\text{limited}} \; = \;\mathop {\lim }\limits_{n \to \propto } S_{\text{weighted}}^{n} .$$
(14)
Examples of weighted supermatrix and limit supermatrix are given to show the relations among the criteria and alternatives for one expert’s opinion (Tables 7 and 8). At the end, the weighted overlay approach was used for applying a weight priority of the criteria to generate a land suitability map for cassava production in the GIS environment.
Table 7

The weighted supermatrix for determining the relation among the criteria and alternatives for cassava production using ANP

Alternatives

Criteria

  

N

S1

S2

S3

Elevation

LULC

NDVI

Rainfall

River

Road

Slope

Soil

Alternatives

N

0.000

0.000

0.000

0.000

0.055

0.055

0.055

0.055

0.055

0.055

0.055

0.055

S1

0.000

0.000

0.000

0.000

0.565

0.565

0.565

0.565

0.565

0.565

0.565

0.565

S2

0.000

0.000

0.000

0.000

0.262

0.262

0.262

0.262

0.262

0.262

0.262

0.262

S3

0.000

0.000

0.000

0.000

0.118

0.118

0.118

0.118

0.118

0.118

0.118

0.118

Criteria

Elevation

0.089

0.090

0.089

0.089

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

LULC

0.194

0.194

0.194

0.194

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

NDVI

0.194

0.194

0.194

0.194

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Rainfall

0.040

0.035

0.040

0.040

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

River

0.020

0.023

0.020

0.020

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Road

0.023

0.024

0.023

0.023

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Slope

0.078

0.079

0.078

0.078

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Table 8

The limited supermatrix for determining the relation among the criteria and alternatives for cassava production using ANP

Alternatives

Criteria

  

N

S1

S2

S3

Elevation

LULC

NDVI

Rainfall

River

Road

Slope

Soil

Alternatives

N

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

S1

0.283

0.283

0.283

0.283

0.283

0.283

0.283

0.283

0.283

0.283

0.283

0.283

S2

0.131

0.131

0.131

0.131

0.131

0.131

0.131

0.131

0.131

0.131

0.131

0.131

S3

0.059

0.059

0.059

0.059

0.059

0.059

0.059

0.059

0.059

0.059

0.059

0.059

Criteria

Elevation

0.045

0.045

0.045

0.045

0.045

0.045

0.045

0.045

0.045

0.045

0.045

0.045

LULC

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

NDVI

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

0.097

Rainfall

0.018

0.018

0.018

0.018

0.018

0.018

0.018

0.018

0.018

0.018

0.018

0.018

River

0.011

0.011

0.011

0.011

0.011

0.011

0.011

0.011

0.011

0.011

0.011

0.011

Road

0.012

0.012

0.012

0.012

0.012

0.012

0.012

0.012

0.012

0.012

0.012

0.012

Slope

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

Soil

0.181

0.181

0.181

0.181

0.181

0.181

0.181

0.181

0.181

0.181

0.181

0.181

2.5.3 GIS analysis

Suitability assessment criteria were used as the reclassified raster data layers for land cover, slope angles, elevation levels, soil types, rainfall levels, distance from rivers, distance from roads and the vegetation index. All of the reclassified raster data were combined with weighted overlay tools. This reclassification was used to simplify or change our interpretation of raster data by changing a single value to a new value or by grouping ranges of values into single values. Each criteria source map was reclassified into four classifications. The classification used the following suitability classes: highly suitable (S1), moderately suitable (S2), marginally suitable (S3), and not suitable (N). Spatial data were converted into raster layers and were then processed in ArcGIS® (ESRI, USA). They were then classified into four classes as integer rasters that represented different suitability levels based on the assigned threshold values (Tienwong et al. 2009). Weighted overlays are overlay analysis tools used to identify the best or most preferable locations for cassava production. The criteria included in the weighted overlay analysis were not equal in importance. The weights of key criteria were calculated using the AHP/ANP application. Using the reclassification and weighted overlay method, a spatial analysis was conducted, and a suitability map for cassava production was created (Eckert and Sujata 2011; Gatrell et al. 2011).

2.6 Ground truth information and field survey

Primary data from the field survey were collected through questionnaires, interviews, and surveys. GPS data for cassava production locations were collected. Ground references were collected to determine the locations of cassava fields located in the highly suitable areas of Serang city. A field survey and a focus group discussion held with cassava farmers and regional sellers were completed in November 2016.

2.7 Sustainability evaluation

Several indicators and frameworks are commonly used for sustainability evaluation (Ahamed et al. 2009; Bell and Morse 2008; Ahamed et al. 2015; Von Wirén-Lehr 2001). In this study, we focused on pillars of agro-ecological sustainability indicators that are related to ecological, social, and economic factors and are associated with several criteria, such as availability, accessibility, affordability, and profitability. These criteria were considered to evaluate the sustainability of cassava production between 2010 and 2015 (Fig. 6).
Fig. 6

Criteria of sustainability evaluation of cassava production

3 Results

In the GIS analysis, the reclassified rasters were used with AHP and ANP weights and ranked accordingly. The CR was the indicator of judgments to refer to the AHP weight, whether consistent or not. In the AHP analysis, a CR of 6.1% was found, which was less than 10%, referred the consistency of expert opinions was acceptable. Among the eight sub-criteria identified, the AHP application ranked soil as the first priority (34%) followed by land cover (18%), the vegetation index (16%), rainfall (11%), elevation level (8%), slope (7%), distance from roads (3%), and distance from rivers (3%) when selecting suitable lands for cassava (Table 9). The ANP model also included a consistency test and observed 6.3%, which was also less than 10% to assess the degree of consistency of the experts. The ANP application ranked soil as the first priority (36%) followed by land cover (18%), the vegetation index (14%), rainfall (11%), elevation (8%), slope (6%), distance from rivers (4%) and distance from roads (3%) (Table 9).
Table 9

Priority criteria weights according to expert’s opinions for selecting land suitability in cassava production

Criterion names

Weights of criterion

Experts initials

(Experience of experts, years)

Expert A

Expert B

Expert C

Expert D

Expert E

Mean

(11 years)

(10 years)

(20 years)

(21 years)

(15 years)

AHP

ANP

AHP

ANP

AHP

ANP

AHP

ANP

AHP

ANP

AHP

ANP

Soil

0.356

0.387

0.408

0.436

0.339

0.345

0.355

0.361

0.244

0.246

0.340

0.355

LULC

0.214

0.223

0.181

0.175

0.198

0.198

0.194

0.194

0.102

0.100

0.178

0.178

NDVI

0.184

0.156

0.170

0.165

0.198

0.198

0.194

0.089

0.067

0.064

0.162

0.134

Elevation

0.109

0.100

0.085

0.069

0.099

0.096

0.091

0.089

0.034

0.032

0.083

0.077

Slope

0.074

0.059

0.072

0.037

0.080

0.080

0.079

0.079

0.038

0.039

0.069

0.059

Rainfall

0.031

0.036

0.042

0.022

0.043

0.040

0.042

0.037

0.398

0.407

0.111

0.108

Road

0.014

0.022

0.023

0.019

0.024

0.023

0.024

0.024

0.054

0.052

0.028

0.028

River

0.020

0.017

0.021

0.078

0.021

0.020

0.021

0.022

0.063

0.061

0.029

0.040

CR

0.080

0.080

0.058

0.065

0.033

0.039

0.043

0.040

0.091

0.091

0.061

0.063

The weighted overlay was used for applying a weight priority of the criteria to generate the land suitability map for cassava production. The reclassified raster data layers of land cover, slope angles, elevation levels, soil types, rainfall, distance from rivers, distance from roads and the vegetation index were combined with weighted overlay tools and AHP/ANP weights to generate suitability map (Fig. 7). A suitability map for cassava production was created from a weighted overlay, and we found in the AHP analysis that 41.60% (11094 ha) of the study area was highly suitable for cassava production, 30.87% (8233 ha) was moderately suitable and 9.83% (2623 ha) was marginally suitable. Whereas, the result of ANP analysis found that 44.62% (11901 ha) of the study area was highly suitable for cassava production, 27.17% (7246 ha) was moderately suitable and 10.51% (2803 ha) was marginally suitable. Additionally, the same result of AHP and ANP show 17.69% (4718 ha) of the land area was found occupied by residences and settlements (Fig. 7 and Table 10). Highly suitable areas for cassava production covered 41.60% (11094 ha) of the total area of the Serang city. These areas were mainly dry lands with moderately well-drained soils. Soils in this group were loamy with topsoil that was leveled and bounded for paddy rice. There is high possibility to use these areas and can be used to grow cassava after they are drained to avoid waterlogging. The moderately suitable area covered 30.87% (8233 ha) of the total area of Serang. These areas were poorly drained and coarsely textured with alluvial terraces. Marginally suitable areas for cassava production cannot support cassava plantations. Only 9.83% (2623 ha) of the land area was categorized as marginally suitable. Deep and coarsely textured soils positioned on slopes of less than 20% of the mentioned areas. Soil fertility levels were moderately low. Upland crops and fruit trees are often found with low levels of fertility, a lack of water during dry seasons, soil erosion on steep slopes, and high levels of acidity in some areas.
Fig. 7

a, b Land suitability distribution using a weighted overlay

Table 10

Suitable area for cassava production in the Serang City, Banten Province, Indonesia

Suitability Class

AHP

ANP

Percentage area (%)

Area (ha)

Percentage area (%)

Area (ha)

Highly suitable

41.60

11094

44.62

11901

Moderately suitable

30.87

8233

27.17

7246

Marginally suitable

9.83

2623

10.51

2803

Not suitable

17.69

4718

17.69

4718

The 4718 ha (17.69%) area of land that was classified as unsuitable for cassava production due to the presence of settlements and residences cannot be replaced with cassava fields. This area included the coastal area in the northern part of Serang and is characterized by sandy soils with high mineral contents. Although cassava can grow under high nitrogen (N), potassium (K) and organic matter (OM) application conditions, to obtain high-quality yields, appropriate management strategies must be applied to boost cassava production in coastal areas. The weighted overlay map used to locate suitable cassava production areas could serve as a reference map for predicting production methods that could support measures to increase local food production in the city of Serang. According to the GPS locations for cassava production recorded in November 2016, most cassava-growing areas were concentrated in the southern part of the region (Fig. 8).
Fig. 8

a, b Validation of land suitability analysis using ground reference information

In the sustainability evaluation, several sub-criteria (e.g., land use, production, population, distance, market, price, productivity, and income) were considered. These data were collected from primary and secondary sources. Over the period examined, production and land use were unsustainable due to a shift from agricultural to settlement land use. Although cassava production has been located in the most suitable areas, we found that the land of cassava fields from 2010 to 2015 decreased 3.38% annually based on our collected data (Table 11). Furthermore, the NVDI images based on Landsat-4 TM and Landsat-8 OLI showed the vegetation conditions, which reflect the land use change and physical features that cover the Earth’s surface (land cover) (Fig. 9). Most land in the city of Serang was cultivated land with plantation fields, irrigated paddy fields, and rain-fed areas. Additionally, protected areas were occupied by settlements.
Table 11

Agricultural data assessment from 2010 to 2015 for Serang City, Banten Province, Indonesia (BPS 2017)

Factors

Sub-factors

2010

2011

2012

2013

2014

2015

Trend of change (%)

Annual rate of change (%)

Availability

Land use (Ha)

321

253

327

391

211

62

− 67.62

− 3.38

Production (ton)

4600

3289

4400

6374

3175

4162

− 5.00

− 0.24

Accessibility

Population (people)

577785

598407

611897

618802

631101

643205

5.35

0.26

Road condition (%)

54.87

54.87

54.87

52.25

55.50

53.83

− 0.95

− 0.04

Affordability

Traditional market (unit)

6

6

6

6

6

6

0

0

Cassava price (USD/kg)

0.075

0.075

0.075

0.09

0.12

0.12

23.07

1.15

Profitability

Productivity (ton/ha)

14.33

14.52

14.58

15.33

15.03

67.12

64.81

3.24

 

Farmer income (USD/kg)

0.036

0.076

0.114

0.152

0.152

0.152

61.70

3.08

Fig. 9

a, b Land use changes in Serang City Drawn from Landsat

4 Discussion

We found that most land areas suitable for cassava production were located in the southern part of Serang in the Banten province. This result could be because the soil steepness levels in this area are less than 15%, and this condition could affect soil formation. From the ground truth survey, cassava farmers can grow cassava in rotation with other crops to prevent depletion of nutrients from soil. The production of cassava in new areas has faced several barriers, especially regarding labor and the conversion of peat land and forests in agricultural areas. Future yields can be maximized through the implementation of several management practices (e.g., minimum tillage, contour ridging, fertilization, strip-cropping, and intercropping with government support and rural appraisals from experts). Our study results illustrate the effectiveness of spatial assessments for evaluating suitable land use for sustainable cassava production. Therefore, geospatial technologies that combine GIS, remote sensing and AHP could be used to support land suitability assessments of cassava production. Geospatial modeling has limitations in obtaining highly accurate validation results due to a lack of ground reference information of previous years. As such, future studies should integrate several indicators based on high-resolution spatial and temporal remote sensing data.

Furthermore, this empirical method accepted key input from experts through AHP-based questionnaires and structured questionnaire surveys for cassava growers and agricultural production officers in the study area, which significantly enhanced the decision-making capabilities of the land-use plan. However, the AHP method has limitations in that it employs suitability determinations that can be subject to bias in both the scope and quality of outputs for the variation of weights. We thought of many ways to provide equal weights after fieldwork was conducted extensively in the city of Serang. Inequality usually varies for site-specific cases and crop selection (such as with cassava) in regional contexts. The judgment of pertinent criteria is complicated, and there are preferences of priority among the criteria. In such a case, AHP has the advantage of weighting the criteria based on experts’ opinions. However, it is very difficult to judge the subjectivity of decision-making during the modeling stages. To overcome the limitation and influences of criteria, we have also employed ANP for further confirmation of weights. Additionally, consistency ratio was introduced for AHP and ANP to validate the judgment of experts. The consistency ratio indicates the degree of coincidence between the AHP or ANP models and experts’ opinions for weighting the criteria in the model. The weights were given to identify the preferences of criteria to analyze in the GIS environment.

In the GIS analysis, weights from AHP and ANP were used to develop the weighted overlay using the criteria. The ground truth information validated the weighted overlay and confirmed the suitable locations of cassava fields in the Serang city. Most of the fields were located in the highly suitable areas and some were in the marginally suitable areas. The validation was required to understand spatial variability of cassava production for regional perspective and identify the causes of decreasing production of cassava. Along with spatial variability, socio-economic factors should be included for increasing cassava production. Present research shows the results of suitable areas for cassava production in the Serang city to establish cassava as an alternate crop to minimize the climate risk of rice production in Indonesia.

5 Conclusions

This study identified suitable areas to evaluate the sustainability of land use for cassava production using a multi-criteria model integrating with GIS, remote sensing and AHP. The multi-criteria model for suitability assessment used eight criteria: LULC, rainfall, distance from rivers, slope angle, elevation level, soil type, distance from roads and NDVI. From these criteria, we found that priority criteria, such as the soil type, LULC, and NDVI, influenced the sustainability of cassava production. All of the criteria were processed through a weighted overlay using AHP to calculate the weights of each criterion. To cut on the bias of AHP, the results also confirmed with the ANP. The land suitability assessment for cassava production indicated that 41.6 and 44.6% of the study area was highly suitable using AHP and ANP, respectively. Furthermore, the sustainability of cassava production was analyzed using several indicators classified into four categories: availability, accessibility, affordability, and profitability. The results show that the land use for cassava cultivation areas declined annually 3.38% between 2010 and 2015. The results obtained from this research are very significant in the decision-making processes to increase the production of cassava in suitable areas of the Serang city. The production scenario is one of the most important points for the suitability understanding for increasing regional production of cassava in Indonesia. The model can be further expanded spatially by including a fuzzy approach with AHP and ANP to overcome the limitation of the multi-criteria model.

Notes

Acknowledgements

We would like to thank the University of Tsukuba to support this research to develop the multi-criteria modeling for land suitability analysis for Cassava Production in Indonesia. We also express our sincere thanks to the, Indonesian Geospatial Agency, the United States Geological Survey (USGS) and European Space Agency (ESA) for geographical and satellite data information. We sincerely thank the Indonesia Endowment Fund for Education (LPDP) for providing scholarship to continue this research in Japan. We also expressed our gratitude to Indonesian experts and field surveyors to participate in this research.

References

  1. Ahamed T, Khan MIN, Takigawa T, Koike M, Tasnim F, Zaman JMQ (2009) Resource management for sustainable development: a community-and GIS-based approach. Environ Dev Sustain 11(5):933–954.  https://doi.org/10.1007/s10668-008-9159-y CrossRefGoogle Scholar
  2. Ahamed T, Noguchi R, Takigawa T, Tian L (2013) Bioproduction engineering: a road map for sustainable agricultural practices. Nova, New YorkGoogle Scholar
  3. Ahamed T, Noguchi R, Takigawa T (2015) Sustainability: integrating environment, agriculture and renewable energy for food security. Nova Publisher, New York, pp 10–11Google Scholar
  4. Ahmed B (2015) Landslide susceptibility mapping using multi-criteria evaluation techniques in Chittagong Metropolitan Area, Bangladesh. Landslides 12(6):1077–1095CrossRefGoogle Scholar
  5. Akıncı H, Özalp AY, Turgut B (2013) Agricultural land use suitability analysis using GIS and AHP technique. Comput Electron Agric 97:71–82.  https://doi.org/10.1016/j.compag.2013.07.006 CrossRefGoogle Scholar
  6. Ariningsih E (2018) Cluster-Based Cassava Production Improvement in West Java and South Sulawesi Provinces. AKP 14(2):125–148Google Scholar
  7. Azizi A, Malekmohammadi B, Jafari HR, Nasiri H, Parsa VA (2014) Land suitability assessment for wind power plant site selection using ANP-DEMATEL in a GIS environment: case study of Ardabil province, Iran. Environ Monit Assess 186(10):6695–6709CrossRefGoogle Scholar
  8. Bell S, Morse S (2008) Sustainability indicators, measuring the immeasurable?. Earthscan, LondonGoogle Scholar
  9. BPS (2017) Regional Statistics Banten Province 2017 Badan Pusat Statistik (The Central Statistics Agency of Indonesia). BPS, BantenGoogle Scholar
  10. Brandt P, Kvakić M, Butterbach-Bahl K, Rufino MC (2015) How to target climate-smart agriculture? Concept and application of the consensus-driven decision support framework “targetCSA”. Agric Syst.  https://doi.org/10.1016/j.agsy.2015.12.011 Google Scholar
  11. Bunruamkaew K, Murayam Y (2011) Site suitability evaluation for ecotourism using GIS & AHP: a case study of Surat Thani province, Thailand. Procedia Soc Behav Sci 21:269–278.  https://doi.org/10.1016/j.sbspro.2011.07.024 CrossRefGoogle Scholar
  12. Campo BVH, Hyman G, Bellotti A (2011) Threats to cassava production: known and potential geographic distribution of four key biotic constraints. Food Secur 3(3):329.  https://doi.org/10.1007/s12571-011-0141-4 CrossRefGoogle Scholar
  13. Ceballos-Silva Alejandro, Lopez-Blanco Jorge (2003) Delineation of suitable areas for crops using a multi-criteria evaluation approach and land use/cover mapping: a case study in Central Mexico. Agric Syst 77(2):117–136.  https://doi.org/10.1016/S0308-521X(02)00103-8 CrossRefGoogle Scholar
  14. Devendra C, Thomas D (2002) Smallholder farming systems in Asia. Agric Syst 71(1):17–25.  https://doi.org/10.1016/S0308-521X(01)00033-6 CrossRefGoogle Scholar
  15. Eckert J, Shetty S (2011) Food systems, planning and quantifying access: using GIS to plan for food retail. Appl Geogr 31(4):1216–1223.  https://doi.org/10.1016/j.apgeog.2011.01.011 CrossRefGoogle Scholar
  16. Elhag M (2014) Sensitivity analysis assessment of remotely based vegetation indices to improve water resources management. Environ Dev Sustain 16(6):1209–1222.  https://doi.org/10.1007/s10668-014-9522-0 CrossRefGoogle Scholar
  17. Elsheikh R, Shariff ARBM, Amiri F, Ahmad NB, Balasundram SK, Soom MAM (2013) Agriculture land suitability evaluator (ALSE): a decision and planning support tool for tropical and subtropical crops. Comput Electron Agric 93:98–110.  https://doi.org/10.1016/j.compag.2013.02.003 CrossRefGoogle Scholar
  18. FAO (1976) A framework for land evaluation, 1st edn. FAO, RomeGoogle Scholar
  19. FAO (2000) Land resources information systems in Asia. World soil resource reports. FAO, RomeGoogle Scholar
  20. FAO (2013) Save and grow: Cassava A guide to sustainable production intensification. FAO, RomeGoogle Scholar
  21. Feenstra GW (1997) Local food systems and sustainable communities. Am J Altern Agric 12(01):28–36.  https://doi.org/10.1017/S0889189300007165 CrossRefGoogle Scholar
  22. Feizizadeh B, Blaschke T (2014) An uncertainty and sensitivity analysis approach for GIS-based multicriteria landslide susceptibility mapping. Int J Geogr Inf Sci 28(3):610–638CrossRefGoogle Scholar
  23. Ferretti V, Pomarico S (2013) An integrated approach for studying the land suitability for ecological corridors through spatial multicriteria evaluations. Environ Dev Sustain 15(3):859–885.  https://doi.org/10.1007/s10668-012-9400-6 CrossRefGoogle Scholar
  24. Gatrell JD, Reid N, Ross P (2011) Local food systems, deserts, and maps: the spatial dynamics and policy implications of food geography. Appl Geogr 31(4):1195–1196.  https://doi.org/10.1016/j.apgeog.2011.01.013 CrossRefGoogle Scholar
  25. Heumann BW, Walsh SJ, McDaniel PM (2011) Assessing the application of a geographic presence-only model for land suitability mapping. Ecological Inform 6(5):257–269.  https://doi.org/10.1016/j.ecoinf.2011.04.004 CrossRefGoogle Scholar
  26. Howeler RH (1991) Long-term effect of cassava cultivation on soil productivity. Field crops Res 26(1):1–18.  https://doi.org/10.1016/0378-4290(91)90053-X CrossRefGoogle Scholar
  27. Khumaida N, Ardie SW, Sopandie D (2016) Influence of Agro-ecology on growth and performance of several potential mutants of Cassava. Procedia Environ Sci 33:70–77CrossRefGoogle Scholar
  28. Khush GS (2005) What it will take to feed 5.0 billion rice consumers in 2030. Plant Mol Biol 59(1):1–6.  https://doi.org/10.1007/s11103-005-2159-5 CrossRefGoogle Scholar
  29. Kolawole PO, Agbetoye L, Ogunlowo SA (2010) Sustaining world food security with improved cassava processing technology: the Nigeria experience. Sustainability 2(12):3681–3694.  https://doi.org/10.3390/su2123681 CrossRefGoogle Scholar
  30. Malczewski J (1999) GIS and multicriteria decision analysis. Wiley, New YorkGoogle Scholar
  31. Malczewski J (2004) GIS-based land-use suitability analysis: a critical overview. Prog Plan 62(1):3–65CrossRefGoogle Scholar
  32. Malczewski Jacek (2006) GIS-based multicriteria decision analysis: a survey of the literature. Int J Geogr Inf Sci 20(7):703–726.  https://doi.org/10.1080/13658810600661508 CrossRefGoogle Scholar
  33. Noerwijati K, Budiono R (2015) Yield and yield components evaluation of cassava (Manihot esculenta Crantz) clones in different altitudes. Energy Procedia 65:155–161.  https://doi.org/10.1016/j.egypro.2015.01.050 CrossRefGoogle Scholar
  34. Pramanik MK (2016) Site suitability analysis for agricultural land use of Darjeeling district using AHP and GIS techniques. Model Earth Syst Environ 2(2):56CrossRefGoogle Scholar
  35. Qureshi MRN, Singh RK, Hasan MA (2017) Decision support model to select crop pattern for sustainable agricultural practices using fuzzy MCDM. Environ Dev Sustain 1–19.  https://doi.org/10.1007/s10668-016-9903-7
  36. Rinner C, Voss S (2013) MCDA4ArcMap–an open-source multi-criteria decision analysis and geovisualization tool for ArcGIS 10. Cartouche Newslett Can Cartogr Assoc 86:12–13Google Scholar
  37. Rouse JW, Haas RH, Schell JA, Deering DW (1973) Monitoring vegetation systems in the Great Plains with ERTS, Third ERTS Symposium, NASA SP-351 I, 309-317Google Scholar
  38. Saaty (1980) Decision making for leaders: the analytic hierarchy process in a complex world. RWS, PittsburghGoogle Scholar
  39. Saaty TL (1989) Group decision making and the AHP. In: Golden BL, Wasil EA, Harker PT (eds) The analytic hierarchy process. Springer, Berlin, HeidelbergGoogle Scholar
  40. Smyth AJ, Dumanski J (1993) FESLM: an international framework for evaluating sustainable land management. FAO, Rome, p 76Google Scholar
  41. Sydorovych O, Wossink A (2008) The meaning of agricultural sustainability: evidence from a conjoint choice survey. Agric Syst 98(1):10–20.  https://doi.org/10.1016/j.agsy.2008.03.001 CrossRefGoogle Scholar
  42. Tienwong K, Dasananda S, Navanugraha C (2009) Integration of land evaluation and the analytical hierarchical process method for energy crops in Kanchanaburi, Thailand. ScienceAsia 35:170–177.  https://doi.org/10.2306/scienceasia1513-1874.2009.35.170 CrossRefGoogle Scholar
  43. Tiwari DN, Loof R, Paudyal GN (1991) Environmental–economic decision-making in lowland irrigated agriculture using multi-criteria analysis techniques. Agric Syst 60(2):99–112.  https://doi.org/10.1016/S0308-521X(99)00021-9 CrossRefGoogle Scholar
  44. Von Wirén-Lehr S (2001) Sustainability in agriculture—an evaluation of principal goal-oriented concepts to close the gap between theory and practice. Agric Ecosyst Environ 84(2):115–129.  https://doi.org/10.1016/S0167-8809(00)00197-3 CrossRefGoogle Scholar
  45. Vrieling A, de Beurs KM, Brown ME (2011) Variability of African farming systems from phenological analysis of NDVI time series. Clim Change 109(3–4):455–477.  https://doi.org/10.1007/s10584-011-0049-1 CrossRefGoogle Scholar
  46. Wargiono J, Sudaryanto B (2000) Cassava leaves and forage crops for ruminant feed in the establishment of sustainable cassava farming system in Indonesia. In National Workshop-Seminar on Sustainable Livestock Production on Local Feed Resources. pp. 496–503. http://ciat-library.ciat.cgiar.org/Articulos_Ciat/proceedings_workshop_02/496.pdf
  47. Widiatmaka W (2016) Integrated use of GIS, AHP and remote sensing in land use planning for tropical high altitude vegetable crops. J Appl Hortic 18:87–99Google Scholar
  48. Zabihi H, Ahmad A, Vogeler I, Said MN, Golmohammadi M, Golein B, Nilashi M (2015) Land suitability procedure for sustainable citrus planning using the application of the analytical network process approach and GIS. Comput Electron Agric 117:114–126.  https://doi.org/10.1016/j.compag.2015.07.014 CrossRefGoogle Scholar
  49. Zolekar RB, Bhagat VS (2015) Multi-criteria land suitability analysis for agriculture in hilly zone: remote sensing and GIS approach. Comput Electron Agric 118:300–321.  https://doi.org/10.1016/j.compag.2015.09.016 CrossRefGoogle Scholar

Copyright information

© The Japan Section of the Regional Science Association International 2018

Authors and Affiliations

  • Riska Ayu Purnamasari
    • 1
  • Tofael Ahamed
    • 2
  • Ryozo Noguchi
    • 2
  1. 1.Graduate School of Life and Environmental SciencesUniversity of TsukubaTsukubaJapan
  2. 2.Faculty of Life and Environmental SciencesUniversity of TsukubaTsukubaJapan

Personalised recommendations