Land suitability assessment for Paulownia cultivation using combined GIS and Z-number DEA: A case study

Highlights

• A hybrid GIS-ZDEA method is proposed to determine suitable locations for Paulownia cultivation.

• The feasibility and suitability maps for Paulownia cultivation are derived based on three categories of criteria.

• The high degree of fluctuation in weather condition is modelled by Z-number.

• The computational experiments demonstrate that the proposed model can improve the results by 8%

Abstract

Paulownia has emerged as a prospering biomass resource for the production of renewable energy. Many countries have decided to cultivate this tree in order to reduce air pollution and secure the increasing energy demand. In this paper, a hybrid approach, including geographical information system (GIS) and mathematical modeling is proposed to determine suitable locations for Paulownia cultivation. The feasibility and suitability maps for Paulownia cultivation have been derived based on three categories of criteria, including (1) Certain Non-Compensatory Criteria (CNCC) (2) Uncertain Compensatory Criteria (UCC), and (3) Certain Compensatory Criteria (CCC), and a four-stage algorithm is proposed. This algorithm sets out to (1) recognize Paulownia growth conditions (2) identify feasible regions considering CNCC (3) evaluate the efficiency of candidate locations by employing data envelopment analysis (DEA) with respect to UCC, and (4) display suitability map according to CCC. The high degree of fluctuation in weather conditions is an important factor that significantly influences the DEA results. Therefore, Z-number is adopted to represent the uncertainty of input data in DEA modeling. To validate the proposed approach, the obtained efficiencies from Z-number, fuzzy number, and crisp DEA models are compared to those obtained from the actual data of a five years period. The results indicate that about 160,000 km² land area is suitable to cultivate Paulownia in Iran that shows great potential for renewable energy production in this country. On the other hand, the computational results demonstrate that applying Z-number decreases the error of the weather forecast by 8% so that land suitability assessment can be carried out more accurately. As a result, this reliable method can prevent agricultural productivity loss due to selecting inappropriate locations.

Introduction

In the last few years, there has been a tremendous increase of interest in the production of biomass feedstocks for renewable energy purposes due to the obligation that intended to cut down the role of fossil fuels and the associated environmental impacts (Welfle et al., 2020). Iran, along with other developing countries, is seeking to reduce fossil fuel consumption in order to reduce environmental pollution.

Recently, a number of researchers have dealt with a tree named “Paulownia” that has attracted many interests owing to the various benefits including (1) high resistance against poor growth conditions (Chinese Academy of Forestry, 1986); (2) fast-growing as well as high yield in short-rotation (Ayrilmis and Kaymakci, 2013); (3) a substantial fiber content of Paulownia to produce paper (Ashori and Nourbakhsh, 2009); (4) beneficial for medicinal purposes such as curing bronchitis, decreasing coughs and control high blood pressure (Chinese Academy of Forestry, 1986); (5) high carbon sequestration (Basu et al., 2016), and most importantly, (6) great potential of energy. Therefore, Paulownia would be recognized as a promising renewable feedstock for biofuel production (Zachar et al., 2018). In this regard, a critical question arises about how to find the appropriate locations for Paulownia cultivation, and this question is the key issue of this research. It is worth noting that, selecting biomass supply locations as a strategic level decision is an important topic in the network design problem (Ahmed and Sarkar, 2018). Moreover, the impact of prescribing suitable biomass cultivation sites on transportation costs is inevitable.

Extensive solutions have been developed to evaluate the appropriateness of different locations for facility location, cultivation planning, and particularly biomass cultivation over the past years. A fraction of these studies has undertaken mathematical programming models to cope with the location problem. For this purpose, Sarker et al. (2009) presented a multi-objective optimization model to address the crop-planning problem with the objectives of maximizing the total gross margin and minimizing total working capital objective functions. They investigated several quantifiable factors, which affect the crop-planning such as land cover, yield rate, demand, resource availability, capital availability, and operational cost. Although the non-quantifiable criteria (e.g., rainfall, weather condition, flood, cyclone, and other natural calamities) have a vital role in the crop-planning, these factors have been neglected due to the difficulty of quantifying. Another multi-objective optimization formulation for the crop-planning problem was further proposed by Adeyemo and Otieno (2010), which aims to minimize the total irrigation water (m²), maximize the total net income from farming, and maximize the amount of total agricultural outputs in tons. To solve the proposed multi-objective model, they developed an evolutionary algorithm, namely multi-objective differential evolution algorithm (MEDA) that can be adopted for crop planning problem, especially in water-deficient areas. In order to find suitable locations for cultivating Jatropha curcas L. (JCL) as a promising bioenergy crop, Babazadeh et al. (2015) implemented a non-radial DEA model considering critical (i.e., non-compensatory), as well as non-critical climatic, economic, and social criteria. Waterlogging conditions, low soil pH, and long frost period were defined as critical criteria to filter primary candidate locations where human development index (HDI) and population, cultivation cost per hectare, and annual rainfall were among social, economic, and ecological criteria, respectively. In a recent research study on plant location problem, Bojic et al. (2018) presented a linear programming (LP) model to address the problem of selecting optimal locations for lignocellulosic bioethanol plants with the objective of minimizing transportation costs. In addition to the economic aspect, environmental and social criteria have also been considered in this study to select bioethanol plants in Serbia. They considered two different modes of transportation (i.e., road and inland waterway) to analyze the direct costs, as well as indirect costs (i.e., pollution, noise, and accidents) of transport service to compare the considered transportation modes according to their economic impact on the society.

On the other hand, insufficient knowledge about input parameters of optimization models aims to determine optimal cultivation locations impose restrictions on the solution approach for decision making. Consequently, in order to achieve more realistic and reliable results, researchers must consider the uncertainty of input data in location optimization problems (Franco et al., 2015, Mohammadi et al., 2019). Stochastic programming and fuzzy approaches are among common analytical methods which have been adopted by many researchers to deal with the uncertain condition in these problems (Mousakhani et al., 2017). For instance, in order to optimize crop planning decisions, including the selection of crops to be grown and the annual devoted acreages to cultivate, Galán-Martín et al. (2015) developed a decision-making tool based on multi-stage stochastic LP model aims at maximizing the net present value of the farmers’ net return. According to the mandatory requirements introduced by the new European Union, namely greening rules, three scenarios are defined for farmers’ profit regarding their decisions at the beginning of the planning period. In another research, to cope with multiple uncertain paramters in the crop area planning model, including water withdrawal, crop efficiency, price, and future irrigation volume, Zeng et al. (2010) proposed a fuzzy optimization approach. They formulated a fuzzy multi-objective linear programming (FMOLP) model to maximize the farmer’s economic net return, minimize the evapotranspiration, and minimize the deviation of the grain yield from the specified target in a fuzzy environment.

To examine the feasibility of plant cultivation in candidate locations, considering geographical factors is crucial (Voets et al., 2013). Additionally, mathematical modeling may be inapplicable or inaccurate to optimize the cultivation locations due to the disability to incorporate geographic criteria precisely into the solution approach (Sharma et al., 2018). To be more specific, a comprehensive investigation of the studied area requires analyising a large set of candidate locations, which increases the computational burden. On the other hand, few candidate locations without considering the geographical characteristics of the candidate locations cause inefficiency in strategic planning. To deal with this problem, Geographic Information System (GIS) has received great attention as a powerful tool to integrate the data of different indicators, manage, and process spatial information and geographic layers of the concerned area (Selim et al., 2018). This beneficial tool can be applied in agricultural planning, including (1) assessment of plants’ adaptability to areas with different conditions, (2) application of seeds, fertilizer, and pesticide, which are related to site-specific or precison farming in an environmentally friendly manner, and (3) water management, through spatial decision support systems. As a result, the application of GIS can improve agricultural production in terms of not only economic objectives but also environmental issues (Akinci et al., 2013). Moreover, GIS helps the researchers and practitioners that focus on location optimization problem of biomass plants and infrastructures in different areas. Applying GIS to estimate the feasibility of biomass power plants establishment, Shi et al. (2008) presented a case study to evaluate and select the optimal locations. GIS was also used in this research to specify the supply area of candidate sites according to the road transportation distance. They considered transportation cost, as well as geographic and economic criteria (e.g., land use and accessibility to the roads and transmission line) to determine final sites. Among the criteria, land use as a geographic criterion has restricted the planners from using the decision-making tools solely; therefore GIS would be a helpful alternative solution for removing the mentioned barrier. On the other hand, although sustainability assessment has become an active area in the biofuel industry in recent years, influential challenges associated with sustainability assessment across this industry have been disregarded in some cases. In that respect, Perpiña et al. (2013) identified, evaluated and weighted environmental (e.g., lithology and vegetation cover), economic (e.g., biomass quantity and transport costs), and social metrics (e.g., affected population and visual impact) to apply in a combined GIS and multi-criteria assessment technique for selecting suitable areas for locating biomass plants. In a similar study, in order to determine the importance of each factor, Bagdanavičiūtė et al. (2018) integrated analytic hierarchy process (AHP) and GIS techniques for initially weighting the identified criteria and then prioritize the locations to select appropriate sites for the cultivation of zebra mussel. They analyzed different environmental and socioeconomic criteria for the cultivation of zebra mussel such as salinity, land use, land cover, accessibility to support services and waterways, etc. Similar work was also performed by Gigović et al. (2017), which determined the best locations for wind farms by adopting DEMATEL-ANP and GIS techniques. They identified eleven economic, environmental, and social criteria such as distance from power lines and slope, wind speed average and land cover/use, and distance from telecommunications and population density, respectively. To provide a comprehensive insight on the relevant literature, Table 1 presents a systematic comparison between the previous studies, which account for location optimization problem of the biomass cultivation and supply sites, and the current research.

Given the aim, which is to assess the suitability of the land, it is required to evaluate and prioritize the study area in terms of criteria for Paulownia cultivation. However, mathematical programming (linear programming) will not be able to solve this problem for two main reasons: (1) conflicts among these criteria, and (2) their different measurement units (Özcan and Toklu, 2009). To this end, multi-criteria approaches can be a powerful solution to precise planning. Knowingly, most of multi-criteria decision-making (MCDM) approaches require expert weighting of criteria and thus this may increase the bias of the solution. To avoid unrealistic weighting, the DEA method can be used. This method is able to assign appropriate weights to corresponding criteria proactively. It also has advantages such as ease of interpretation of the results (efficiency scores), ability to evaluate criteria with different measurement units of (significant flexibility in selecting criteria), etc. (Sigala, 2003).

As previously pointed out, geographical criteria have a vital role in the cultivation planning of biomass sources. However, mathematical modeling is inefficient to be implemented solely to identify the optimum locations for cultivation since this approach is incapable of considering geographical criteria appropriately. In this regard, GIS can be applied along with mathematical modeling to analyze the geographical data in location optimization problems effectively and accurately. On the other hand, the issue of uncertainty and ambiguity of input data for weather criteria should be taken into account, because high fluctuation in criteria values may cause an irrecoverable deviation from the goals. To explain more, the meteorological content of the forecasts have a high impact on agriculture production. However, in spite of notable progress in the prediction systems, planning in real-life agriculture suffers from the vagueness of future weather conditions (Calanca, 2014). To avoid possible errors caused by the fluctuation in the meteorological values in the future, the methodology of previous studies has been limited to fuzzy number type 1. To do so, experts have been providing three numbers (minimum, average, and maximum) to forecast the weather (i.e., expected temperature, expected rainfall, etc.) for each region in the future. Based on the results, this has led to a significant improvement to deal with the uncertainty in weather parameters. But it should not be overlooked that experts are also capable of presenting reliability percentages for these values. Although its use could improve the accuracy of weather parameters estimation, previous studies have been lacking this valuable information in handling the uncertainty in this area. For this purpose, Z-number (Zadeh, 2011) is adopted in this paper to represent weather forecasting. Z-number is an ordered pair of fuzzy numbers (A, B) associated with an uncertain variable x, which has the capability to describe the reliability of uncertain information. The first member of Z-number, A, plays the role of a fuzzy restriction, where, the second member, B, is the reliability of “A”.

To the best of our knowledge, a practical combination of GIS and mathematical modeling for selecting optimal locations for biomass plant cultivation under uncertainty has not been addressed in the literature (see Table 1). This paper presents a hybrid approach by applying GIS and DEA to evaluate and determine the capability and propriety of candidate regions for paulownia cultivation regarding weather uncertainty. The novelty of this paper lies in the development of a methodology to apply Z-number as a type of fuzzy number to increase the accuracy of weather forecasting and thus to control the uncertainty in location optimization problems. On the other hand, all relevant criteria for Paulownia cultivation, including geology, geographic (i.e., pedology and weather), economic and social aspects are identified and taken into consideration.

The rest of this paper is organized as follows: In Section 2, the studied problem is explained, and the proposed approach is presented. Section 3 discusses the obtained results and displays the feasible and suitable cultivation regions. Finally, the overview of conclusion and managerial implications has been surveyed in Section 4.