Modelling Best Oil Palm Site Planting in Njimom, West-Cameroon: A GIS-Analysis Combining Weighted Linear Combination, Fuzzy Analytical Hierarchy Process and Utility Function

Oil Palm (Elaeis guineensis Jacq.) has recorded a boom production the last decades and its main productive zone is inside the tropics that meet the best biophysical conditions. Investors as well as geospatial practitioners are increasingly interested on the best growing and harvesting conditions. So said, the aim of this paper is to select the best oil palm planting site through the best methods combination. The study area is the district of Njimom located in the west-Cameroon, transitional between the equatorial and the climatic zones. In the same GIS environment, the Weighted Linear Combination (WLC) and Fuzzy Analytic Hierarchy Process (FAHP) respectively highlight the subtle differences between capability and suitability, while the Utility Function (UF) helps to assess the consideration of sustainability aspects. The first results consist in eight layers representing natural conditions, that is rainfall, temperatures, sunshine, slope, elevation, soil richness, soil moisture and forest cover, recoded in six classes ranked from 5 to 0 according to the FAO standardised scale. They are crossed using the straightforward method of WLC to give the “Capability layer”. The second results consist in three layers related to the social-economical constraints for production, as built-up How to cite this paper: Mfondoum, A.H.N., Mfondoum, R.B.N., Wokwenmendam, P.N., Gbetkom, P.G. and Ntengo, M. (2019) Modelling Best Oil Palm Site Planting in Njimom, West-Cameroon: A GIS-Analysis Combining Weighted Linear Combination, Fuzzy Analytical Hierarchy Process and Utility Function. Journal of Geographic Information System, 11, 138-165. https://doi.org/10.4236/jgis.2019.112011 Received: March 14, 2019 Accepted: April 13, 2019 Published: April 16, 2019 Copyright © 2019 by author(s) and Scientific Research Publishing Inc. This work is licensed under the Creative Commons Attribution International License (CC BY 4.0). http://creativecommons.org/licenses/by/4.0/ Open Access A. H. N. Mfondoum et al. DOI: 10.4236/jgis.2019.112011 139 Journal of Geographic Information System area, distance to road and distance to rivers. These layers are recoded in binary with 1 and 0, they are weighted by the FAHP priority vector and membership approach to give the “suitability layer”. Then, the number of occurrences of each aspect of the sustainability is counted in each of the two preceding processes to perform the UF. The resulting value, that is 0.542 for the capability process and 0.315 for the suitability process, serves to weight their respective layers, and their sum gives the final map with the best oil palm site planting in the northern part of the study area, on about 34,950 ha, representing 44.8% of Njimom district.


Introduction
Oil Palm (Elaeis guineensis Jacq.) is one amongst others world's most important tropical tree crops. It is mainly grown for the industrial production of vegetable oil [1]. The different uses include the manufacture of soaps, detergents and other toiletries in oleo chemical industries for the Kernel Oil Palm [2] [3], as well as ingredient in many foods, in leather, meal and chemical industries, for the Crude Oil Palm [4]. According to [5], an estimated 50 percent of packaged products sold in supermarkets contains some of the ubiquitous oil. More recently, it has been involved in the biodiesel production. Further, [6] [7] [8] [9] estimate that future increases in demand for vegetable oils will be primarily satisfied by palm oil rather than by other vegetable oils. This explains the boom of the production in the last decades, representing more than 30% of the global vegetable oil production [10], and making it the world's most widely consumed edible oil [4] [11]. Demand for palm oils is expected to double by 2020 [12].
The main productive zone is located inside the tropics between 10˚N and 10˚S of the Equator, where biophysical and environmental conditions are combined for its growth [1] [11]. For these reasons, the 43 first productive nations are tropical countries located in Southeast Asia, Africa, Central and South America [3] [13]. In most of these countries, governments increasingly promote its cultivation as a major contributor to poverty alleviation, as well as food and energy independence [14]. Then, palm trees are grown in two ways. They are observed in village gardens where they provide oil for local consumption with relatively lower yield and oil quantity. However, for industrial purposes, palm trees are mainly grown and harvested over large uniform areas of 3000 to 5000 ha [1] [11]. The grower or investor must then use modern methods for best harvesting and economical benefits [11]. He should integrate at the same time the two others dimensions of the sustainability, which are the social development and the environment protection or restoration. Geospatial techniques show ability to face this challenge and the processing's results are widely requested by enterprises, NGO's or others individuals who want to invest in any agricultural activity.

General Framework
According to [1] [11] [43] [44], oil palm plantation needs specific natural and economical conditions. For the natural aspects, climate, soil, sun exposition and physical surroundings are mainly concerned. The oil palm grows well where it is hot all the year round, with temperatures between 25 -28 degrees Celsius, and never under 15 degrees Celsius. The palm trees should be permanently exposed to a lot of sunshine for a strong photosynthesis, so the leaves will grow larges, the fruits will ripen well, and there will be more oil in the fruits. Optimal rainfall conditions are 2000 -2500 mm per year with a minimum of 100 mm per month, with 1000 mm and five months dry period per year as the worst moisture liveable.The land should be flat because oil palm thrives best in lowlands below 300 -400 m of altitude to avoid or minimize risk erosion. Soils should be deep and permeable to enable the good development of roots, and rich enough in mineral salts in order to produce many large clusters of fruit. The moisture supply is important to regulate the temperature usually through the presence of a surrounding dense forest.
Moreover, the economics aspects depend on the type of plantation to be created. [11] distinguishes three types of oil palm plantations. A small-scale palm oil farm may cover 7.5 hectares, for usually local farmers. The medium-scale farm ranges from 10 to 500 hectares. This type of farm normally uses modern agronomic practices such as plant spacing, cover cropping, fertilization, ring weeding, pruning. Large-scale farms cover an area exceeding 500 hectares. These are state owned enterprises, established to meet the internal consumption needs of the country and provide a surplus for export. For the medium and the large sizes, factories and mills should be close of the exploitation, as well as roads infrastructures for easy transport. So said, [45] and [46] have proposed some basics frameworks of growth conditions suitability classification ( Table 1). The ranking is made on a proposed scale of 5 to 0 from the optimal to the worst conditions, with mid-scales fuzzy linguistic appreciation (Table 2).

Casting of Methods and Background Related
The oil palm growing different conditions are enough expressive of a multi-criteria decision-making (MCDM) problem. It is mainly involved in combining information from several criteria to form a single index of evaluation [47].
Depending on the GIS analysis context, it can be qualified of resource allocation analysis or suitability/capability analysis, as in this paper [39] [40]. Then, the choice of fundamentals and supporting methods leading the analysis is based on the three aspects brought out by the planting conditions, that is, the natural ca- The WLC or simple additive weighting is one of the widely used multi-criteria evaluation (MCE) methods as a GIS-based decision rules for land suitability analysis [48] [49]. Practically, the suitability components and conditions are transposed onto maps and standardized according to the analyst' scale choice. Weighs of relative importance are assigned to each of them. Then, both weighs and standardized maps are combined using the following Equation (1): The AHP was developed in the 70's by [50] as a helping tool for decision makers to find out the ''best'' alternative out of several ones by considering a number of conflicting criteria [49] [50]. The process is based on the three principles, that is, decomposition, comparative judgment and the synthesis of data [50] [51] [52] [53]. Depending on the study, the fuzzy set could be incorporated with the AHP [54]. The idea of the fuzzy set theory is to deal with vagueness and no specificity inherent in human formulation of preferences, constraints, and goals [55]. The information is expressed by linguistic or fuzzy values rather than by numbers and this approach handles multiple criteria and combinations of qualitative and quantitative data [56]. The most popular fuzzy number is the Triangular Fuzzy Number (TFN), which is represented by three parameters as l, m, u, corresponding to the smallest possible value, the most promising value, and the largest possible value. For a fuzzy number denoted A  = (l, m, u), the membership function is defined as follow (Equation (2), Figure 1): Furthermore, linguistic variables that allow the translation of natural language into logical or numerical statements are hierarchically organized as a Linguistic Fuzzy Decision Network, based on the expert's preferences or the causal relationship between the values [57] (Table 3, Figure 2, Table 4, Figure 3).

• The Utility Function (UF)
How useful is each objective to the sustainability of the oil palm cultivation?
The answer to this question explains the integration of the utility function. UF is a logical attribution of values to the different baskets of consumption, such a way that the most desirable baskets receive higher values than those who are less [58]. The principle is a ranking of the consumer preferences. It is stated that a basket ( ) , y y , only if utility level U of ( ) is higher than the one of ( ) The UF used in this study is written as follow (Equation (4)): where c and d are positive numbers with the relationship:     (5,7,9) (1/9, 1/7, 1/5) 8  Very strongly extreme importance (6,8,9) (1/9, 1/8, 1/6) 9  Extreme importance (EXI) (7,9,9) (1/9, 1/9, 1/7)  [39] and [40], while integrating the sustainability to bring out another approach than the ones of [59] [60]. The second is most technical, by targeting the best methods combination to embody the said three theoretical considerations in the same GIS analysis, as tried by [59] and [60].

Study Area Preceding Methods Implementation
The study area is the district of Njimom, located between latitudes North

Conditions-Layers Generation, Ranking and Grouping
The layers were produced according to the standards scales (see Table 2   As all product of level 1 that is raw, the image need some preprocessing. Bands blue (B2) to short wavelength Infrared (B7) used in this study were stacked. Then, radiometric calibration from 16 bit to 8 bit and atmospheric corrections were performed. After, to enhance the spatial resolution, the panchromatic band (B8) was preprocessed and merged to the multispectral stacked image.
In the raster calculator, the sum of AVI and SI was computed to enhance the sensing of forest mark. Then, the BI was subtracted from that new layer to minimize the soil effects.
o Soils quality -Two neo-bands were produced.
Then, the stream network was digitized. A focal statistics analysis following a radius neighborhood of 20 meters was performed. To finish with this step, the drainage density of the study area was calculated according to Equation (12). Considering that soils should be well-drained to enable well-fixed roots, but not permanently drained to jeopardize growing and harvesting [11], the LWSI and DD layers were integrated in a raster calculator and respectively weighted at 2/3 (67%) and 1/3 (33%), to obtain the soil moisture.
o Built-up -Some studies have shown that built-up indices give more accurate information than the supervised or unsupervised classification [64]. For instance, the Dry Built-up Index (DBI) recently developed was important in highlighting cities features in dry climates [65]. Its formula is given by Equation (13).
( ) With NDVI meaning the Normalized Difference Vegetation Index, and expressed following the Equation (14).
( ) However, the study area is not exactly a dry one with sands and dunes soils and a sparse steppe vegetation, but natural features as shrubs and grasslands, o Roads network -Assuming that the design of a modern oil palm plantation request the proximity of a factory for beneficial transformation [11], the easy transport of the products need a furnished road network. The study area is opened up thanks to rural tracks and the national road 6 (see Figure 1) and it was stated that the farm factory should be close to the road by 1 km. The road network was digitized and rasterized using the Euclidian distance function on 4000 m, to cover the whole study area. A threshold of 1000 m was then applied and recoded in two classes of suitability, 1 inside this distance, and unsuitability, 0 that is over this distance. Technically, the layers were assembled in natural capability and social-economical suitability. On one hand, the group of layers belonging to the natural capability were reclassified, and recoded on an appreciation scale of 0 to 5 denoting the lowest to the highest capability; and because they are fixed conditions or less flexible for the productivity, the layers were crossed following the straightforward method of WLC sum method, to obtain the "Capability layer".
On the other hand, the social-economicalconditions-layers were aggregated after performing the AHP process. The pairwise comparison matrix has been built stating that local mills around the river network are 3 times less important than modern factory, but 5 times more than the built-up; while nearness between network road and modern factory importance is the highest, that is respectively 3 times and 7 times more than local mills and built-up (Equation (16)).

River network 1 1 3 5 Road network
These relative weighs were normalized to obtain the priority vector. The consistency of judgments were assessed through ratio of the consistency index (CI) to the random index (RI) in Equation (17) ( ) 0.10 The "Suitability layer" had resulted from sum of these three components weighted by their consistent priority vector in a raster calculator. The visual patterns and numerical results were confirmed by submitting the output image to the linear fuzzy membership operation.
Once the "Capability layer" and the "Suitability layer" were produced, it was important to assess the level of considering sustainability. The first step has consisted in counting the number of occurrences of the social, economic or environmental dimension in each layer processing (Table 6). Then after, the utility function was applied to come up with their individual sustainability values.
In both cases, x 1 , x 2 and x 3 refer respectively to social, economic and environmental aspects of the sustainability. Inside each group of layers, the ratio has been made between each aspect number of occurrences and the total. Thus, the cube root of all results was computed, and the new values were multiplied to find the sustainability utility coefficient (SUC) for each group. The non-occurring or null values were computed as 1.
As last step, in a raster calculator, each layer was multiplied by its so-called sustainability utility coefficient (SUC), before performing their weighted sum.
The diagram flowchart presented in Figure 6 synthesizes the main steps of the methodology.

Conditions-Layers Raw Values and Classes
An overall of eleven individual conditions-layers maps result from the processing above. Eight of them belong to the natural capability and three to the social-economical suitability for the oil palm planting. Inside the two groups, there are non-composed and composed or neo-bands layers.  Concerning the social-economical suitability, the non-composed layers are the distance to the road network (0 -3837 m) and to the rivers (0 -1500) (Figure 8).
They express the investors' needs and the population perception. The only composed layer is the built-up map (−1.37 -1.477), whose values approximate the area to be avoid by the plantation (Figure 8).
When compared to the standards of oil palm farming [11], these conditions are generally averagely meet (see Table 2) for oil palm plantation in the area.
Only the elevation (696 -1533 m), that should better be between 400 and 500 m, as well as the soils that the best mineral content is clay, show some first view restrictions (see Table 2). Then, as stated above, the best way to assess that was the recoding of the layers according to the standardized scale.
The natural capability conditions-layers are reclassified into six new classes.
The fuzzy linguistic scale is decreasing, with "Highest", "High", "Moderate", "Marginal", "Low", and "Lowest" capability. On the base of their original processed data, all the conditions-layers highest to lowest values were completely integrate the model, except the elevation whom data ranking were inverted, giving highest scores and qualifiers to the lowest altitudes. For the layers that extremes data were inexistent, the number of classes was adapted to the score values, for better results (Table 7, Figure 9).   The social-economical suitability conditions-layers have two classes, "Suitable" and "Unsuitable". The highest values of the built-up are unsuitable for the oil plantation. The distance of 4000 m around the road is the suitable for the factory for the easy access. The distance over 100 m of the river is suitable for the farming and the factory to preserve local oil palm production activity and then avoid conflicts with local population (Table 8, Figure 10).

Capability and Social-Economical Combined Layers
The capability layer is resulting from the WLC shows trends with values going from 23 to 45. The highest capability is recorded in the northern part of the study area. Its scores correspond to highest soil richness, soil moisture and forest cover, average sunshine, rainfall and temperatures, lowest altitude and slope ( Figure 11).
The suitability layer is the result of the AHP performing (Table 9). Then, the three recoded layers multiplied by their priority vectors are respectively, the river network layer by 0.2828, the road network layer by 0.6434 and the built-up layer by 0.0738. Their linear fuzzy membership sum is in the interval between 0 and 1, with a visible domination of the road network layer ( Figure  11).

Sustainability Utility Coefficients (SUC) Weighing and Final Best Site Maps
Following the UF computing, each layer obtains a SUC rated on a fuzzy scale of 0 to 1. None of both groups fulfils the sustainability dimensions at 100%. Natural capability processing express it at 54.2%, that is normalised value of 0.542, while social-economical processing express it at 31.5%, that is a normalised value of 0.315 (Table 10). The two layers are recoded into six classes for the final map scale of appreciation. Then, using the SUC normalized values, the capability layer is multiply by 0.542 and the suitability layer by 0.315. The sum of two layers in the raster calculator had given the last map.    Distance to road (m) ≤4000 >4000 Distance rivers (m) >100 ≤100 Table 9. AHP matrix priority vectors.  The final map is laid out in four different designs ( Figure 12). The first one shows the main trends from the best to the least or worst conditions. The values are comprised between 5.142 in the best case and 0.857 in the worst. The main area for the high values is in the north of Njimom district. This map is recoded into six classes to discriminate and prioritize the sites in the second map. The ranking is from the first to the sixth priority for the investor, with areas comprised between 35,700 ha and 153 ha ( Figure 13). When applying a binary recoding following the "Yes"/"No" logic to this map, considering the highest class as valid and others as invalid, its enables to highlighting the best site of the study area in the third map. The first priority area is the best site, on an area of 35,700 ha. Ten plantations over eleven related to the local mills used for the analysis are within the limits of the site. The sum of their estimated minimum areas, when taking seize as 7.5 ha according to [11] for small farmers (7.5 × 10 = 750) (see Table 2), is subtracted from this 35,700 ha, and the approximate overall exploitable area is about 34,950 ha; that is 44.8% of the 78,000 ha (780 km 2 ) district total area ( Figure 14). The selected site is then utilisable for the large size modern oil palm plantation (See Table 2). At last, the limit of the best site is drawn,

Best site
Other sites

Discussions and Outlook
According to the main purpose of selecting the best oil palm planting site, this study has been done following as possible the international ranking standards.
The main outcome of the processing locates the potentially most indicated site on the northern lowlands of Njimom district as shown in Figure 12. Its visual comparison with the inputs layers shows a large influence of the "Capability layer" trends (see Figure 11). One-step backward and still based on the visual comparison, seven on eight conditions-layers are highly highlighting the northern area as most favourable, except the forest cover layer that records the medium and lowest values there (see Figure 7 & Figure 9); while layers as slope and elevation most clearly, but also rainfall and temperatures, deeply influence the shape of the chosen site.
Further, the main explanation of the "Capability layer" trends domination over the "Suitability layer" ones on the best site selection is the weighing process using the sustainability utility coefficients (SUC). Indeed, with respectively more than 50% (0.542) and just 30% (0.315), the outcome is more representative of the first layer than the second one. Consequently, when estimating the areas of the highest capability value ( Figure 11) and the best site area, they are respectively 23,349 ha and 34,950 ha. The difference of 11,601 ha represents the contribution of the "Suitability layer" to the final selection, which is 33.2% or 1/3 of the total area, versus 66.8% or 2/3 represented by the above "Capability layer" value.
On another point, although the site is the most adequate according to the local available conditions, it does not meet 100%, because the layers of slope and elevation do not record the highest ranking of 5. Then the highest score ranking of the final map is at the most 4 according to the international standards.
The same way, concerning the purpose of erosion, the said standards and all the studies read on the purpose of oil palm land suitability most focus on avoiding any risk than integrating it to the modelling process. To pursue with this study, the main outlook is on one hand to build a complete model of the Revised Universal Soil Loss Equation (RUSLE) and integrate it to the model. On the other hand, integrating the status of the land degradation in the process, following the model proposed by [67] can bring more information to the site selection.
These two components might reinforce or mitigate the outcome.
The analysis has also opened the way to apply the DBI to extract the built-up, and finally to test the inverse ratio of this formula for better results (see Equations (13) and (15)). This is a point the authors are already working on, to deepen the understanding of raw materials built-up reflectance parameters, related to thermal infrared and blue bands of the electromagnetic spectrum, in the remote sensing of diverse dry climates cities.

Conclusion
This paper has experimented to combine the WLC, FAHP and UF methods in the same GIS environment. The goal was to choose the best oil palm planting site. The three directions were capability and suitability, headed by the idea of sustainability. Then it appears that in the district of Njimom, with transitional biophysical features between the equatorial and the tropical climate, some requirements are fulfilled for oil palm planting in its northern side. Therefore, using the straightforward approach of WLC helps to follow the standardized scale of adequate conditions as rainfall, temperatures, sunshine, slope, elevation, soil richness and moisture, and forest cover, which should be respected for better growing and harvesting. The same way, using a flexible analysis as FAHP helps to integrate the human thoughts and choices in the analysis as built-up, factory location or local populations' mills. Another enhancement is the level of considering sustainability, 0.857, that is further used as an evaluator of the overall social, economic and environmental dimensions of a plantation project. At last, combining the three methods has brought up the best site for oil palm plantation on an area of 34,950 ha, predisposing the district of Njimom to host modern and sustainable agriculture for local development.