Comprehensive solid waste dumpsite selection in arid cities of Northeastern Ethiopia: A spatial-MCDA approach

ABSTRACT

In the coming three decades, world waste will be expected to increase by 70%. Coupled with the absence of waste management technologies, unregulated dumping remained a threat to sustainable development in developing countries. To this end, this study emphasized to trace socially acceptable and environmentally friendly solid waste dumpsite for the northeastern arid cities of Ethiopia. The tuneful integration of GIS and MCDA was used for the end-to-end processing of 17 variables. At the earlier, to increase alternative dumpsites and seeking for safe zone outside the cities, 20 km buffer at Semera-logia city was used as a spatial scope of the study. Fifteen key informants (KIs) from related departments of the city and Samara University were selected to validate the model result. In the end, Ordinal Regression (OR) analysis was used to test the significance of the criteria’s on suitable dumpsite selection. The pair-wise comparison matrix shows that variables such as distance from surface water (11.32%), distance from the settlement (9.88%), distance from boreholes (8.87%), and aquifer productivity (8.51%) were the top weighed criteria for the study area. The weighted overlay result shows that 10.75% and 38.64% of the study area are found to be convenient and prohibited for dumping, respectively. Indeed, the KIs were agreed on 77% of the validation points and the agreement decreases from points of not-suitable to highly suitable. Except for aquifer productivity, all the 16 variables are significantly and negatively affect dumpsites suitability. This result can be further used as a preliminary database for experts to design landfill structures and also management operations for decision-makers.

Implications: The identification of potential dumpsite is an important step towards maximizing efficiency and building sustainability in the arid areas. The decision-making process shall be based on the empirical investigation. The decision-making process on dumpsite selection were well thought out both the socio-economic and environmental factors. In owing process, the empirical investigation that was conducted focusing on socio-economic dimension jeopardized other components of the environments. The approach that considers both socio-economic and spatial factors was all-inclusive. Overall, decision-makers could have a good insight to work on prioritizing dumpsite alternatives and implement feasible waste management actions.

Introduction

Background

Under the existing waste management scenario of the world, the waste disposal is expected to increase by 70% from 2.14 billion tonnes (2016) to 3.17 billion tonnes in 2050 (World Bank 2019) and it is a threat to sustainable development (Getahun et al. 2011). This implies that the impact of unsustainable waste management is worse in developing countries (World Bank 2019). The Sub-Saharan Africa region generated 174 million tons of waste in 2016 and more than 40% was organic (Kaza et al. 2018). Waste management is an essential urban service (Kaza et al. 2018). In Africa, less than half of the generated wastes are collected from door-to-door (Kassahun and Birara 2018) or informal points (Kaza et al. 2018).

Proper waste management is essential for building sustainable and livable cities, but it is a challenge for many developing countries and cities (Kassahun and Birara 2018; Umar and Naibbi 2019; World Bank 2019). Globally 33% of waste is still openly dumped (Kaza et al. 2018). In low-income countries, over 90% of waste is often disposed of in unregulated dumps or openly burned (World Bank 2019). Despite its location, waste dumping is the final hierarchy in the waste management process (Ekmekçioĝlu, Kaya, and Kahraman 2010). Around the world, almost 40% of waste is dumped to landfills (Kaza et al. 2018). However, proper landfilling is often lacking in developing countries (Hoornweg and Bhada-Tata 2012; Umar and Naibbi 2019). In this region, open dumping is prevalent and 93% of waste is burned or dumped on roads, open land, or waterways (Kaza et al. 2018). In developing countries like Ethiopia, the waste disposal system is unplanned and worsened (Kassahun and Birara 2018; Umar and Naibbi 2019). In cities like Mekelle, Bahir-Dar, Addis Ababa and Jimma, waste disposal areas are low-lying open spaces, roadsides and uncontrolled sanitary landfills (Adane and Muleta 2011; Getahun et al. 2011; Kassahun and Birara 2018; Regassa, Sundaraa, and Seboka 2011; Tadesse, Ruijs, and Hagos 2008). Besides, 74% of existing disposal sites are reported as unsuitable in Nigeria (Umar and Naibbi 2019).

The poorly managed solid waste operation causes different problems in different cities of developing countries (Regassa, Sundaraa, and Seboka 2011; World Bank 2019) Improper waste disposal causes health hazards, environmental contamination, climate change and urban violence (Mohammedshum et al. 2014; Munyai and Njabulo 2020; Regassa, Sundaraa, and Seboka 2011; Salinah, Naing, and Sarimah 2020; World Bank 2019). Effective waste management is expensive, which takes 20–50% of municipal budgets (World Bank 2019). However, landfilling is becoming more prevalent in the sub-Saharan Africa region (Kaza et al. 2018). Construction and use of landfills are commonly the first step toward sustainable waste management (Kaza et al. 2018).

Problem definition

Due to the rise in environmental awareness and education, investigating areas suitable for waste disposal is vital to the sustainable development of a city. Governments are pursuing sustainable waste disposal methods and looking for ways to expand local capacity and continuing effort to improve solid waste disposal practices (Kaza et al. 2018). The solid waste collection point ought to be convenient for the community (Yalew 2012) and needs careful planning (Regassa, Sundaraa, and Seboka 2011).

Proper solid waste management (SWM) requires convenient place for appropriate disposal (Kassahun and Birara 2018). Municipal solid waste (MSW) disposal is a complex and multidisciplinary problem that environmental, social, and technical as well as economic aspects need to be considered (Ekmekçioĝlu, Kaya, and Kahraman 2010). The convectional multi-criteria decision analysis (MCDA) techniques have largely been non-spatial and it lacks well-developed, analytical capabilities to support decision-making processes (Sharifi and Retsios 2004). The applications of geographic information systems (GIS) and remote sensing techniques are efficient and low cost in selecting the appropriate site of dumping (Mohammedshum et al. 2014).

The decision made by municipality officers on dumpsite selection needs, integrating the socioeconomic and environmental information. Waste disposal requires integrated systems that are efficient, sustainable, and socially supported (Fatoyinbo et al. 2020; Fernando and Claudio 2020; World Bank 2019). Site selection is a good application area for integrating the MCDA and GIS. Spatial-MCDA in a user-friendly environment is vital and enables us to arrive at an effective, most acceptable solution, efficient and policy-relevant information (Danesh et al. 2019; Ekmekçioĝlu, Kaya, and Kahraman 2010; Jensen and Christensen 1986; Muttiah, Engel, and Jones 1996; Sener 2004; Sharifi and Retsios 2004).

Related studies in Ethiopian cities confirmed improper SWM where (low-lying) open fields, roadsides, and uncontrolled sanitary landfills are common waste disposals and in further posing diseases and disturbing ecological integrities (Ali 2016; Ambaneh 2016; Asefa and Mindahun 2019; Beshir 2014; Erasu et al. 2018; Hailu, Hanchiso, and Bereta 2019; Kassahun and Birara 2018; Lohri, Camenzind, and Zurbrügg 2014; Mekuria, Muralitharan, and Ali 2019; Mohammed and Elias 2017; Mohammedshum et al. 2014; Mussa and Suryabhagavan 2019; Solomon 2018; Taddese and Balcha 2019; Tassie, Endalew, and Mulugeta 2019; Wakuma 2018; Wondim and Dzwairo 2018).

Despite the failure of execution, the FDRE constitution, Article 44.1 verbalized an environmental right as “All persons have the right to a clean and healthy environment” (FDRE 1994) and again environmental pollution control proclamation no. 300/2002 article 5.1 stated that “All urban administrations shall ensure the collection, transportation, and, as appropriate, the recycling, treatment or safe disposal of municipal waste through the institution of an integrated municipal waste management system” (FDRE 2002). Following that, specifically solid waste management proclamation no. 513/2007 was issued with the aim “to enhance at all level capacities to prevent the possible adverse impacts while creating economically and socially beneficial assets out of solid waste” (FDRE 2007). In detail, the responsible government body, Ministry of Urban Development and Construction, developed and set out a solid waste management manual (MUDC 2012). However, with different factors, comprehensive solid waste management in most cities of the country remains risky.

In an arid environment, glass containers, tin cans, plastic bags, and food-related solid wastes are most dominant. For instance, Dire Dawa and JigJiga city administrations devised solid waste dumpsites which are located 7 km far-away from the center of the city (MUDC 2016a, 2016b). However, in Dire Dawa, lack of administering the dumpsite causes a public health problem (MUDC 2016a). And also in JigJiga city, the dumpsite covers 4 ha and served for the past 13 years without a fence. Still, waste materials are dumped into the nearby Toga river and late military camp. Increasing the extent of the dumpsite and illegal burning of the waste are reported as a challenge in Jigjiga city (MUDC 2016b).

Around the study area, Samara-Logia city administration, massive amounts of solid wastes are dumped into river channels, drainage lines, sacred sites, and generally, free space out of the city are preferred areas (Ali 2016; Mussa and Suryabhagavan 2019). Mussa and Suryabhagavan (2019) conveyed the dumpsite suitability map of Logia town using criteria such as road network, soil map, land use/land cover (LU/LC), slope, geology (permeability), river network, fault, built-ups, depth of wells. The study pointed highly suitable to suitable areas toward the northern part of Logia town that is near to Semera city. The major gaps of the study are (i) Semera and Logia are under one administration body which tend to grow each other and the middle space where the so-called highly suitable dumpsites would be part of the city, (ii) the decision criteria employed in the study are most common in humid areas. For instance, the wind direction is mostly blowing from north and northeast to Logia town (south) which carries waste materials back to the town. (iii) Urban expansion is a common event in an emerging area. This would have a significant impact on solid waste dumpsite suitability modeling. Hence, this indicates that dumpsite mapping must not be limited to the current boundary of a town rather considers future development options.

Therefore, important and contextually parameters suited to arid environments stretching from socioeconomic to ecological issues have to be experimented to locate sustainable solid waste dumpsites in the study area. The overall objective of this study is to locate urban solid waste dumpsite attaining ecological friendly, economically feasible, and socially acceptable via spatial-MCDA system contextualized for the arid city, Semera-logia.

Material and methods

Study area description

Semera-Logia town, the study area, is an administrative seat of Afar National Regional State of Ethiopia. Geographically, the study area is situated in 40.7°– 41.2°E longitude and 11.5°– 12.0° N latitude (Figure 1). The Awash river is the most exploited river which crosses Semera-Logia and Dubti towns then flows to lake Abe at the lower plain (surface cease) of the basin (Yoseph, Bekele, and Mariam 2018a). The mean annual temperature at Dubti, Semera, and Logia site shows 30°c, 31°c, and 28°c, respectively, of which the hottest month ranges from May to September/October. The available mean annual precipitation record reveals near to 16 mm at all sites (Climate Data 2019; Yoseph, Bekele, and Mariam 2018a). From flat slope up to a gradient of 55° and hot desert climates type characterizes the study area.

Figure 1. Location map of the study area

Out of the total population of Dubti district projected for 2017, 61.34% (61,521) of people are urban dwellers where Semera, Logia, and Dubti towns take a larger share (CSA 2013). Semera-Logia town is located on the grand import/export path of Ethiopia. The major economic sources of the town include trade, office work, and construction, and other small-to-large-business (Mussa and Suryabhagavan 2019). The major soil type includes Solonchaks, Leptosols, Regosols, and Fluvosols where the latter soil types are dominated by alluvial deposition and opportune for irrigated agriculture (FAO 2006). The study area accesses an annual river nearby and also pumps groundwater (GW) via boreholes and dug-wells (Yoseph, Bekele, and Mariam 2018a).

Data and data collection

Semera-Logia town is abandoned in terms of the massive solid waste disposals. Additionally, solid waste materials are dumped every day from individual homes, offices, shops, and other business centers (municipal waste). It is presumed that the future quantity of solid waste and disposal system is sought for systematic planning. Therefore, Semera-Logia town was purposively selected for comprehensive dumpsite selection. By assuming the future urban development and looking dumpsite out of the town, the current (2020) boundary of the towns were buffered with 20 km as a spatial scope.

All the required criteria considered for the acceptable solid waste dumpsite selection were collected from different sources, i.e., reports, books, articles, web pages, raw data providers, and field surveys (Table 1). According to current literature, in the first stage, 38 variables were identified and some variables were again merged and removed to suit the local context and finally 17 major socio-economic and environmental variables were selected for the study (Table 1). All data required for the study were collected and prepared in the late months of 2019. The spatial data pre-processing was undergone to harmonize the selected variables (data) in terms of projection (WGS 84/UTM zone 37 N) and resolution (30 m).

Table 1. MCDA parameters of solid waste disposal site selection

Criteria	Contextually stated suitability ranks						Literature
	Weight (%)	Highly Suitable	Suitable	Moderately Suitable	Less Suitable	Not Suitable
Socioeconomic criterions
Distance from settlement ^a	9.88	>3000	2000–3000	1000–2000	500–1000	<500	(Ali 2016; Ambaneh 2016; Hailu, Hanchiso, and Bereta 2019; Danesh et al. 2019; Pasalari et al. 2019; Josimović and Marić 2012; Sharifi and Retsios 2004)
Access to road (intermediate) ^a	6.75	1000–1500	1500–2000	2000–3000	>3000	0–1000	(Mohammedshum et al. 2014; Ambaneh 2016; Hailu, Hanchiso, and Bereta 2019; Asefa and Mindahun 2019; Danesh et al. 2019; Pasalari et al. 2019; Josimović and Marić 2012; Şener, Sener, and Karagüzel 2011)
Distance from protected areas ^a	6.75	>2500	2000–2500	1500–2000	1000–1500	<1000	(Ambaneh 2016; Danesh et al. 2019; Pasalari et al. 2019; Josimović and Marić 2012)
Distance from cultivated land ^b	5.97	>2500	2000–2500	1500–2000	1000–1500	<1000	(Josimović and Marić 2012)
Distance from electric towers ^a	4.94	>1000	800–1000	600–800	400–600	<400	(Josimović and Marić 2012)
LU/LC ^a	4.69	Bareland	Grassland	Shrubland	Woodland	CL, WL, WB & S	(Ambaneh 2016; Asefa and Mindahun 2019; Danesh et al. 2019; Pasalari et al. 2019; Josimović and Marić 2012; Şener, Sener, and Karagüzel 2011; Sharifi and Retsios 2004)
Distance from rural residence ^a	`	>1500	1000–1500	500–1000	250–500	<250	(Josimović and Marić 2012)
Environmental criterions
Distance from surface water (pond, river, reservoir) ^a	11.32	>1600	1200–1600	800–1200	200–800	<200	(Mohammedshum et al. 2014; Ambaneh 2016; Hailu, Hanchiso, and Bereta 2019; Asefa and Mindahun 2019; Danesh et al. 2019; Pasalari et al. 2019; Josimović and Marić 2012)
Distance from boreholes ^b	8.87	>1600	1200–1600	800–1200	200–800	<200	(Mohammedshum et al. 2014; MUDC 2012; Danesh et al. 2019; Pasalari et al. 2019; Josimović and Marić 2012)
Aquifer productivity (considering borehole sites) ^b	8.51	Aquitard – minor aquifer	Extensive & moderately fissured	Extensive & highly productive fissured	Extensive & highly productive porous	Extensive & moder. product. porous	(Hailu, Hanchiso, and Bereta 2019)
Soil type ^d	5.31	-	Solonchaks	Leptosols	Regosols	Fluvosols	(Ambaneh 2016; Asefa and Mindahun 2019)
Lithology ^b	5.30	Basaltic volcanic	Conglomerate	Rhyolite	Alluvial deposit	Swamp	(Mohammedshum et al. 2014; Ambaneh 2016; Mekuria, Muralitharan, and Ali 2019; Danesh et al. 2019; Josimović and Marić 2012; Şener, Sener, and Karagüzel 2011; Simsek, Kincal, and Gunduz 2006)
Slope aspect (facing to wind direction) ^c	4.49	SE, S, SW, W, NW	-	-	-	N, NE	(Danesh et al. 2019; Pasalari et al. 2019; Josimović and Marić 2012; Şener, Sener, and Karagüzel 2011)
Soil depth ^d	4.36	40–60 m	25–40	15–25	60–100 m	<15	(Danesh et al. 2019)
Elevation (m) ^c	3.52	331–408	408–479	479–572	572–676	>676	(Danesh et al. 2019)
Slope gradient (degree)^c	2.61	0–10	10–20	20–30	40–55	-	(Mohammedshum et al. 2014; Hailu, Hanchiso, and Bereta 2019; Danesh et al. 2019; Pasalari et al. 2019; Şener, Sener, and Karagüzel 2011; Simsek, Kincal, and Gunduz 2006; Sharifi and Retsios 2004)
Distance from faults ^a	2.37	0–500	501–1000	1001–1500	1501–2000	>2000	(Danesh et al. 2019; Pasalari et al. 2019)
	100

Notes: Superscript letters on the criteria variable indicates the sources of raw data. ^a Google earth & field visit (digitizing) – shape file; ^b Ethiopia Geological survey – scanned map at scale of 1:250,000 (Yoseph, Bekele, and Mariam 2018b); ^c USGS – 30 m resolution (earthexplorer.usgs.gov); ^d Shape file from Ethiopian Ministry of Water, Irrigation & Energy & (FAO 2006)

Data organization and analysis

Criteria preparation and suitability mapping

Following variable selection, in GIS environment, variables at which distance is measured were buffered with suitability class distance and again categorical variables, i.e., land use/land cover (LU/LC) and soil type were processed and then suitability class Id was assigned altogether to generate criteria suitability maps. Subsequently, the suitability class Id of the variables are used to convert into raster format to ease further analysis (Figures 2; Figures 3; Table 1).

Figure 2. Solid waste dumpsite selection modeling workflow

Figure 3. Variables suitability map

The landscape and land cover patterns of the study area are less heterogeneous. This helps the LU/LC mapping process in terms of techniques to be used. Therefore, the Google Earth satellite image of 2019 was imported to the QGIS interface and on-screen-digitizing technique was used to detect the LU/LC features of the study area. Finally, based on woody biomass inventory and strategic planning project definition (WBISPP 2004), eight LU/LC types such as bareland, grassland, shrubland, woodland, cultivated land (Cl), wetland (WL), waterbody (WB), and settlement (S) were identified.

The socioeconomic criteria include distance from the settlement, access to road (intermediate), distance from protected areas, distance from cultivated land, distance from electric towers, LU/LC, and distance from the rural residents. This criterion addresses community safety, health, and acceptance issues in the study area. Such as distance from residents (rural and urban) and protected areas (parks, religious places, airport, and institutions) protect stench; distance from road considers costs of transportation and safety of road adjacency; distance from electric tower protects power transmission towers from accidental damage incineration; LU/LC and distance from cultivated land safeguards the safety of productive lands (Table 1).

Environmental criteria are considered to protect potential contamination through waste leachate, pollution posing failure in ecological integrity, and ensure long-term stability of landfill structures. This criterion includes distance from surface water (pond, river, reservoir), distance from boreholes, aquifer productivity, soil type, lithology, slope aspect (facing to wind direction), soil depth, elevation (m), slope gradient (degree), distance from faults. The suitability ranking was stated to avoid potential costs of maintaining landfill structures and polluting soil and water resources (Table 1).

The Analytical Hierarchy Process (AHP) techniques of MCDA were used to evaluate the relative importance and weight of the variables. AHP is consists of three main operations, including hierarchy construction, priority analysis, and consistency verification (Saaty 2008).

In the first phase, the corresponding weight of all of the 17 variables was calculated using a pairwise comparison matrix. This is done using the relative importance of two variables considering one (equally important), two (moderately important), three (strongly important), four (very strongly important), and five (extremely important) (Table 1) (Saaty 2008).

In the second phase, the consistency ratio (CR) (eq.1)(1) $CI=\left({\lambda }_{max}-n\right)/\left(n-1\right)$(1) and the consistency index (CI) (eq.2)(2) $CR=CI/RI<0.1$(2) were calculated to verify pairwise assessment (Alonso and Lamata 2006; Saaty 2008). In calculating CI, the value of the random index (RI) value for 17 variables was obtained from Alonso and Lamata (2006) which is equal to 1.6086. Finally, the calculated CI and CR values are 0.15 and 0.09, respectively. Therefore, the CR confirmed reasonable consistency of pairwise assessment (Saaty 2008).

(1) $CI=\left({\lambda }_{max}-n\right)/\left(n-1\right)$(1)

(2) $CR=CI/RI<0.1$(2)

Where λ_max is maximum eigenvalues, n is number of parameters.

At the end of the workflow, the suitability values of each variable were prepared for overlay operation and multiplied by the corresponding weight of the variables to generate the final weighted overlay and suitability map (Figure 2).

Validation

Different validation procedures are adapted to verify the certainty of spatial-MCDA results. Among that web-based public participation GIS was employed to select areas suitable for restaurant establishment. The study compared a map provided by the citizen and expert which revealed low-level of consistency (Vanolya, Jelokhani-niaraki, and Toomanian 2019).

In this study, a 30 km long and six transect lines with 5 km spacing were established. Afterward, random points were distributed along the transect lines and 96 representative points were taken (Figure 1) and the suitability level was sampled from the final suitability map. Finally, the sample points labeled by the suitability level were printed and provided for key informants to receive their dis/agreement on the expert-generated dumpsite suitability map.

Here, the key informants were selected based on their knowledge to interpret a map and understand the effects of improper waste dumping in the environment and community. Therefore, a total of fifteen (15) key informants from Semera, logia, and Dubti towns who are working in related departments and experts from Samara University were selected. Finally, a frequency table was used to summarize the responses.

Ordinal regression (OR) analysis

Depending on the nature of the endogenous variable which is the Likert or ordered variable, the ordinal regression, multinomial logistic regression, and binary logit (if the response variable is recoded into different variables i.e. dummy response) models can be used. If the dependent variable has a category which is a natural ordering, the ordinal regression model fits the data (Gujarati 2004; Rao 2008). To choose a link function, it is helpful to examine the distribution of values for the outcome variable (Elamir and Sadeq 2010).

Hence, to run the OR model, intersecting polygons of the 17 independent variables and the dependent variable (dumpsite suitability) maps were used and the spatially indexed category values were imported to statistical package for social science (SPSS v. 20). To identify the link function, the histogram of the dumpsite suitability data was observed and the “not suitable” category shows extreme values. For this type of dependent variable, Cauchit link function is typical fit this data (Elamir and Sadeq 2010; Wainaina 2012).

In the end, the Chi-square (X²) test, model goodness of fit, and significance of the predictor variables were generated and interpreted.

Results and conclusion

Dumpsites suitability modeling

In developing countries like Ethiopia, waste generation and improper dumping are increasing with population growth and urbanization. The waste generated is also dumped along the river, roadsides, and open areas. GIS and remote sensing techniques are tools that are employed in this kind of preliminary studies due to its ability to manage a large volume of spatial data from a variety of sources. Identifying suitable landfills using GIS may not be accepted by the local communities and decision-makers. The issue of selecting an optimal site for dumpsite is depended on the socio-spatial data. Therefore, the comprehensive approach that incorporates expert judgments (AHP-MCDA) and GIS is substantial. The study demonstrated the effectiveness of MCDA (i.e. AHP techniques) and GIS in the site identification and selection process.

Among the environmental and socio-economic criteria used in this study (Figures 3; Table 1), expert’s preference shows that, distance from surface water (11.32%), distance from the settlement (9.88%), distance from boreholes (8.87%), aquifer productivity (8.51%), distance to the road (6.75%) and distance from protected areas (6.75%) were given more weights and recognized as the most important criteria for landfill site selection (Table 1).

Recent studies revealed that, in addition to residential areas, environmental variables are considered and weighted larger than socioeconomic parameters (Asefa and Mindahun 2019; Danesh et al. 2019; Pasalari et al. 2019; Şener, Sener, and Karagüzel 2011). It is known that the quality and quantity of available water are limited in arid areas. This study has given more attention to water-related variables to protect sensitive resources of the area from solid waste contamination. In addition to that, wind blowing is a day-to-day phenomenon in arid areas. In this study, the slope aspect gets intermediate weight (4.49%). However, it is less weighted in studies such as Pasalari et al. (2019).

Accordingly, the spatial-MCDA output map shows that 38.64% of the study area is found to not suitable for landfill or dumpsite. This contamination-free area is only for harmonious environmental integrity and citizens’ safety (Figure 4). About 10.75% and 13.39% of the areas were highly suitable and suitable for the landfill/dumpsite, respectively. Related studies in the field reported that majority of modeled areas and existing waste disposal sites are not convenient in-terms of social safety, environmental cleanness, and economic feasibility (Asefa and Mindahun 2019; Fatoyinbo et al. 2020; Khodaparast, Rajabi, and Edalat 2018; Şener, Sener, and Karagüzel 2011). On the contrary, Erasu et al. (2018) reported that 57.5% of solid waste is properly disposed of into landfill sites, whereas the remaining proportion of wastes are illegally dumped at roadsides and open spaces.

Figure 4. Solid waste dumpsite suitability map

In most cases, a higher number of parameters are considered as decision-making alternatives, the more reliable decisions could be made (Danesh et al. 2019; Pasalari et al. 2019). On the contrary, a fewer number of variables could reduce insight to indicate a precise location for dumping (Asefa and Mindahun 2019). Hence, this study considers multiple criteria and provides alternative sites for proper dumping of overwhelming solid wastes. Selection of specific location and design periods is left for experts working on the field and local decision-makers.

Validation

The response from the KIs revealed that 77% of the total points indicating the modeled dumpsite suitability level was acceptable (Table 2). In addition, the majority (51%) of the responses shows agreement on areas not suitable for dumping. Therefore, the consistencies of responses are higher on not suitable to less suitable areas for dumping.

Table 2. Summary of validation responses

Suitability Level	Sample points	Agree		Disagree		Total
		Count	% of total	Count	% of total
Highly Suitable	11	92	8	73	22	165
Suitable	11	145	13	20	6	165
Moderately Suitable	13	122	11	73	22	195
Less Suitable	19	179	16	106	32	285
Not Suitable	42	569	51	61	18	630
Total	96	1107	77	333	23	1440

Notes. Number of respondents: 15

According to the KIs response, model results for highly suitable and suitable areas were found extremely far from the center were subjected to disagreement. This shows that dumpsite should be placed neither near to the city, nor extremely distant areas.

Ordinal regression analysis

The significant X² statistic indicates that the model gives a significant improvement over the baseline intercept only model (Appendix B). This tells us that the model gives better predictions than if we just guessed based on the marginal probabilities for the outcome categories.

Based on the model fitting information (i.e. X² =150,797.4 and sig. = 0.000) can be concluded that there is an association between the dependent and independent variable(s) in cauchit link function.

The goodness of fit test table shows consistency between the observed data and the fitted model. Pearson’s X² statistic for the model is intended to show whether the observed data are consistent with the fitted model. Since the significance values are small (sig. =0.000) the data and the model predictions are similar and observed data were consistent with the estimated values in the fitted model (Appendix B).

The Nagelkerke value indicates that about 94.8% of the variance in the outcome variable (dumpsite suitability) is explained by the predictor variables (Appendix B). The likelihood ratio X² for testing the overall model (null hypothesis) is 19,894.299 with 186 degrees of freedom and a p-value of less than 0.0000. This tells us that the overall model fits better than we would expect by chance alone (Appendix B).

Appendix A summarizes the effect of each predictor variable. The sign of the coefficients of the independent variable gives important insights into the effects of the predictors/response variable in the model. The OR model shows that distance from borehole, distance from cultivated land, elevation, distance from the electric tower, distance from faults, lithology, LU/LC2019, distance from the protected area, distance from the road, distance from rural residents, distance from settlement/town, slope aspects, slope gradient/degree, soil depth, soil type and distance from surface water negatively and significantly affects the dumpsite suitability.

Therefore, the spatial-MCDA integrated approach can be concluded as practical in locating feasible and sustainable dumpsites and further used to be a preliminary database for experts to design landfill structures and also management operation for decision-makers.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Teshome Betru Tadesse

Teshome Betru Tadesse has received his master of science (M.Sc.) degree in Watershed Management from Hawassa University, Ethiopia. He has authored and co-authored five articles. And currently, he is a lecturer and researcher at Samara University, Ethiopia.

Setiye Abebaw Tefera

Setiye Abebaw Tefera has received a master of art (MA) degree in Environment and Development from Addis Ababa University, Ethiopia. He has authored and co-authored three articles. And currently, he is a lecturer and researcher at Samara University, Ethiopia.

Mengist Tesfaye Kidane

Mengist Tesfaye Kidane has been working as an Assistant Lecturer in the Department of Civil Engineering, Samara University, Ethiopia. And currently, he is a lecturer and researcher at Samara University, Ethiopia.

Appendix

Appendix A. Association of variables with dumpsite suitability on the ordinal regression model

Parameters		Estimate	Wald	df	Sig.
Threshold	Dumpsite_suitability = 1	−115.704	4239.623	1	.000
	Dumpsite_suitability = 2	−94.894	4156.890	1	.000
	Dumpsite_suitability = 3	−78.589	4050.981	1	.000
	Dumpsite_suitability = 4	−63.096	3818.561	1	.000
Location	aquifer_productivity = 1	−16.647	183.460	1	.000
	aquifer_productivity =2	−1.260	.617	1	.432
	aquifer_productivity =3	−.207	.032	1	.858
	aquifer_productivity =4	9.306	63.255	1	.000
	aquifer_productivity =5	0^a	.	0	.
	D.from_boreholes = 1	−1310.559	.	1	.
	D.from_boreholes =2	−6.162	5.856	1	.016
	D.from_boreholes =3	−18.941	259.771	1	.000
	D.from_boreholes =4	−10.132	118.578	1	.000
	D.from_boreholes =5	0^a	.	0	.
	D.from_cultland = 1	−45.435	32.836	1	.000
	D.from_cultland =2	−22.769	327.368	1	.000
	D.from_cultland =3	−12.950	1247.783	1	.000
	D.from_cultland =4	−6.694	146.157	1	.000
	D.from_cultland =5	0^a	.	0	.
	elevation = 1	−15.134	3010.325	1	.000
	elevation = 2	−10.152	2006.281	1	.000
	elevation = 3	−7.134	1492.279	1	.000
	elevation = 4	−3.413	1062.052	1	.000
	elevation = 5	0^a	.	0	.
	D.from_ectower = 1	−20.724	3644.786	1	.000
	D.from_ectower =2	−15.571	1964.613	1	.000
	D.from_ectower =3	−10.148	1119.019	1	.000
	D.from_ectower =4	−5.126	317.247	1	.000
	D.from_ectower =5	0^a	.	0	.
	D.from_fault = 1	−9.560	2831.255	1	.000
	D.from_fault =2	−6.995	2188.541	1	.000
	D.from_fault =3	−4.808	1181.807	1	.000
	D.from_fault =4	−2.112	256.697	1	.000
	D.from_fault =5	0^a	.	0	.
	lithology = 1	−48.274	499.656	1	.000
	lithology = 2	−35.921	752.802	1	.000
	lithology = 3	−36.144	454.921	1	.000
	lithology = 4	−23.420	376.073	1	.000
	lithology = 5	0^a	.	0	.
	lulc2019 = 1	−20.451	1052.554	1	.000
	lulc2019 = 2	−16.151	131.712	1	.000
	lulc2019 = 3	−10.757	3597.175	1	.000
	lulc2019 = 4	−5.237	2697.689	1	.000
	lulc2019 = 5	0^a	.	0	.
	D.from_protareas = 1	−30.510	282.587	1	.000
	D.from_protareas =2	−23.177	478.266	1	.000
	D.from_protareas =3	−14.749	571.260	1	.000
	D.from_protareas =4	−7.974	268.579	1	.000
	D.from_protareas =5	0^a	.	0	.
	D.from_road = 1	−28.297	3821.709	1	.000
	D.from_road =2	−21.184	3362.606	1	.000
	D.from_road =3	−13.918	2511.633	1	.000
	D.from_road =4	−6.617	946.837	1	.000
	D.from_road =5	0^a	.	0	.
	D.from_ruresident = 1	−18.378	323.877	1	.000
	D.from_ruresident =2	−14.941	437.070	1	.000
	D.from_ruresident =3	−9.066	572.088	1	.000
	D.from_ruresident =4	−4.419	275.495	1	.000
	D.from_ruresident =5	0^a	.	0	.
	D.from_settlement = 1	−39.799	957.556	1	.000
	D.from_settlement =2	−31.277	1349.795	1	.000
	D.from_settlement =3	−20.625	1889.433	1	.000
	D.from_settlement =4	−10.440	1858.584	1	.000
	D.from_settlement =5	0^a	.	0	.
	slope_aspect = 1	−18.213	4381.864	1	.000
	slope_aspect =5	0^a	.	0	.
	slope_gradient =2	−10.105	6.073	1	.014
	slope_gradient =3	−5.473	936.023	1	.000
	slope_gradient =4	−2.413	605.917	1	.000
	slope_gradient =5	0^a	.	0	.
	soil_depth = 1	−13.891	896.429	1	.000
	soil_depth = 2	−9.391	2158.409	1	.000
	soil_depth = 3	−5.046	647.245	1	.000
	soil_depth = 4	0^a	.	0	.
	soil_type = 1	−15.358	1253.660	1	.000
	soil_type = 2	−10.894	1745.370	1	.000
	soil_type = 3	−5.328	477.025	1	.000
	soil_type = 4	0^a	.	0	.
	D.from_Surf.water = 1	−47.223	4181.615	1	.000
	D.from_Surf.water =2	−35.187	4093.695	1	.000
	D.from_Surf.water =3	−24.135	3950.219	1	.000
	D.from_Surf.water =4	−11.805	3182.375	1	.000
	D.from_Surf.water =5	0^a	.	0	.

Notes. Link function: Cauchit

^aThis parameter is set to zero because it is redundant.

Appendix

Appendix B. Model parameters information and fitness tests

Model fitting information	2 Log Likelihood = 19,894.299; X^2= 150,797.354; Df = 62 and sig. = 0.000
Goodness of fit	Pearson X^2= 118,428.037; Df = 38,282 and Sig. = 0.000
Pseudo R^2	Cox and Snell = 0.900; Nagelkerke = 0.948 and McFadden = 0.770
Test of parallel line	−2 Log Likelihood = 19,894.299 (null hypothesis); X^2= 1043.552; Df =186 and Sig = 0.000