Solid Waste Disposal Site Selection using Boolean Logic
- Lynette Dias
- Feb 19
- 9 min read
Colombia is a country in South America that shares its borders with Venezuela, Peru, Brazil and Ecuador. It has a population of 49.65 million as of 2018 (www.worldbank.org). Each day the country produces 32,000 tons of solid waste, or 0.68kg/capita per day. The country’s biggest cities- Bogotá, Cali, Medellín, and Barranquilla generate over 10,000 tons/day. In 2016 the Colombian Government planned to revolutionise its solid waste management plans and make it more sustainable. Some of the hurdles in the plan were finding ways to improve landfilling technologies and develop a more efficient waste transport system (www.rwsenvironment.eu).
The city of Chinchina lies in the Caldas department of Colombia with a population of approximately 150,000. Like in most other cities in Colombia, the solid waste generated here is usually dumped in ravines and rivers. The closest landfill to the city is the Relleno Sanitario La Esmeralda which it shares with 15 other municipalities and is 27km from it (www.chinchina-caldas.gov.co). Increasing environmental awareness led the city of Chinchina to develop a proper waste disposal site keeping in mind both ecological and social constraints.
This study aims to identify suitable areas for the development of a waste disposal site using Boolean logic modelling based on a list of criteria prepared by a team of specialists in geology, geomorphology, hydrology, civil engineering, and regional planning. Additionally, the suitable sites have to meet the areal capacity criterion and should be accessible by road.
Study Area
The study area covers an area of approximately 68km2 with the city of Chinchina located in the South Western part. Chinchina is one of Colombia’s most important coffee producing regions. The city is located in a valley surrounded by coffee plantations. Hence coffee factors its way into the site selection criterion. The terrain has a gentle slope in most parts.
The Process
The first step involved data preparation. The datasets available were vectors (landslide, landuse, lithology, roads), a raster (slope) and a table (drillhole). The selection criteria and their corresponding dataset are shown in table 1. The site selection process had three stages:
1. Identifying suitable areas by creating binary spatial criteria maps for each factor and then combining them.
2. Selecting the areas deemed suitable that also met the areal capacity criterion.
3. Choosing the final sites as those areas which met the areal capacity criterion and were accessible by road.
The entire methodology is enumerated in the flowchart in Figure 1.
Binary Selection Criterion Mapping
The process used to visualise each criterion in a Geographical Information System (GIS) was by creating a binary spatial criterion map using Boolean logic modelling. This process yields a map with pixels that have two output values which may be 0 or 1. If the criterion is satisfied then the pixel value is 1 if not it gets a value 0.
For the vector layers, the binary map was created by first creating a new attribute in the attribute table for the layer and setting the value as 0 or 1 based on if the pixel satisfied or didn’t satisfy the criterion for that layer. This was done because qualitative variables like the classes in the vector layers have to be represented numerically so as to be used in mathematical operations (Carranza, 2008). This attribute was then selected for the burn in value field when converting the layer to a raster. All the converted rasters were given a 12.5m spatial resolution so that they could be combined into one map later. The procedure for each individual map is explained below.
Site Selection Criteria
| Criterion | Input dataset |
Slope | <20 degrees | Slope raster |
Landslide | Not active or dormant | Landslide distribution vector |
Landuse | Pasture or shrubs or bare | Landuse vector |
Proximity to BUA | >300m | Landuse vector |
Proximity to city | <2km | Landuse vector |
Geology: |
| Lithology vector+ drillhole table |
Soil thickness | ≥5m per day | drillhole table |
Clay content | >50% | drillhole table |
Permeability | <5m per day | drillhole table |
Area | ≥1 hectare (=10,000m2) | |
Accessibility | Easily accessible by road | Open street Maps road vector |
Slope map
This layer was already in raster format with each pixel having a DN value equal to the slope in degrees for that location on the ground. It was reclassified using the reclassify by table tool to satisfy the slope criterion. Slopes from 0-20 degrees were given a value 1 and those from 20-75 degrees were given a value 0. This yielded a binary slope criterion map.
Geology map
The two datasets used to create the geology criterion map were a shapefile of the 14 different lithologies present in the study area as polygons, and a table with drillhole data from various boreholes across the study area. The selection criterion was based on thickness, permeability and percentage of clay found in a given layer all of which was stored in the drillhole table, hence the two layers had to be joined.
The drillhole table was first loaded into QGIS and then exported as a comma-separated values file. This converted it into a vector point layer and enabled it to be displayed in QGIS over the lithology layer. A spatial join was then created between the lithology layer and the drillhole layer using the join attributes by location tool. All fields were summarised and ‘min’ and ‘max’ values were selected as the summaries to be calculated.
A new attribute ‘value’ was created in the attribute table and those lithologies which satisfied the lithology criterion were given a value 1. The fields were selected using the expression:
("THICKNESS_min">=5) AND ("PERCCLAY_max">50) AND ("PERMEABILITY_max"<5)The vector was then converted to a raster using a burn in value of 1 on the value attribute yielding a binary geology criterion map.
Landslide map
The layer was present as a shapefile with polygons of the 4 landslide classes- active, dormant, stable and none. According to the criterion areas with active landslides or those that could become active in the future were not desirable. A new attribute ‘value’ was created in the attribute table and those polygons which were stable or had no landslides were given a value 1. The fields were selected using the expression:
"slide" = 'No landslides' OR "slide" = 'Stable' The vector was then converted to a raster using a burn in value of 1 on the value attribute yielding a binary landslide criterion map.
Proximity to City map
The outline of Chinchina city was extracted as a raster layer from the landuse shapefile. For this, in the ‘city value’ attribute all pixels which had ‘Chinchina’ as a field in the attribute ‘city’ were given 1. The landuse vector layer was converted to raster using ‘city value’ as the field for burn in value. This yielded a map with 1 for pixels in the city and 0 for those outside it. Then the proximity (raster distance) tool was used to select pixels at a distance of 2km from the city. These pixels were given a value 1 as we want a site at most 2km from the city which is the city proximity criterion. Pixels of areas >2km were given the value 0. The city itself was given a value 0 because we cannot have a waste disposal site in the middle of the city. This process gave us a city proximity criterion binary map.
Proximity to built-up area (BUA) map
The built-up areas (BUAs) were extracted as a raster layer from the landuse shapefile. In the ‘BUA value’ attribute all pixels which had ‘Buil-up area’ as a field in the attribute ‘landuse’ were given 1. The landuse vector layer was converted to raster using ‘BUA value’ as the field for burn in value. This yielded a map with 1 for pixels in built up areas and 0 for those outside it. The raster proximity tool was used to highlight areas less than 300m from built-up areas. These values were given a value 0 to satisfy the built-up area proximity criterion. The result was a built-up area proximity criterion binary map.
Landuse map
Three new attributes were created- ‘city value’, ‘BUA value’, and ‘landuse value’. For the attribute ‘landuse value’ the classes of little ecological and economic importance like pasture and shrubs were given the value 1 according to the landuse criterion. Classes of economic importance like coffee and ecological importance like forests were given the value 0. Built-up area was also given the value 0. The vector layer was then converted to raster using a burn in value of 1 for the attribute ‘landuse value’ to create a binary landuse criterion map.
Suitable Areas Map
All the binary criterion maps were combined using the raster calculator tool. Pixels of the same ground resolution cell in all the layers, which satisfied all the criterion and had a value 1 got a value 1 in the final map whereas those that had even one 0 got a final value of 0. The result was a binary map showing suitable areas for waste disposal that satisfy all the criterion.
The expression used was:
"slope_binary@1" AND "geo_binary@1" AND "landslide_binary@1" AND "city_proximity_binary@1" AND "bua_proximity_binary@1" AND "landuse_binary@1"
Areal Capacity of suitable sites
The suitable areas binary map was converted to a vector layer. The area of each polygon in the vector layer was computed using the field calculator in the attribute table and is shown in the table below. An attribute called ‘value’ was added and all polygons with areas of at least 1 hectare were given a value 1 in this attribute. Then the vector was converted to a raster using ‘value’ as a field for the burn in value resulting in a map of suitable areas for waste disposal that met the areal capacity criterion.
The map shows two suitable regions that have an area larger than 1 hectare (=10,000m2). Larger areas are preferred because developing the site for use as a landfill takes time and hence it is desirable to have a larger site that can be used for at least 30 years and can support the increase in solid waste of a growing population.
Larger areas also help provide a buffer to the areas surrounding the site (Preliminary Landfill Site
Suitability Report, 2004).
Site | Area |
Site 1 | 70,731 m2 |
Site 2 | 15,614 m2 |
Site 1 would be preferred over 2 in this case since it is almost 4 times bigger.
Accessibility of suitable sites
According to the regional and local screening method for selecting sites, the sites must be at most 1km from a road to make it economically viable (Derakhshandeh et al., 2014), and the Colombian Government states that the highest weightage for accessibility must be given to sites between 0-5km from the road (MAVDT, 2005).
For this step, a vector layer of roads was downloaded from Open Street Maps. The layer had different types of roads which were divided into 2 main categories: major-primary, secondary, tertiary, and trunk roads, and minor- residential and unclassified. For this study, the distance from minor roads was set as 150m and major roads as 300m to be considered accessible and hence a 300m buffer was created around the major roads and 150m around the minor roads. Each buffer was then displayed over the suitable area raster to analyse the accessibility.
Both sites are accessible by major roads as well as minor roads however site 1 is accessible from one and very close to another minor road making it a better choice, in case of a roadblock or traffic jam on one road.
Accuracy
The accuracy of the site selection was checked by overlaying the selected site polygons over Google Earth imagery. Both sites were located very close to coffee plantations and part of Site 1 was in a plantation. This could be the result of inaccurate or outdated landuse maps which would have to be edited for a better result.
Conclusions
This study led to the identitfication and selection of two suitable sites for waste disposal. Six binary criterion maps were created and then their results were integrated into one map showing the suitable areas. The sites were then narrowed down to two after evaluating their areas and their accesssibility from major and minor roads. The accuracy obtained was not quantified, but the wrong classification of a landuse criterion could imply that the datasets provided need to be updated.
This study using Boolean Logic led to the identitfication and selection of two suitable sites for waste disposal. However, a weighted overlay or a fuzzy logic overlay could result in a map showing not just suitable and unsuitable areas, but a range of suitability that would be more beneficial to the city governance.
The focus of this study was waste disposal site selection for the use of a city, however that isn’t the end of the road for city planners. The next task would be a field survey and acquiring land permits and having just two suitable sites would create problems in case both sites were on private land or had some unforeseen issue. This study used Boolean Logic, however, a weighted overlay or a fuzzy logic overlay could result in a map showing not just suitable and unsuitable areas, but a range of suitability that would be more beneficial to the city governance.
References
Carranza, E. J. M. (2008). Geochemical anomaly and mineral prospectivity mapping in GIS. (Handbookof exploration and environmental geochemistry; Vol. 11, pp. 28-29). Amsterdam: Elsevier.
Derakhshandeh, M., & Taleb Beydokhti, T. (2014). Management of landfill locating of urban waste. European Online Journal of Natural and Social Sciences: Proceedings, 3(3 (s)), pp-32.
Guam environmental protection agency and department of public work (2004). Preliminary landfill site suitability report, pp.11.
MAVDT, Decree 838, Location of areas for the final disposal of solid waste., Ministry of the Environment, Housing and Territorial Development, Official Gazette No. 45862, (2005). (Translated from Spanish)
World Bank GDP Growth. Retrieved October 24, 2020, from
Rijkswaterstaat Environment project. Retrieved October 22, 2020, from https://rwsenvironment.eu/projects/colombia/colombia/
Solid Waste Disposal Plan Retrieved October 24, 2020, from
Comments