The study employed a mixed approach, incorporating both quantitative and qualitative research approaches, allowing for a multi-dimensional exploration of the research variables in Kasarani study area. Quantitative data collected included the quantity of solid waste generated, household size, number of bedrooms, area of the housing units, number of units per floor, number of units per block, number of people per floor, floor area, occupancy density, plot size, ground coverage, number of people in the cluster, cluster area, cluster population density, and frequency of solid waste collection. Qualitative data, gathered through observations and interviews, provided insights into waste separation, waste dumping, and open waste burning, helping the researcher understand residents’ solid waste management practices. Additionally, both quantitative and qualitative methods were used to evaluate the types of waste storage containers, waste transportation methods, road conditions, and locations of disposal sites. 105 housing units, 30 blocks, and 30 clusters obtained through simple random sampling were the subject of the study. The KoBoCollect application, which was used as the study’s data recording tool, incorporated skip-logic and validation criteria to enhance efficiency and accuracy when administering the interview guide. A simple analysis of the collected data was performed in KoBoToolbox before downloading in .CSV format for further analysis in Excel, SPSS, and ArcGIS. Data presentation was done in the form of tables, figures, and texts. As well, the study employed a multiple regression model to analyse the effect of spatial attributes and the solid waste management system, specifically focusing on waste generation. The independent variables considered at the housing unit level were: household size, number of rooms, housing unit size, household income, and occupancy density. The independent variables considered at block level were: block area (m²), number of people in the block, number of housing units, ground coverage, block occupancy density. The dependent variable at each level was the amount of waste generated. The study’s indicator system’s formation involved a thorough analysis of existing literature on waste generation and supervisory consultations to ensure that the chosen variables are relevant and capable of providing meaningful correlations with solid waste generation.

Expression of the model:

$Y={\beta }_0+{\beta }_1{X}_1+{\beta }_n{X}_n+\epsilon$(1)

where:

Y is the amount of waste generated;

${\beta }_0$ is a constant equivalent to the dependent variable if no independent variable is influencing the dependent variable;

$X_1$ is the first independent variable;

$X_n$ is the n^th independent variable;

${\beta }_1$ is the coefficient of $X_1$;

${\beta }_n$ is the coefficient of $X_n$;

$\epsilon$ is the error.

The primary contributors to solid waste within the study area included food waste, food soiled paper, plastic, textile (e.g., clothing, linens, and towels), metal, glass, rubber, leather, cardboard, special waste (e.g., sanitary towel, diaper, cleaning products, sharp objects such as needles and beauty-related waste products), and electronic waste (e-waste), such as bulbs, batteries, phones, and sockets among others. Organic waste contributed more than 50% of the solid waste generated in the study area. The approximate volume of waste generated per week per housing unit was 8.64 kg, with the least recorded output of 1 kg and the highest amount of solid waste generated of 20 kg. The per capita waste generation rate was calculated by dividing total average amount of solid waste generated per household by the average household size per the number of days the waste was generated, see Equation (2). The formula was also employed to calculate the per capita waste generation in studies conducted by Dangi et al. (2011) and Dikole & Letshwenyo (2020) .

$g=\frac{G}{d\cdot h}$(2)

where:

g is the amount of waste generated per person per day;

G is the average amount generated by a housing unit per week;

d is the number of days the housing unit waste volume was averaged;

h is the average household size;

G = 8.64 kg;

d = 7 days (1 week);

h = 4.

Thus,

$g=\frac{8.64}{7\times 4}=0.31\text{kg}$

Equation (3) illustrates how the approximate volume of waste generated per block per week was obtained:

$G_B=G\times H_B$(3)

where:

$G_B$ is the approximate average volume waste generated per block;

G is the average amount generated by a housing unit per week;

$H_B$ is the average total number of housing units per block.

$G=8.64\text{kg}$

$H_B=23$

Thus,

$G_B=8.64\text{kg}\times 23$

$G_B=198.72\text{kg}$

Equation (4) illustrates how the approximate volume of waste generated per cluster per week was computed:

$G_C=G_B\times B_C$(4)

where:

$G_C$ is the approximate average volume waste generated per cluster;

$G_B$ is the average amount generated per block per week;

$B_C$ is the average total number of blocks per cluster.

$G_B=198.72\text{kg}$

$B_C=25$

Thus,

$G_C=198.72\text{Kg}\times 25$

$G_C=4968\text{kg}$

Kasarani residents store their generated solid waste in single storage containers. 48% are stored in lidded plastic containers, 28% in plastic bags, and the remaining 14% in unlidded bins, see Figure 2 and Figure 3 . These types of households’ storage containers discourage waste segregation at source since most residents lump together their generated waste. Regrettably, mixed solid waste presents challenges to waste recyclers in sorting and processing, impeding recycling efforts ( Sheau-Ting et al., 2016 ). Additionally, this practice fosters waste contamination, indirectly impacting human health. For instance, some of the waste pickers retrieve organic waste from the disposal sites and sell it to pig farmers in the neighbourhood, who in turn sell the pigs to the nearby slaughterhouses. From the slaughterhouse, the meat is supplied to butcher shops in the neighbourhood. Ultimately, the consumers who purchase the pork meat are indirectly affected by the contaminated food waste. Furthermore, the unlidded solid waste storage containers have several adverse effects on the neighbourhood. They encourage intrusion by rodents and houseflies, which can spread diseases and create unsanitary conditions.

82% of the study area blocks were devoid of waste storage containers. Similar to the inadequate space for other essential facilities, like parking facilities, in a block due to high ground coverage, residential blocks lack space for storage containers. Most developers prioritize the optimum number of housing units over them. Unfortunately, this has a ripple effect on waste segregation at the housing unit level. The residents were demotivated to segregate their solid waste at the household level since it would be lamped together at the gate on waste collection day.

Commendably, 98% of the Kasarani study area receives waste collection services; this stands out compared to the 54.6% of Kenya’s urban population that receives the same services and the 80% of the whole Nairobi City County population ( KNBS, 2019 ). 60% of them receive services from community-based organizations. The remaining 31% receive the services from private companies. The county government doesn’t offer waste collection service to this neighbourhood. The solid waste is majorly collected once a week for an average fee of Ksh. 150 per month. Generally, the blocks are devoid of space for solid waste storage or collection, as mentioned earlier. Thus, on the block’s scheduled day for waste collection, the residents carry their garbage from their housing units to the plot’s entrance, awaiting collection by the contracted waste collector, see Figure 4 . However, in cases of delays in waste collection by the contracted waste collectors, the residents experience bad odors and flies in the surrounding area.

90% of the solid waste generated in the study is transported via cart to the disposal areas, see Figure 5 . The remaining 10% of the garbage is transported via private trucks. The cart pullers resort to dumping their waste in their preferred accessible space within the study area due to the extensive distance to the county-designated space (Dandora dumpsite), which is 15 km from the study area. The study area lacks a designated space for waste disposal, collection, or transfer. The human-pulled carts are open and prone to leaks, resulting in waste dropping and leachate dripping during transportation, contaminating the environment, e.g., soil and water sources, and posing risks to human health.

The study area is characterized by open-space waste dumping, including on vacant plots, riparian reserves, railway reserves, and street dumping, see Figure 6 . The open dumping of solid waste is attributed to the extensive distance the cart pullers have to cover to reach the designated disposal site, 15 km from the study area. It is also ascribed to the lack of solid waste collection facilities, which act as a linkage between residential blocks, waste recyclers, and the designated disposal site. Open dumping poses significant environmental and public health hazards. The compounded waste contaminates the soil, water sources, and air. As well, the disposal of waste in the rivers and drainage channels causes blockage, affecting storm water flow and leading to flooding during rainy seasons. To reduce the volume of solid waste accumulated in the dumping areas, residents burn the accumulated waste, see Figure 7 . This mode has health and environmental adverse effects, such as the emission of greenhouse gases (GHG), such as ammonia (NH₃), carbon dioxide (CO₂), carbon monoxide (CO), hydrogen sulphide (H₂S), methane (CH₄), nitrogen dioxide (NO₂), sulphur dioxide (SO₂), plus particulate matter in the air ( Jakhar et al., 2023 ; Ramadan et al., 2022 ).

Spatial attribute at housing unit level. The spatial attributes at the housing unit level included household size, number of rooms in the housing unit, housing unit area size (m²), household income (Ksh.), and housing unit occupancy density (persons/m²).

Correlation analysis. To explore the relationships between spatial attributes and the amount of solid waste generated at housing unit level, a correlation analysis was conducted. The variables included in this analysis were: household size, number of rooms, housing unit area size (m²), household income (Ksh.), housing unit occupancy density, and amount of solid waste generated by the housing unit. Pearson’s correlation coefficient was used to measure the strength and direction of the linear relationships between these variables. The results are summarized in Table 1 .

From Table 1 , household size, number of rooms, housing unit size (m²), and household income (Ksh.) had a positive correlation with the amount of solid waste generated by a housing unit at 0.709, 0.534, 0.547, and 0.474, respectively. This means that as they increase, the amount of solid waste also increases, and vice versa. The increase in household size with the amount of waste generated by the housing unit is in line by another study by Trang et al. (2017) . In the study area, housing units with smaller household sizes often depend on nearby motels

Table 1. Correlation matrix table showing correlations between the spatial attributes and solid waste generation at housing unit level.

**. Correlation is significant at the 0.01 level (2-tailed).

or informal food vendors for cooked meals and pre-prepared vegetables, thereby minimizing the amount of solid waste they generate. The increase in solid waste generated by household income levels is supported by other studies ( Visvanathan & Trankler, 2003 ). Housing unit occupancy density had a weak negative correlation with the amount of solid waste generated at −0.135. This means that when the housing unit occupancy density increases, the amount of solid waste decreases insignificantly.

Regression analysis. A stepwise regression analysis was conducted to model the relationship between housing unit spatial attributes and the solid waste management system, specifically focusing on waste generation. The independent variables considered in this analysis were: household size, number of rooms, housing unit size, household income, and occupancy density. These variables were entered into SPSS for the analysis. The dependent variable was the amount of solid waste generated by the housing units. The stepwise regression approach was chosen for its ability to identify the most significant predictors by systematically adding or removing variables based on their p-values. This approach helps in refining the model to include only those variables that have a statistically significant impact on the dependent variable. The results of the regression analysis are summarized in Table 2 , Table 3 and Table 4 .

Table 2. Model summary regression of solid waste generation and spatial elements at housing unit level.

a. Predictors: (Constant), Household size; b. Predictors: (Constant), Household size, Housing unit area.

Table 3. Sample ANOVA for regression of solid waste generation and spatial elements at housing unit level.

a. Dependent Variable: Amount of waste; b. Predictors: (Constant), Household size; c. Predictors: (Constant), Household size, Housing unit area.

Table 4. Regression coefficients for solid waste generation and spatial elements at housing unit level.

a. Dependent Variable: Amount of waste.

From the stepwise regression, two models were generated, illustrated by Equations (5) and (6). In the first model, the household size was identified as the most significant predictors by systematically excluding the number of rooms, housing unit size, household income, and occupancy density basing on their p-values. In the second model, household size and housing unit area size were identified as the most significant predictors whereas number of rooms, household income, and occupancy density were excluded. The F and T tests were used to determine the significance of the models and parameters. Cumulatively, the spatial elements predictors explain 49.8% and 53% of the variance waste generation in the two models respectively, Table 2 . The regression model was significant as F < 0.10, Table 3 . In the model, household size and housing unit area size (m²) elements were significant at t < 0.10, Table 4 .

Model 1

$ASWG=0.907+0.709HS+1.49$(5)

Model 2

$ASWG=0.774+0.585HS+0.227HUA+1.44$(6)

where:

ASWG is the amount of solid waste generated at the household level;

HS is the household size;

HUA is the housing unit area (m²).

Spatial attribute at block level. The spatial attribute at block level entailed block area (m²), number of people in the block, number of housing units, ground coverage, block occupancy density (persons/m²).

Correlation analysis. To explore the relationships between neighbourhood spatial attributes and the amount of solid waste generated at block level, a correlation analysis was conducted. The variables included in this analysis were: block area (m²), number of people in the block, number of housing units, ground coverage (%), block occupancy density (people/m²), and amount of solid waste generated (kg) in the block. The results are summarized in Table 5 .

Table 5. Correlation matrix table showing correlations between the spatial attributes and solid waste generation at block level.

**. Correlation is significant at the 0.01 level (2-tailed).

The Pearson Correlation (r) explained the strength of the correlation. From Table 5 , block area (m²), ground coverage, block occupancy density, number of housing units, and the number of people in the block, have a positive correlation with waste generation element of 0.178, 0.261, 0.832, 1, and 1 respectively. This means that when they increase, the amount of solid waste also increases, and vice versa. For instance, a residential block with a higher number of housing units generates more volume of waste compared to a block with a lower number of housing units.

Regression analysis. A regression analysis was also conducted at block level to develop the waste generation model. The F and T tests were used to determine the significance of the model and parameters, respectively. Cumulatively, the spatial elements predictors explain 92.9% of the variance waste generation, see Table 6 . The regression model was significant as F < 0.10, see Table 7 . In the model, the block area and occupancy density were significant at t < 0.10, see Table 8 .

Model:

$AG=-179.268+0.525BA+0.059GC+0.986OD+27.96$(7)

where:

AG is amount of waste generated;

BA is block area;

GC is the ground coverage;

OD is occupancy density.

a. Predictors: (Constant), Occupancy density, Block area, Ground coverage.

Table 7. ANOVA for regression of waste generation and spatial elements at block level.

a. Dependent Variable: Amount of waste; b. Predictors: (Constant), Occupancy density, Block area, Ground coverage.

Table 8. Regression coefficients for waste generation and spatial attributes at block level.

a. Dependent Variable: Amount of waste.

The occupancy density and block area emerged as the significant predictors of variations in waste generation element at 98.6% and 52.5% respectively, see Table 8 .

Spatial attribute at block level. The spatial attribute at cluster level entailed cluster area (m²), cluster population, and cluster population density.

Correlation analysis. To explore the relationships between neighbourhood spatial attributes and the amount of solid waste generated at cluster level, a correlation analysis was conducted. The variables included in this analysis were: cluster area (m²), cluster population, cluster population density (persons/m²), and amount of solid waste generated by the cluster. The results are summarized in Table 9 .

Table 9. Correlation matrix table showing correlations between the spatial attributes and solid waste generation at cluster level.

*. Correlation is significant at the 0.05 level (2-tailed); **. Correlation is significant at the 0.01 level (2-tailed).

The Pearson Correlation (r) explained the strength of the correlation. From Table 9 , cluster area (m²), population of the cluster, cluster population density, have a positive correlation with waste generation element of 0.642, 1, and 0.310, respectively. This means that when they increase, the amount of solid waste generated also increases, and vice versa. For instance, a cluster that is highly populated generates more solid waste compared to a cluster that is less populated.

All residential blocks to have a 9n kg capacity solid waste storage container (where n is the number of housing units and 9 is the average amount of waste generated by a housing unit). As well, the block’s storage container should have six compartments to accommodate the segregated solid waste from the nth housing units. These include: organic waste, plastic waste, metal and glass waste, electronic waste, sanitary waste (diapers and sanitary towels) and hazardous waste, and other inorganic waste. This will motivate the residents to continually segregate their waste at source. This practice is key for solid waste material recycling due to the minimal contamination that arises from solid waste compounding. Figure 8 illustrates partitioned housing units’ storage container that will feed to the compartmentalized block’s solid waste storage container:

Averagely, a residential block generates about 198.72 kg. A rickshaw with a 200-kg capacity can accommodate the amount of solid waste generated by the block. The rickshaw will transport the solid waste to a collection center within a radius of 1 km. The rickshaw should also be partitioned to encourage further solid waste segregation. Figure 9 illustrates a rickshaw of 200 kg capacity collecting waste from the block storage container.

Solid waste collection facilities should be installed in the high-density residential neighbourhood. Six residential clusters should be served by one collection facility of 30-ton capacity. The capacity of the solid waste collection facility was computed with the formula below:

$C=AN$(8)

where:

C is the capacity of the collection facility;

A is the average amount of waste generated by a cluster;

N is the number of clusters to be served by the collection facility;

A = 5 tonnes;

N = 6.

$C=5\text{tonnes}\times 6=30\text{tonnes}$

The collection facilities should be partitioned to allow further waste segregation. The section for organic solid waste should be substantial since it takes a significant share of the solid waste generated (more than 50%). These segregated organic solids can be sold to the pig farmers within the study area since they are not contaminated. The remaining less fibrous organic solid waste can be used for biogas production, and the more fibrous waste can be used for making compost to be used as organic fertilizer. Similarly, the segregated inorganic solid waste can be recovered by waste recyclers. The remaining segregated waste is to be taken to the material recovery facility, where the other waste recyclers can get their recoverable. Figure 10 illustrates the proposed solid waste collection facilities.

Figure 11 illustrates how six clusters can be served by a solid waste collection facility that is also compartmentalized to allow material recovery and reduce the amount of solid waste heading to the designated disposal site, where further material recovery should happen.

A 30-ton truck should be involved in transporting solid waste material from the collection facility to the material recovery facility, located in Dandora, which enhances further waste material recovery, see Figure 12 . Only materials that have reached their lifespan should be sustainably disposed of through a sanitary landfill.

All roads linking the residential blocks to solid waste collection facilities should be at least 6 meters wide and be upgraded to cabro standard so as to ease waste transportation in all-weather conditions.