prioritization of road maintenance funding in south africa ... · informative); however, the final...

1
Prioritization of Road Maintenance Funding in South Africa By Oscar Gonzalez & Diogo Prosdocimi May 3, 2016 To more clearly display the results, we will zoom into one particular district within South Africa (Namawka). However, our analysis did create the Road Maintenance Prioritization network for the entire country. Seeing the detail for a large and dense road network (i.e., entire country) is not as visually useful. Figure 1: Network Distance Impedance Service Areas for Schools in Namakwa District (in minutes) Figure 2: Network Distance Impedance Service Areas for Healthcare Facilities in Namakwa District (in minutes) Figure 3: Census Tract Job Count in Namakwa District The road network for the spatial join of census tract job density value (from 0 to 5) was created and used in the suitability analysis to create the composite network with values from 0 to 15, but it was not able to be displayed due to technical issues with ArcGIS. Below is the composite road network result for the entire Namakwa District. Figure 4: Composite Map of Road Network Accessibility for Road Maintenance Prioritization in South Africa’s Namakwa District This was displayed in the symbology of standard deviations for visual purposes (a legend of 16 options was not visually informative); however, the final results do assign 0 to 15 values to each road segment. Therefore, the government of South Africa (including the District government of Namakwa) has a simple ordinal system of accessibility that it may use to prioritize road maintenance projects. Figure 5: Final Result Distribution of Accessibility Weight Score for all Roads throughout South Africa (in kilometers) All points and spatial data were geocoded and/or re-projected into the Hartebeesthoek_1994 datum (WGS84 projection), which is the standard for spatial datasets in South Africa. 2. Network Analysis With the Network Analyst toolkit, we created service areas with 5 different impedances for the school and healthcare facilities layers (see table below). Then, we assigned each road segment the weight value from 0 to 5 (5 being the closest to the facility) based on the impedance to the facilities. For employment, we spatially joined the weighted value of census data jobs (again, from 0 to 5 with 5 being the highest amount of jobs for that tract) onto the road segments within each census tract. 3. Suitability Analysis Ultimately, we conducted a simplified opportunities-based suitability analysis with the three layers that we generated above in the Network Analysis. Each layer converted each road segment into a weighted field of 0 to 5 (5 being the most accessible for that layer). We simply added these polyline layers by adding the weight fields together (see pictogram below). This process effectively creates a composite map where each road is assigned a value of accessibility weight from 0 to 15, which creates our Road Maintenance Prioritization system. This study had some notable limitations that should be addressed in further research, or modeling, of the topic. Primarily, accessibility was defined with only three parameters. Other major social service facilities (i.e., police stations, fire stations, family resource centers, institutions of higher education, child care facilities, food access) were not able to be included in this analysis. Also, the network data that was available did not include congestion or juncture data, so it does not account for stop signs, traffic lights, and other factors that affect flow and congestion. The government of South Africa, and its districts, currently does not have a comprehensive system or established goals for determining which roads to prioritize for the use of limited government funding for road maintenance projects. While most countries and transportation planning departments have a well-defined strategic plan for Road Maintenance Prioritization (RMP), the government of South Africa does not. In fact, it relies primarily on elected officials making claims for their desired RMP and having the elected bodies vote on the most compelling case. There are no established standards for RMP beyond selecting the roads, or road segments, that received more votes based on the persuasiveness of an elected official. Currently, researchers at the University of Cape Town, such as Don Ross and Matthew Townshend, are attempting to nudge the government into creating and implementing a system for RMP that has defined goals regarding accessibility, instead of relying on political will. For this study, we shall focus on three major factors for accessibility, arguing for the government of South Africa to prioritize its road maintenance funding based on accessibility to: (1) schools, (2) healthcare facilities, and (3) employment. Ultimately, we will create an ordinal model in which the government of South Africa may more efficiently choose which roads to prioritize for government funding (on a 0 to 15 scale). BACKGROUND & PROBLEM METHODS [continued] Ross, D. & Townshend, M (2014). Can South Africa road authorities satisfy constitutionally protected basic access needs without sacrificing economic growth? http://www.academia.edu/19750954/Can_South_African_road_authorities_satisfy_constitutionally_protected_basi c_access_needs_without_sacrificing_economic_growth RESULTS [continued] RESULTS REFERENCES DATA We obtained our data from the following sources: For data on schools, we used data from an educational survey conducted by the Department of Education (DOE) http://www.education.gov.za For data on healthcare facilities, we used data from 2011-2012 South Africa Hospital Survey (SAHS) https://africaopendata.org/dataset/south-african-hospitals-survey- 2011-2012 For data on employment, population, and other demographic variables, we used data from the Council for Scientific and Industrial Research (CSIR) http://www.csir.co.za Weight Schools Healthcare Facilities Employment 5 1 min 5 min 11898+ jobs 4 3 min 10 min 1189-11897 jobs 3 5 min 15 min 120-1188 jobs 2 10 min 30 min 13-119 jobs 1 15 min 45 min 2-12 jobs 0 15+ min 45+ min 0-1 jobs METHODS We utilizing three modeling techniques to create our accessibility-based RMP system for the entire country of South Africa: 1. Geocoding While some of the data included geographic coordinates, for some of the facilities (251 of the hospitals and 2,0382 of the schools), we only had addresses. We created an Address Locator, and with the road network of South Africa (from CSIR), we matched all but 31 of the hospitals (98.99% match rate) and 789 of the schools (96.9% match rate). For employment, we spatially joined the census data on number of jobs in each census block to the centroid of each block. The geocoding results are below: Layer / Type Total Number Matched Match Rate Schools 25,720 24,931 96.9% Healthcare facilities 3,077 3,046 98.99% Employment Over 9.5 million in 25,000 census tracts Over 9.5 million in all tracts 100% 0 0.45 0.9 0.225 Decimal Degrees 4 Legend Namakwa Employment # of Jobs 0 1 2 - 6 7 - 33 34 - 180 181 - 963 964 - 5163 0 0.45 0.9 0.225 Decimal Degrees 4 Legend 1 3 5 10 15 Namakwa Roads 0 0.45 0.9 0.225 Decimal Degrees 4 Legend Namakwa Roads 5 10 15 30 45 0 0.45 0.9 0.225 Decimal Degrees 4 Legend Composite Analysis Total < -0.50 Std. Dev. -0.50 - 0.50 Std. Dev. 0.50 - 1.5 Std. Dev. 1.5 - 2.5 Std. Dev. > 2.5 Std. Dev. Namakwa LIMITATIONS

Upload: others

Post on 27-Sep-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Prioritization of Road Maintenance Funding in South Africa ... · informative); however, the final results do assign 0 to 15 values to each road segment. Therefore, the government

Prioritization of Road Maintenance Funding in South AfricaBy Oscar Gonzalez & Diogo Prosdocimi

May 3, 2016

To more clearly display the results, we will zoom into one particular district within South Africa (Namawka). However, our analysis did create the Road Maintenance Prioritization network for the entire country. Seeing the detail for a large and dense road network (i.e., entire country) is not as visually useful.

Figure 1: Network Distance Impedance Service Areas for Schools in Namakwa District (in minutes)

Figure 2: Network Distance Impedance Service Areas for Healthcare Facilities in Namakwa District (in minutes)

Figure 3: Census Tract Job Count in Namakwa District

The road network for the spatial join of census tract job density value (from 0 to 5) was created and used in the suitability analysis to create the composite network with values from 0 to 15, but it was not able to be displayed due to technical issues with ArcGIS.

Below is the composite road network result for the entire Namakwa District.

Figure 4: Composite Map of Road Network Accessibility for Road Maintenance Prioritization in South Africa’s Namakwa District

This was displayed in the symbology of standard deviations for visual purposes (a legend of 16 options was not visually informative); however, the final results do assign 0 to 15 values to each road segment. Therefore, the government of South Africa (including the District government of Namakwa) has a simple ordinal system of accessibility that it may use to prioritize road maintenance projects.

Figure 5: Final Result Distribution of Accessibility Weight Score for all Roads throughout South Africa (in kilometers)

All points and spatial data were geocoded and/or re-projected into the Hartebeesthoek_1994 datum (WGS84 projection), which is the standard for spatial datasets in South Africa.

2. Network AnalysisWith the Network Analyst toolkit, we created service areas with 5 different impedances for the school and healthcare facilities layers (see table below). Then, we assigned each road segment the weight value from 0 to 5 (5 being the closest to the facility) based on the impedance to the facilities. For employment, we spatially joined the weighted value of census data jobs (again, from 0 to 5 with 5 being the highest amount of jobs for that tract) onto the road segments within each census tract.

3. Suitability AnalysisUltimately, we conducted a simplified opportunities-based suitability analysis with the three layers that we generated above in the Network Analysis. Each layer converted each road segment into a weighted field of 0 to 5 (5 being the most accessible for that layer). We simply added these polyline layers by adding the weight fields together (see pictogram below).

This process effectively creates a composite map where each road is assigned a value of accessibility weight from 0 to 15, which creates our Road Maintenance Prioritization system.

This study had some notable limitations that should be addressed in further research, or modeling, of the topic. Primarily, accessibility was defined with only three parameters. Other major social service facilities (i.e., police stations, fire stations, family resource centers, institutions of higher education, child care facilities, food access) were not able to be included in this analysis.

Also, the network data that was available did not include congestion or juncture data, so it does not account for stop signs, traffic lights, and other factors that affect flow and congestion.

The government of South Africa, and its districts, currently does not have a comprehensive system or established goals for determining which roads to prioritize for the use of limited government funding for road maintenance projects. While most countries and transportation planning departments have a well-defined strategic plan for Road Maintenance Prioritization (RMP), the government of South Africa does not. In fact, it relies primarily on elected officials making claims for their desired RMP and having the elected bodies vote on the most compelling case. There are no established standards for RMP beyond selecting the roads, or road segments, that received more votes based on the persuasiveness of an elected official.

Currently, researchers at the University of Cape Town, such as Don Ross and Matthew Townshend, are attempting to nudge the government into creating and implementing a system for RMP that has defined goals regarding accessibility, instead of relying on political will.

For this study, we shall focus on three major factors for accessibility, arguing for the government of South Africa to prioritize its road maintenance funding based on accessibility to: (1) schools, (2) healthcare facilities, and (3) employment.

Ultimately, we will create an ordinal model in which the government of South Africa may more efficiently choose which roads to prioritize for government funding (on a 0 to 15 scale).

BACKGROUND & PROBLEM METHODS [continued]

Ross, D. & Townshend, M (2014). Can South Africa road authorities satisfy constitutionally protected basic access needs without sacrificing economic growth? http://www.academia.edu/19750954/Can_South_African_road_authorities_satisfy_constitutionally_protected_basic_access_needs_without_sacrificing_economic_growth

RESULTS [continued]

RESULTS

REFERENCES

DATA

We obtained our data from the following sources:• For data on schools, we used data from an educational

survey conducted by the Department of Education (DOE)• http://www.education.gov.za

• For data on healthcare facilities, we used data from 2011-2012 South Africa Hospital Survey (SAHS) • https://africaopendata.org/dataset/south-african-hospitals-survey-

2011-2012• For data on employment, population, and other

demographic variables, we used data from the Council for Scientific and Industrial Research (CSIR) • http://www.csir.co.za

Weight Schools Healthcare Facilities

Employment

5 1 min 5 min 11898+ jobs

4 3 min 10 min 1189-11897 jobs

3 5 min 15 min 120-1188 jobs

2 10 min 30 min 13-119 jobs

1 15 min 45 min 2-12 jobs

0 15+ min 45+ min 0-1 jobs

METHODSWe utilizing three modeling techniques to create our accessibility-based RMP system for the entire country of South Africa:

1. Geocoding While some of the data included geographic coordinates, for some of the facilities (251 of the hospitals and 2,0382 of the schools), we only had addresses. We created an Address Locator, and with the road network of South Africa (from CSIR), we matched all but 31 of the hospitals (98.99% match rate) and 789 of the schools (96.9% match rate). For employment, we spatially joined the census data on number of jobs in each census block to the centroid of each block. The geocoding results are below:

Layer / Type Total Number Matched Match Rate

Schools 25,720 24,931 96.9%

Healthcare facilities 3,077 3,046 98.99%

Employment Over 9.5 million in 25,000 census tracts

Over 9.5 million in all tracts

100%

0 0.45 0.90.225Decimal Degrees

4

Legend

Namakwa

Employment# of Jobs

012 - 67 - 3334 - 180181 - 963964 - 5163

0 0.45 0.90.225Decimal Degrees

4

Legend1351015NamakwaRoads

0 0.45 0.90.225Decimal Degrees

4

LegendNamakwaRoads510153045

0 0.45 0.90.225Decimal Degrees

4

Legend

Composite AnalysisTotal

< -0.50 Std. Dev.-0.50 - 0.50 Std. Dev.0.50 - 1.5 Std. Dev.1.5 - 2.5 Std. Dev. > 2.5 Std. Dev.Namakwa

LIMITATIONS