Date post: | 22-Dec-2015 |
Category: |
Documents |
Upload: | ralf-mcdaniel |
View: | 218 times |
Download: | 0 times |
Estimating Intrazonal Impedances in
Macroscopic Travel Demand Models
Matthew Bediako OkrahTechnical University of Munich
15th Transportation Planning Applications Conference
20 May 2015
Contents• Background
• Proposed Method
• Application and Suitability Checks
• Significant Observations
2
Background
• Existing approximate methods - Nearest Neighbour Techniques- Functions of Area Size
• Questions- Good estimates of intrazonal impedances?- Good estimates of intrazonal flows?
3
Intrazonal Trips
4
Intrazonal Trips
5
Proposed Method
6
Select Nodes
Calculate Weight Matrix Weighted Distance Matrix
Assign Node Weights
Compute Intrazonal Distance
Calculate Distance Matrix
Proposed Method
• Variants of Method- Unweighted Nodes- Nodes Weighted by Closeness Centrality- Nodes Weighted by Degree Centrality
7
Application of Method
• Dachau with
102 TAZ
• Area= 13 sq.mi.
• Pop= 46,000
8
Suitability Check 1
• Aggregate 102 Zones into:- 25 Zones- 10 Zones- 5 Zones- 2 Zones
• Intrazonal distance with different methods
• Compare estimated intrazonal distances
9
10
25-Zone10-Zone5-Zone2-Zone
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
RM
SE (km
)
UnweightedDegreeClosenessArea-BasedNearest Neighbor
RMSE in Estimated Intrazonal Distances
Suitability Check 2
• Trip Distribution with the different methods - Home-Work Trips - Non-Home-Based Trips- Home-Shopping Trips
• Between modeled and observed trips;- Compare TLFDs (Coincidence Ratio)- Compare Total Number of Intrazonal Trips
11
12
10987654321
25
20
15
10
5
0
Distance (km)
Perc
ent
Observed (ATL = 1.6 km)Estimated (ATL = 1.5 km)
Coincidence Ratio = 0.83
TLFD of NHB Trips (Unweighted)
Estimated Intrazonal Trips = 1215
13
10987654321
25
20
15
10
5
0
Distance (km)
Perc
ent
Observed (ATL = 1.6 km)Estimated (ATL = 1.5 km)
Coincidence Ratio = 0.83
TLFD of NHB Trips (Degree)
Estimated Intrazonal Trips = 1195
14
10987654321
20
15
10
5
0
Distance (km)
Perc
ent
Observed (ATL = 1.6 km)Estimated (ATL = 1.5 km)
Coincidence Ratio = 0.82
TLFD of NHB Trips (Closeness)
Estimated Intrazonal Trips = 1384
15
10987654321
20
15
10
5
0
Distance (km)
Perc
ent
Observed (ATL = 2.6 km)Estimated (ATL = 3.4 km)
Coincidence Ratio = 0.55
TLFD of NHB Trips (Area-Based)
Estimated Intrazonal Trips = 2136
16
10987654321
25
20
15
10
5
0
Distance (km)
Perc
ent
Observed (ATL = 1.6 km)Estimated (ATL = 1.5 km)
Coincidence Ratio = 0.82
TLFD of NHB Trips (Nearest Neighbor)
Estimated Intrazonal Trips = 1548
17
Home-Shopping(700)Non-Home-Based(1280)Home-Work(280)
60
40
20
0
-20
-40
-60
Perc
ent
UnweightedDegreeClosenessArea-BasedNearest Neighbor
Errors in Estimated Number of Intrazonal Trips
Conclusions
• Unweighted provides better estimates of
intrazonal impedances
• Better intrazonal impedance no guarantee for
better estimates of intrazonal flows
• Nearest neighbour method sufficient given
ease of measurement
• Need for smaller zones to avoid intrazonal
trips18
Thanks for your Attention
19