Distribution
The distribution in ULTImodel is performed with distribution.DistributeTraffic.
Target values
Using the class TargetValues, the total travel volume is split into the four sectors
national long- and short distance and international OD and transit.
Input: country DataFrame with attributes for classification and totals.
import distribution.DistributeTraffic as DistributeTraffic
import pandas as pd
country_layer = pd.read_csv('country_layer.csv')
country_layer.head()
>>> output
cntr_code index border_share border_length border_length_shared border_crossings_count neighbors population area car_mio_pkm freight_mio_tkm
1 DK 0 0.015467 4088.436575 63.236291 5.0 2 8.396369e+06 43653.093630 60000 15000
2 FI 0 0.269796 8715.390137 2351.374349 136.0 3 9.547163e+06 337512.028767 65000 17000
3 NO 0 0.101108 24434.055949 2470.468692 176.0 3 8.527085e+06 322504.893423 65000 20000
5 SE 0 0.213012 9981.249404 2126.124196 128.0 3 1.621385e+07 449736.409751 100000 50000
Initialize with TargetValues and the country_layer
tv = DistributeTraffic.TargetValues(country_layer)
Target values for personal transport in person kilometers traveled for DK
pt=tv.targets_personal_transport(cn='DK', target_col='car_mio_pkm')
pt
>>> output
{'short': 38917497147.80412,
'long': 20929430184.74586,
'int': 146360460.06437054,
'trans': 6712207.385638687}
Target values for freight transport in vehicle kilometers traveled and trips for DK,
short-distance is further subdivided in sectors (LCV, MFT, HFT)
ft=tv.targets_freight_transport(cn='DK', target_col='freight_mio_tkm')
>>> output
{'lcv': 2400000000.0,
'mft': 449999999.99999994,
'hft': 273347783.1152192,
'trips_short': 79150845.0484885,
'long': 461086483.1505175,
'trips_long': 3347303.631950553,
'int': 416644412.4854073,
'trips_int': 3024671.9565818245,
'trans': 93921321.24885575,
'trips_trans': 403289.59421090997}
Gravity model
Long-distance trips are distributed with GravityModel, producing an OD trip matrix.
Input: TAZ GeoDataFrame with attraction attributes, cost matrix np.array, target values
(see TargetValues)
import distribution.DistributeTraffic as DistributeTraffic
import geopandas as gpd
import numpy as np
taz = gpd.GeoDataFrame.from_file('DK-taz-atts.gpkg')
taz.head()
>>> output
nuts_id cntr_code nuts_name geometry population index_nat index_int id length
0 DK011 DK Byen København MULTIPOLYGON (((12.50479 55.63734, 12.48025 55... 1.199214e+06 0.588597 0.676324 0 19.055466
1 DK012 DK Københavns omegn MULTIPOLYGON (((12.24627 55.70664, 12.25874 55... 8.048375e+05 0.749279 0.860955 1 111.140495
2 DK013 DK Nordsjælland MULTIPOLYGON (((12.24627 55.70664, 12.21994 55... 6.529701e+05 0.665808 0.765043 2 367.978524
3 DK014 DK Bornholm MULTIPOLYGON (((14.88943 55.22813, 14.94373 55... 4.738833e+04 0.091693 0.105359 3 59.449975
4 DK021 DK Østsjælland MULTIPOLYGON (((12.24627 55.70664, 12.16537 55... 3.613609e+05 0.392641 0.451162 4 134.617609
cost_mtx = np.load('cost-matrix.npy')
cost_mtx.shape # (origin, destination, cost), cost: 0=travel time, 1=distance
>>> (69, 69, 2)
pt = {'short': 38917497147.80412,
'long': 20929430184.74586,
'int': 146360460.06437054,
'trans': 6712207.385638687}
ft = {'lcv': 2400000000.0,
'mft': 449999999.99999994,
'hft': 273347783.1152192,
'trips_short': 79150845.0484885,
'long': 461086483.1505175,
'trips_long': 3347303.631950553,
'int': 416644412.4854073,
'trips_int': 3024671.9565818245,
'trans': 93921321.24885575,
'trips_trans': 403289.59421090997}
Initialize with GravityModel
gm = DistributeTraffic.GravityModel(cost_mtx, taz, taz_cn='cntr_code')
Personal transport trip matrix
trips_pt = gm.trip_distribution_pt(target=pt['long'], cn=cn, mob_rate=25, occ_rate=1.3) # cars (personal transport)
trips_pt[:3, :3]
>>> output
Scaling factor personal transport for DK: 0.940780924347826
array([[ 0. , 12769235.98231415, 4371570.57699052],
[14108225.57493652, 0. , 3533537.04995944],
[ 3868113.92794895, 3632011.70280941, 0. ]])
Freight transport trip matrix
trips_ft = gm.trip_distribution_ft(target_vkt=ft['long'], target_trips=ft['trips_long'], cn=cn) # trucks (freight transport)
trips_ft[:3,:3]
>>> output
Freight transport gravity model result for DK: Relation to target vkm 0.7624308514138889
array([[ 0. , 30357.34659726, 25671.69472289],
[30455.28146353, 0. , 33256.92980559],
[25480.89080153, 33418.32286164, 0. ]])
Intrazonal trips
Short-distance trips are assigned to the road network within a cell (single) or within multiple cell within a pre-defined distance (multiple). The VKT are equally distributed using weights for road types.
Input: TAZ GeoDataFrame with attraction attributes, network edges GeoDataFrame, target values
(see TargetValues)
import distribution.DistributeTraffic as DistributeTraffic
import geopandas as gpd
taz = gpd.GeoDataFrame.from_file('DK-taz-atts.gpkg')
taz.head()
>>> output
nuts_id cntr_code nuts_name geometry population index_nat index_int id length
0 DK011 DK Byen København MULTIPOLYGON (((12.50479 55.63734, 12.48025 55... 1.199214e+06 0.588597 0.676324 0 19.055466
1 DK012 DK Københavns omegn MULTIPOLYGON (((12.24627 55.70664, 12.25874 55... 8.048375e+05 0.749279 0.860955 1 111.140495
2 DK013 DK Nordsjælland MULTIPOLYGON (((12.24627 55.70664, 12.21994 55... 6.529701e+05 0.665808 0.765043 2 367.978524
3 DK014 DK Bornholm MULTIPOLYGON (((14.88943 55.22813, 14.94373 55... 4.738833e+04 0.091693 0.105359 3 59.449975
4 DK021 DK Østsjælland MULTIPOLYGON (((12.24627 55.70664, 12.16537 55... 3.613609e+05 0.392641 0.451162 4 134.617609
edges = gpd.GeoDataFrame.from_file('edges.gpkg')
edges.head()
>>> output
highway speed_kph index nuts_id geometry length tt ultimo_id type
0 0 primary 63.0 NaN DK032 LINESTRING (8.75547 55.16922, 8.75502 55.16959... 17.558210 1003.0 1 3
1 0 primary 63.0 NaN DK032 LINESTRING (8.75547 55.16922, 8.75591 55.16884... 7.344446 420.0 2 3
2 0 primary 80.0 NaN DK032 LINESTRING (8.75496 55.31883, 8.75514 55.32270... 2.645051 119.0 3 3
3 0 primary 63.0 NaN DK032 LINESTRING (8.75496 55.31883, 8.75013 55.31339... 17.558210 1003.0 4 3
4 0 primary 63.0 NaN DK032 LINESTRING (8.75496 55.31883, 8.76117 55.32033... 19.593661 1120.0 5 3
pt = {'car': 38917497147.80412}
ft = {'lcv': 2400000000.0,
'mft': 449999999.99999994,
'hft': 273347783.1152192,
'trips_short': 79150845.0484885}
Initialize with IntraZonal
iz = DistributeTraffic.IntraZonal(taz, edges)
Assign VKT from target to network by determining VKT per TAZ and distributing said VKT on
all roads within the TAZ with road_type_weighted_single
net_r, taz_r = iz.road_type_weighted_single(target=pt, veh_types=['car'])
net_r[['ulitmo_id', 'nuts_id', 'car_short']].head()
>>> output
ultimo_id nuts_id car_short
0 1 DK032 2.845761e+06
1 2 DK032 2.845761e+06
2 3 DK032 2.845761e+06
3 4 DK032 2.845761e+06
4 5 DK032 2.845761e+06
taz_r[['nuts_id', 'car_sub']].head()
>>> output
nuts_id car_sub
0 DK011 3.445021e+08
1 DK012 5.091953e+08
2 DK013 6.505540e+08
3 DK014 3.849412e+07
4 DK021 2.271284e+08
Assign VKT from target to network by determining VKT per TAZ and distributing said VKT on
all roads within the TAZ and surrounding TAZ within a pre-defined distance with road_type_weighted_multiple.
It is possible to set a higher weight to the main TAZ during assignment.
The distance between TAZ is based on the distance matrix. Furthermore, the area of each TAZ in km² is calculated so that smaller TAZ are guaranteed to have their surroundings included.
# read distance matrix
distance_mtx = np.load('distance-matrix.npy')
# calculate area
iz.taz.to_crs(epsg=3035, inplace=True)
iz.taz['area'] = iz.taz.area # area per TAZ (smaller TAZ always get surrounding TAZ included, independent from distance)
iz.taz.to_crs(epsg=4326, inplace=True)
# intrazonal assignment
net_r, taz_r = iz.road_type_weighted_multiple(target=pt, matrix_dis=distance_mtx, veh_types=['car'],
weights = pd.DataFrame({'type': [0, 1, 2, 3], 'weight': [0.75, 1.5, 3.5, 3.5]}),
index='population', fac_cell=1.5, distance=50000, occ_rate=1.3)
net_r[['ultimo_id', 'nuts_id', 'car_short']].head()
>>> output
ultimo_id nuts_id car_short
0 1 DK032 2.189047e+06
1 2 DK032 2.189047e+06
2 3 DK032 2.189047e+06
3 4 DK032 2.189047e+06
4 5 DK032 2.189047e+06
taz_r.head()
>>> output
car_sub nuts_id
0 6.505052e+07 DK011
1 3.573749e+08 DK012
2 9.930296e+08 DK013
3 2.961086e+07 DK014
4 3.355154e+08 DK021
