production exchange consumption allocation system modeling ...€¦ · production exchange...
TRANSCRIPT
PECASProduction Exchange Consumption Allocation System
Modeling Review and Update
Mike Alexander, AICP
Dr. JD Hunt, HBA Specto
Jim Skinner, AICP
Wei Wang, GISP
Q1: What (land use) model are COGs/MPOs
already implementing ?
• ARC is in process of implementing a PECAS model
(HBA Specto) for small-area forecast allocation of
control totals derived from a REMI model (with
Technical Advisory Group input).
• PECAS will replace the DRAM/EMPAL model used
since the early 1990s. A custom product
developed in tandem with PB, called the TAZ-
Disaggregator (TAZ-D) is being used as a ―bridge
model‖ while PECAS is under development .
Q2, 5: What stage of implementation is the ARC?
How is PECAS to be used?
• PECAS has two modules: Activity Allocation (AA) and Spatial
Development (SD) . We have run both modules, integrated
and for 20-counties (thus full model) uncalibrated through
time for a three year test period of 2005-2008. Current
status is 2nd and 3rd stage calibration of both modules, and
base data refinement
• The goal for development is, by end of 2010, an integrated
run through plan horizon year of 2040, with calibration of
parameters nearing completion, and best available base
data in place.
• Our modeling goal is to use PECAS as small-area allocation
model in conformity work for next RTP/RDP (est. 2014-15
delivery date), as well as more ongoing scenario work and
TIP modeling
Q3,4,6: What are the major issues of implementation and
how they are being overcome ?
• Data demands- , cost data, equation
parameters
– Building relationships (involving education) to collect the data from
constituents
– Close cooperation of other internal divisions (Land Use, Transportation)
in building impedance and zoning information
– Collaborative post-processing of in-house or public sector data
(Estimates, Census Data)
– Purchase of data from third-parties in private sector (Means, Epermits,
IMPLAN)
• Module calibration
– Automated procedures developed or being developed by consultants
• Output analysis (to inform calibration)
– Custom spreadsheets and charting
– CSV Data Loader
– Web Mapping and Analysis Interface
SD: Parcel Level Data County Parcels
Barrow 28,184
Bartow 42,167
Carroll 50,633
Cherokee 93,866
Clayton 88,723
Cobb 228,690
Coweta 55,348
DeKalb 230,888
Douglas 39,140
Fayette 42,808
Forsyth 77,639
Fulton 341,017
Gwinnett 260,371
Hall 77,103
Henry 72,839
Newton 44,374
Paulding 59,670
Rockdale 34,780
Spalding 29,616
Walton 36,561
Total 1,934,417
– 20-County parcel features
We tried to print it, but the machine kept crashing….
SD: Parcel Level DataWingap format tax data
So which table has the year a structure was built???
Zoning Clean-up Will It Ever End??
UHHH……. NO
AA—Transportation Cost Coefficients
Transport
Coefficients:
κHp,c
amount of
commodity
consumed per
visit
transport
money cost
coefficient
Household Obtained
Services $ / visit utils / $
Summary
calculated
inputs
calculated
inputs
Travel
Segment
Travel
Segment $ / $-util
Commodity Name
Commodity
UnitsSkim Name
Skim
Units
CG10RetailOutput $HBNW utils 0.02708 138.97 -0.5315000
CS13OtherServOutpu
t $HBNW utils 0.02569 146.48 -0.5315000
CS14HealthOutput $HBNW utils 0.00282 1,335.48 -0.5315000
CS15GSEdOutput $HBNW utils 0.01687 223.03 -0.5315000
CS16HiEdOutput $HBNW utils 0.01265 297.41 -0.5315000
AA-Price Estimations
AA-Trip Length CalibrationCommodity BuyingSelling Group
CG01AgMinDirection selling Work-Other
CG02AgMinOutput buying CV Heavy
CG03ConDirection selling Work-Other
CG04ConOutput buying CV Heavy
CG05MfgDirection selling Work-Other
CG06MfgOutput buying CV Med/Light
CS07TCUDirection selling Work-Other
CS08TCUOutput buying CV Med/Light
CS09WsOutput buying CV Med/Light
CS10RetailOutput buying HB Shop
CS11FIREOutput buying Work-Other
CS13OthServOutput buying HB Other
CS14HealthOutput buying HB Other
CS15GSEdOutput buying HB School
CS16HiEdOutput buying HB Univ
CS17GovOutput buying Work-Other
CL23WhiteCollar selling HB Work
CL24Services selling HB Work
CL25Health selling HB Work
CL26Retail selling HB Work
CL27BlueCollar selling HB Work
CL28Military selling HB Work
SD: Rents– Local effect coefficients
– Fulton County -Institutional spaces -Last sold price
exponential I (r2=0.195) INS (r2=0.199) O (r2=0.155) R (r2=0.153) RD (r2=0.194) RH (r2=0.298)
neg exponential B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile)
(Constant)2.44 0.595 4.098 1.08 0.397 2.719 1.403 0.101 13.854 3.525 0.183 19.29 2.775 0.044 63.602 3.009 0.047 63.522
Major Road-0.368 0.126 -2.913 0.5 -0.094 0.171 -0.547 0.5 -0.606 0.095 -6.4 0.5 -0.519 0.084 -6.158 0.5 -0.083 0.006 -14.56 0.5 0.056 0.01 5.58 0.5
Freeway1.504 0.335 4.483 0.25 -0.052 0.117 -0.442 3.00 -0.008 0.008 -1.037 3.00 -0.026 0.015 -1.736 3.00
Freeway Exit-0.984 0.571 -1.724 0.25 -0.525 0.424 -1.238 5 -0.431 0.176 -2.445 0.25 -0.282 0.041 -6.95 0.25 -0.027 0.033 -0.81 0.25
MARTA0.993 0.837 1.187 0.25 -0.428 0.214 -2.003 2 0.713 0.161 4.44 1 0.252 0.128 1.968 0.5 -0.116 0.011 -10.44 1 0.056 0.013 4.269 1
School0.045 0.013 3.454 0.25 0.063 0.025 2.532 0.25
Greenspace-0.788 0.236 -3.334 1.00 0.581 0.113 5.14 1.00 0.04 0.007 5.835 0.25 0.033 0.01 3.428 0.25
eAge0 0.003 -0.208 -0.003 0.002 -1.596 -0.005 0.001 -4.593 -0.001 0 -12.31 -0.005 0 -42.01
luz11.243 0.183 6.782 0.824 0.297 2.769 0.339 0.076 4.449 0.004 0.114 0.038 -0.716 0.014 -51.74
luz20.678 0.334 2.03 0.36 0.299 1.206 -0.161 0.122 -1.318 -0.012 0.007 -1.774 -0.665 0.015 -45.57
luz31.053 0.455 2.315 1.037 0.231 4.493 -0.571 0.089 -6.43 0.013 0.128 0.101 0.291 0.009 31.637 -0.589 0.01 -56.6
luz40.024 0.312 0.076 0.967 0.442 2.188 -0.667 0.109 -6.105 0.229 0.11 2.092 0.676 0.01 67.374 -0.122 0.009 -13.14
luz50.461 0.19 2.432 -0.002 0.144 -0.014 -0.735 0.104 -7.074 -0.301 0.009 -34.85 -0.792 0.013 -62.94
luz60.901 0.142 6.332 0.957 0.226 4.234 -0.494 0.155 -3.195 0.5 0.011 44.94
luz7-0.457 0.129 -3.549 0.013 0.231 0.055 -0.248 0.277 -0.898 -0.607 0.095 -6.376 -0.254 0.008 -30.52 -0.916 0.014 -65.66
luz80.299 0.358 0.836 1.278 0.535 2.388 -1.342 0.259 -5.174 -0.55 0.132 -4.162 0.862 0.326 2.646 0.004 0.019 0.212
luz9-0.27 0.126 -2.137 -0.406 0.364 -1.113 -0.76 0.1 -7.614 -0.234 0.009 -27.11 -0.84 0.014 -60.74
luz10-0.369 0.137 -2.69 0.186 0.453 0.41 0.209 0.748 0.28 -0.592 0.217 -2.732 -0.36 0.01 -37.84 -1.051 0.039 -26.74
luz11-0.518 0.159 -3.246 0.753 0.195 3.864 -0.957 0.274 -3.496 -0.588 0.097 -6.094 -0.297 0.009 -31.83 -0.902 0.018 -49.23
luz12-0.162 0.187 -0.866 -0.395 0.316 -1.252 0.017 0.241 0.07 -0.748 0.119 -6.295 -0.451 0.008 -57.56 -1.177 0.016 -72.29
SaleYear00-1.098 0.148 -7.419 -0.748 0.219 -3.421 0.858 0.107 8.042 -0.881 0.093 -9.488 -0.446 0.007 -64.27 -0.211 0.01 -20.2
SaleYear01-0.752 0.155 -4.862 -0.593 0.224 -2.649 0.803 0.111 7.238 -0.82 0.092 -8.89 -0.392 0.007 -55.54 -0.074 0.011 -6.908
SaleYear02-0.632 0.149 -4.227 -0.222 0.231 -0.962 0.771 0.12 6.423 -0.599 0.093 -6.413 -0.325 0.007 -47.23 -0.112 0.011 -10.33
SaleYear03-0.248 0.152 -1.64 -0.115 0.199 -0.577 1.035 0.114 9.077 -0.347 0.087 -3.972 -0.276 0.007 -40.97 -0.122 0.011 -11.51
SaleYear04-0.844 0.139 -6.07 -0.138 0.207 -0.667 0.802 0.097 8.301 -0.324 0.084 -3.855 -0.134 0.006 -20.94 -0.069 0.01 -6.663
SaleYear050.054 0.137 0.397 0.333 0.222 1.498 0.793 0.091 8.744 -0.108 0.085 -1.275 -0.041 0.006 -6.716 0.056 0.01 5.734
SaleYear06
SaleYear070.054 0.15 0.357 0.356 0.201 1.767 1.122 0.099 11.36 0.207 0.082 2.517 -0.074 0.006 -11.74 -0.023 0.01 -2.227
SaleYear08-0.24 0.376 -0.638 -0.134 0.359 -0.373
SD: New SD Database
– Migrated SD databases to PostgreSQL
– New database schema
– 20-County SD database are loaded– 2 million records in the parcel table
– 10 million records in the local effect table
– More pseudo parcels
– AND these numbers are for one year of data….
Web Interface /Query
SQL Table Views for Output instead of .CSVs
Open Source GIS link
Where It Has Taken Us…Today
–An integrated model run through time, with scenario evaluation •AA> SD> AA, 2005 through 2006
•A scenario of increased transportation costs
•AA running Region wide, but physical development only in Fulton County
– Maps of results to follow
•Sets the stage for further calibration (Stage 3)
–Capability for Full 20-county Run
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Comparison Labor Types by Travel Distance
White Collar Blue Collar
Travel Distance in Miles Travel Distance in Miles
Total Value of Labor
Scenario Tests-SD Results
Forecast Area (Superdistricts =building blocks)
END OF PRESENTATION
The LU Allocation Model (PECAS)
– PECAS (Production, Exchange and Consumption Allocation System)
•Developed by Drs Doug Hunt and John Abraham of University of Calgary
•Based on sound economic theory, incorporating I-O modeling approach; achieves equilibrium
•Two Modules, run Sequentially and Annually –
– Activity Allocation (AA) Module: equilibrium exchange and consumption prices are established by larger zone (LUZ)
– Space Development (SD) Module: based on pricing (rents) from AA and development costs, rational ―developer‖ makes decision or non-decision to develop space in given smaller zones (TAZ) until the market ‗clears‘
•Work Reviewed by the REMI/PECAS Technical Advisory Group (TAG)
PECAS:
Activity Allocation (AA)
$
$ $$$
$ $
$ $$ $
$$$ $ $
$ $ $
Pro
du
cin
gS
ec
tors
Goods, Services, Labour and SpaceC
on
su
min
g S
ec
tors
$$ $ $
$
$
$ $
$$ $
$
$$
$
$
Eco
no
mic
Flo
ws
AA Concept
Activity Allocation (AA)
joint discrete utility
p
nnneelpe
Nn
lpelepnpnlpp
a
l
a
l
a VsVVU
121
ppNpnppp ,...,,..., 21
ppNpnppp ccccc ,...,,...,, 21Utility of exchanging and
shipping one unit of
Commodity between l and e
Quantity of commodity
produced or consumed under
production option p
Additional utility
associated with
production option p
Additional utility
associated with
location l for activity a
Stochastic
error terms
PECAS:
Space Development (SD)
more
the sa
me
no c
hange
mid
density
resid
entia
l
com
merc
ial
industria
l
dere
lict
quantity
zoning dictates set
of alternatives
parcel-by-parcel
microsimulation
Space Development:
Simulation of Transitions
SD Module: joint discrete-continuous utility
qshjphjphjp lllTrjTRU
Rent less
amortized
construction
cost per unit
space
Additional Rent
less
development
costs per unit
land
Stochastic
error
terms
Space quantity
(building size) Land
quantity
(parcel size)
Nested logit structure
No change Demolish DerelictAdd spaceNew space type
QuantityQuantity
multi-level nested
discrete-continuous
logit
Design Diagram-Atlanta (Sorry about size)
CO
MM
OD
ITIE
S
CG
01A
gM
inD
irectio
n
CG
02A
gM
inO
utp
ut
CG
03C
onD
irectio
n
CG
04C
onO
utp
ut
CG
05M
fgD
irectio
n
CG
06M
fgO
utp
ut
CS
07TC
UD
irectio
n
CS
08TC
UO
utp
ut
CS
09W
sO
utp
ut
CS
10R
eta
ilOutp
ut
CS
11F
IRE
Outp
ut
CS
13O
thS
erv
Outp
ut
CS
14H
ealth
Outp
ut
CS
15G
SE
dO
utp
ut
CS
16H
iEdO
utp
ut
CS
17G
ovO
utp
ut
CF
18TaxR
eceip
ts
CF
19G
ovS
up
Receip
ts
CF
20In
vestR
eceip
ts
CF
21R
etu
rnIn
vestR
eceip
ts
CF
22C
apita
lTra
nsfe
rReceip
ts
CL23W
hite
Colla
r
CL24S
erv
ices
CL25H
ealth
CL26R
eta
il
CL27B
lueC
olla
r
CL28M
ilita
ry
SP
AC
E
agri
cultu
ral a
nd m
inin
g s
pace
industr
ial
reta
il
off
ice
instit
utio
nal
mili
tary
space
deta
ched r
esid
entia
l
hig
her
density
resid
entia
l
PRODUCING ACTIVITIES LAND / PARCEL ZONING
AI01AgMinMan f d agricultural or mining zoned land
AI02AgMinProd f very low density residential zoned land
AI03ConMan f d low density residential zoned land
AI04ConProd f d medium density residential zoned land
AI05MfgMan f high density residential zoned land
AI06MfgProd f high-rise residential
AI07TCUMan f a a mixed-use high rise residential
AI08TCUProd f d d majority use mixed use commercial
AI09Whole f d school zoned land
AI10Retail f institutional land
AI11FIRE f passive park zoned land
AI12PTSci active park zoned land (ballparks)
AI13ManServ f manufacturing (heavy) industrial zoned land
AI14PBSOff a d other (light) industrial zoned land
AI15PBSRet d office park industrial zoned land
AI16PSInd d office zoned land
AI17Religion d d low density commercial retail zoned land (office in retail)
AI18BSOnsite d d high density commercial retail zoned land (office in retail)
AI19PSOnsite d a transportation communications utilities zoned land
AI20FedGov f d military base land
AI21StLocGov f
AI22Military
AI23GSEdu f
AI24HiEdu f
AI25Health f
AA26FedGovAccounts
AA27StLocGovAccounts
AA28CapitalAccounts
AH29HHlt20_12 e e e e e e
AH30HHlt20_3p e e e e e e
AH31HH2050_12 e e e e e e
AH32HH2050_3p e e e e e e
AH33HH50100_12 e e e e e e
AH34HH50100_3p e e e e e e
AH35HHge100_12 e e e e e e
AH36HHge100_3p e e e e e e
IMPORTS x x x x x x x x x x x x x x x x x x x x x x x x x x x
EXCHANGE LOCATIONS c p c p c p c p p p p p p p p p s s s s s c c c c c c
EXPORTS x x x x x x x x x x x x x x x x x x x x x x x
CONSUMING ACTIVITIES LEGEND
AI01AgMinMan f f f f f f e f = fixed make or use technical coefficient
AI02AgMinProd f f f f f f e e= exchange
AI03ConMan f f f f f f e x = export or import available
AI04ConProd f f f f f f c = exchange occurs at consumption zone
AI05MfgMan f f f f f f e p = exchange occurs at production zone
AI06MfgProd f f f f f f e d = space development possible
AI07TCUMan f f f f f f e g = time and distance for transport disutilities
AI08TCUProd f f f f f f e h = accessibility or related log sum for transport disutilities
AI09Whole f f f f f f e d = space development permitted
AI10Retail f f f f f f f f f f e a = space development allowed with special application
AI11FIRE f f f f f f e e at higher costs
AI12PTSci f f f f f f e
AI13ManServ f f f f f f e
AI14PBSOff f f f f f f e
AI15PBSRet f f f f f f e
AI16PSInd f f f f f f e
AI17Religion f f f f f f e
AI18BSOnsite f f f f f f
AI19PSOnsite f f f f f f
AI20FedGov f f f f f f e e
AI21StLocGov f f f f f f e e
AI22Military f f f f f f e
AI23GSEdu f f f f f f e
AI24HiEdu f f f f f f e
AI25Health f f f f f f e e
AA26FedGovAccounts f f f f f f
AA27StLocGovAccounts f f f f f f
AA28CapitalAccounts f f f f f f
AH29HHlt20_12 f e e
AH30HHlt20_3p f e e
AH31HH2050_12
AH32HH2050_3p
AH33HH50100_12
AH34HH50100_3p
AH35HHge100_12 f e e
AH36HHge100_3p f e e
TRANSPORT FLOWS
home-based work g g g g g g
home-based other g g g g g g
non-home-based g g g g g g g
light truck g g g g
medium truck g g g g
heavy truck g g g g
Figure 1: Diagram Summarizing the Structure of the Atlanta PECAS Model
Generalized Development Process
•Gather behavioral data
•Establish targets
•3-stage calibration
•Agile development
task 1 SD Development
01: Establish parcel-level space rent equations
02: Establish space transition costs calculation system
03: Establish space maintenance costs equations
04: Establish base-year parcel database
05: Establish base-year quantities of space by type in each zone
06: Establish all-year parcel-level inputs for calibration period
07: Identify space transition constants
task 2 Transport Utility Equation Establishment
08: Establish transport utility equations
task 3 Technology Option Point Development
09: Identify household technology option points:
10: Identify industrial technology option points:
11: Identify accounts categories technology option points:
task 4 AA Target Development
12: Develop labor production zonal-level targets:
13: Develop labor consumption zonal-level targets:
14: Develop labor spatial flow targets:
15: Develop commodity production zonal-level targets:
16: Develop commodity consumption zonal-level targets:
17: Develop commodity spatial flow targets:
18: Develop import and exports targets:
task 5 Import Export Calibration
19: Establish imports and exports functions equation parameters:
task 6 AA Inputs Development
20: Develop skim matrices from transport model:
21: Develop X-Vector Attribute values
task 7 Stage 2 Calibration
22: Establish buying and selling utility equation parameters
23: Establish size terms for imports and exports in external zones
24: Establish technology allocation utility equation parameters
25: Establish location allocation utility equation sensitivity parameters
26: Establish location allocation utility equation zonal constants
task 8 Stage 3 Calibration
27: Set-up semi-automated calibration process
PECAS General Task List
2008 Progress: AA Summary
– IMPLAN data = core of input data, along with PUMS and baseline employment data (NAICS)
– Initial integration with transport model achieved (skims)
–Unconstrained module (20-co) run in June ‗08, LUZ level
– Initial constrained module run in December 2008
IMPLAN Data
PUMS Data: Occ by Income
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
120.00%
lt35 35-75 75plus
Retail
Blue Collar
White Collar
PUMS Data: Occ by Industry
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
120.00%
Education Gvmt/Military Industrial/Manuf. Medical Natl resource Office/Prof. Retail
Retail
Blue Collar
White Collar
Travel Demand Model Skims
2008 Progress: SD Summary
– Parcel spatial data and assessor data = core of input data
– Module (Fulton County) run in May 08, parcel level, with zoning
– Inconsistency between three data sources (assessor space, NAICS employment, LandPro acreage)>> reevaluation
– Used LandPro and NAICS employment to develop TAZ level targets for space (key: consistency across 20 counties)
– Then, disaggregated to parcel using Abraham Floorspace Synthesizer
– Initial synthesizer runs for 15 of the 20 counties
2009 Progress
– Focus of work: June-December 2009
– Model Integration
• AA>SD>AA RUN through time
• Transportation Model Output Incorporated and Used.
– Data Development
– Database Development
– 2nd stage Calibration (size terms, dispersion parameters, transition constants)
– Continued Consultant & Client Collaboration
• Gotomeetings semi-weekly
• Three workshops (March, September, November 2009)
• Gotomypc logins – Allow for maintenance of Directory Structure
– Facilitate coding
AA-Trip Length CalibrationCommodity BuyingSelling Group
CG01AgMinDirection selling Work-Other
CG02AgMinOutput buying CV Heavy
CG03ConDirection selling Work-Other
CG04ConOutput buying CV Heavy
CG05MfgDirection selling Work-Other
CG06MfgOutput buying CV Med/Light
CS07TCUDirection selling Work-Other
CS08TCUOutput buying CV Med/Light
CS09WsOutput buying CV Med/Light
CS10RetailOutput buying HB Shop
CS11FIREOutput buying Work-Other
CS13OthServOutput buying HB Other
CS14HealthOutput buying HB Other
CS15GSEdOutput buying HB School
CS16HiEdOutput buying HB Univ
CS17GovOutput buying Work-Other
CL23WhiteCollar selling HB Work
CL24Services selling HB Work
CL25Health selling HB Work
CL26Retail selling HB Work
CL27BlueCollar selling HB Work
CL28Military selling HB Work
AA—Transportation Cost Coefficients
Transport
Coefficients:
κHp,c
amount of
commodity
consumed per
visit
transport
money cost
coefficient
Household Obtained
Services $ / visit utils / $
Summary
calculated
inputs
calculated
inputs
Travel
Segment
Travel
Segment $ / $-util
Commodity Name
Commodity
UnitsSkim Name
Skim
Units
CG10RetailOutput $HBNW utils 0.02708 138.97 -0.5315000
CS13OtherServOutpu
t $HBNW utils 0.02569 146.48 -0.5315000
CS14HealthOutput $HBNW utils 0.00282 1,335.48 -0.5315000
CS15GSEdOutput $HBNW utils 0.01687 223.03 -0.5315000
CS16HiEdOutput $HBNW utils 0.01265 297.41 -0.5315000
AA-Price Estimations
SD: Space Development Module
– Disaggregate process to the parcel level
– Represent developers‘ action
– Connection with AA
– From AA: current year space price at LUZ level
– To AA: quantity of the spaces for next year AA
– Output
– Space is a commodity consumed by the activities in the AA model
– 8 PECAS space types (A/D/S/M/O/R/L/H)
– Rents are space prices
– Zoning rules
SD: Development Events
– Year-by-year step
– Possible development events– E0: no change
– En: new space type and quantity
– Er: alter or renovate
– Ed: derelict
– Two step process for each parcel– Selection of development events and update space type
– Update space amount
– Data needs– Epermits
– Parcel level data
– Rents
SD: Parcel Level Data
– For each parcel:– Area of the parcel
– Existing space type
– Existing space quantity (building floorspace)
– Structure year
– Zoning rules (allowable uses and density range)
– Cost and fees (associated with development of each permitted space type and quantity)
– Challenges (20 Counties: every dataset is different)
– Parcel features and ID
– Parcel attributes (building floorspace, space type…)
– Geocoded points for Clayton…
– Combine parcel with tax assessors’ data
– Updates
– 20-county parcels are cleaned and loaded– About 2 million parcels are cleaned
– Benefit other planning projects
SD: Parcel Level DataWingap format tax data
So which table has the year a structure was built???
SD: Parcel Level Data County Parcels
Barrow 28,184
Bartow 42,167
Carroll 50,633
Cherokee 93,866
Clayton 88,723
Cobb 228,690
Coweta 55,348
DeKalb 230,888
Douglas 39,140
Fayette 42,808
Forsyth 77,639
Fulton 341,017
Gwinnett 260,371
Hall 77,103
Henry 72,839
Newton 44,374
Paulding 59,670
Rockdale 34,780
Spalding 29,616
Walton 36,561
Total 1,934,417
– 20-County parcel features
We tried to print it, but the machine kept crashing….
SD: Derived Space
– Why do we need the derived space?– The quality of the parcel space data: very inconsistent
– The (in) consistency between employment and space
– Mixed use issues
– Derived using NAICS employment and Landpro
– New space totals at LUZ and disaggregate to TAZ
– Then, evaluation…
SD: Synthesizing to Parcels
– FloorSpace Synthesizer Tool
– Based on existing space type, quantity and zoning…
– Calibration
SD: Synthesizing to Parcels
– Calibration tasks:
– Evolution of initial synthesized results
– Directing synthesized development to actual built-on parcels
– Directing the correct space type to the parcel
– Directing the synthesized built space to developed parcel in amount resembling actual quantities
SD: Synthesizing to Parcels
–Synthesized results
SD: Rents
– Space prices are rents for the use of space
– Per unit of space per unit of time
– Rent equation:– Space price at LUZ level in AA (done by AA & SD integration)
– Local-level effects due to the density of development around the parcel
– Age of the structure
– Local Effects: distance from (or proximity to) local-level influences
– Expressway
– Interstate exit
– Major road
– School
– Marta
– Green space
SD: Rents– Local effect coefficients
– Fulton County -Institutional spaces -Last sold price
exponential I (r2=0.195) INS (r2=0.199) O (r2=0.155) R (r2=0.153) RD (r2=0.194) RH (r2=0.298)
neg exponential B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile) B
Std.
Error t-ratio
Max
(mile)
(Constant)2.44 0.595 4.098 1.08 0.397 2.719 1.403 0.101 13.854 3.525 0.183 19.29 2.775 0.044 63.602 3.009 0.047 63.522
Major Road-0.368 0.126 -2.913 0.5 -0.094 0.171 -0.547 0.5 -0.606 0.095 -6.4 0.5 -0.519 0.084 -6.158 0.5 -0.083 0.006 -14.56 0.5 0.056 0.01 5.58 0.5
Freeway1.504 0.335 4.483 0.25 -0.052 0.117 -0.442 3.00 -0.008 0.008 -1.037 3.00 -0.026 0.015 -1.736 3.00
Freeway Exit-0.984 0.571 -1.724 0.25 -0.525 0.424 -1.238 5 -0.431 0.176 -2.445 0.25 -0.282 0.041 -6.95 0.25 -0.027 0.033 -0.81 0.25
MARTA0.993 0.837 1.187 0.25 -0.428 0.214 -2.003 2 0.713 0.161 4.44 1 0.252 0.128 1.968 0.5 -0.116 0.011 -10.44 1 0.056 0.013 4.269 1
School0.045 0.013 3.454 0.25 0.063 0.025 2.532 0.25
GreenSpace-0.788 0.236 -3.334 1.00 0.581 0.113 5.14 1.00 0.04 0.007 5.835 0.25 0.033 0.01 3.428 0.25
eAge0 0.003 -0.208 -0.003 0.002 -1.596 -0.005 0.001 -4.593 -0.001 0 -12.31 -0.005 0 -42.01
luz11.243 0.183 6.782 0.824 0.297 2.769 0.339 0.076 4.449 0.004 0.114 0.038 -0.716 0.014 -51.74
luz20.678 0.334 2.03 0.36 0.299 1.206 -0.161 0.122 -1.318 -0.012 0.007 -1.774 -0.665 0.015 -45.57
luz31.053 0.455 2.315 1.037 0.231 4.493 -0.571 0.089 -6.43 0.013 0.128 0.101 0.291 0.009 31.637 -0.589 0.01 -56.6
luz40.024 0.312 0.076 0.967 0.442 2.188 -0.667 0.109 -6.105 0.229 0.11 2.092 0.676 0.01 67.374 -0.122 0.009 -13.14
luz50.461 0.19 2.432 -0.002 0.144 -0.014 -0.735 0.104 -7.074 -0.301 0.009 -34.85 -0.792 0.013 -62.94
luz60.901 0.142 6.332 0.957 0.226 4.234 -0.494 0.155 -3.195 0.5 0.011 44.94
luz7-0.457 0.129 -3.549 0.013 0.231 0.055 -0.248 0.277 -0.898 -0.607 0.095 -6.376 -0.254 0.008 -30.52 -0.916 0.014 -65.66
luz80.299 0.358 0.836 1.278 0.535 2.388 -1.342 0.259 -5.174 -0.55 0.132 -4.162 0.862 0.326 2.646 0.004 0.019 0.212
luz9-0.27 0.126 -2.137 -0.406 0.364 -1.113 -0.76 0.1 -7.614 -0.234 0.009 -27.11 -0.84 0.014 -60.74
luz10-0.369 0.137 -2.69 0.186 0.453 0.41 0.209 0.748 0.28 -0.592 0.217 -2.732 -0.36 0.01 -37.84 -1.051 0.039 -26.74
luz11-0.518 0.159 -3.246 0.753 0.195 3.864 -0.957 0.274 -3.496 -0.588 0.097 -6.094 -0.297 0.009 -31.83 -0.902 0.018 -49.23
luz12-0.162 0.187 -0.866 -0.395 0.316 -1.252 0.017 0.241 0.07 -0.748 0.119 -6.295 -0.451 0.008 -57.56 -1.177 0.016 -72.29
SaleYear00-1.098 0.148 -7.419 -0.748 0.219 -3.421 0.858 0.107 8.042 -0.881 0.093 -9.488 -0.446 0.007 -64.27 -0.211 0.01 -20.2
SaleYear01-0.752 0.155 -4.862 -0.593 0.224 -2.649 0.803 0.111 7.238 -0.82 0.092 -8.89 -0.392 0.007 -55.54 -0.074 0.011 -6.908
SaleYear02-0.632 0.149 -4.227 -0.222 0.231 -0.962 0.771 0.12 6.423 -0.599 0.093 -6.413 -0.325 0.007 -47.23 -0.112 0.011 -10.33
SaleYear03-0.248 0.152 -1.64 -0.115 0.199 -0.577 1.035 0.114 9.077 -0.347 0.087 -3.972 -0.276 0.007 -40.97 -0.122 0.011 -11.51
SaleYear04-0.844 0.139 -6.07 -0.138 0.207 -0.667 0.802 0.097 8.301 -0.324 0.084 -3.855 -0.134 0.006 -20.94 -0.069 0.01 -6.663
SaleYear050.054 0.137 0.397 0.333 0.222 1.498 0.793 0.091 8.744 -0.108 0.085 -1.275 -0.041 0.006 -6.716 0.056 0.01 5.734
SaleYear06
SaleYear070.054 0.15 0.357 0.356 0.201 1.767 1.122 0.099 11.36 0.207 0.082 2.517 -0.074 0.006 -11.74 -0.023 0.01 -2.227
SaleYear08-0.24 0.376 -0.638 -0.134 0.359 -0.373
SD: New SD Database
– Migrated SD databases to PostgreSQL
– New database schema
– 20-County SD database are loaded– 2 million records in the parcel table
– 10 million records in the local effect table
– More pseudo parcels
– AND these numbers are for one year of data….
Where It Has Taken Us…Today
–An integrated model run through time, with scenario evaluation •AA> SD> AA, 2005 through 2006
•A scenario of increased transportation costs
•AA running Region wide, but physical development only in Fulton County
– Maps of results to follow
•Sets the stage for further calibration (Stage 3)
–Capability for Full 20-county Run
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Scenario Tests-AA Results
Labor Travel ChangeWhite Collar Travel Distance by Total Value of Labor
Travel Distance in Miles
Total Value of Labor
Labor Travel ChangeBlue Collar Travel Distance by Total Value of Labor
Travel Distance in Miles
Total Value of Labor
Comparison
White Collar Blue Collar
Travel Distance in Miles Travel Distance in Miles
Total Value of Labor
Scenario Tests-SD Results
PECAS Task List--Progress
Task group 1 SD Development
01: Establish parcel-level space rent equations
02: Establish space transition costs calculation system
03: Establish space maintenance costs equations
04: Maintain base-year parcel database
05: Establish base-year quantities of space by type in each zone
06: Establish all-year parcel-level inputs for calibration period
07: Identify space transition constants
Task group 2 Transport Utility Equation Establishment
08: Establish transport utility equations
Task 3 Technology Option Point Development
09: Identify household technology option points:
10: Identify industrial technology option points:
11: Identify accounts categories technology option points:
Black= Annual Update Grey= Completed Pre-09 Blue= Completed in 2009 Green = In process/continual
PECAS Task List--ProgressBlack= Annual Update Grey= Completed pre-09 Blue= Complete in 2009 Green = In Process
Task group 4 AA Target Development
12: Develop labor production zonal-level targets:
13: Develop labor consumption zonal-level targets:
14: Develop labor spatial flow targets (CTPP)
15: Develop commodity production zonal-level targets:
16: Develop commodity consumption zonal-level targets:
17: Develop commodity spatial flow targets: (CFS)
18: Develop import and exports targets: (NEEDS REFINING)
Task group 5 Import Export Calibration
19: Establish imports and exports functions equation parameters:
Task group 6 AA Inputs Development
20: Develop skim matrices from transport model: (NEEDS INTG THR TIME)
21: Develop X-Vector Attribute values (e.g. schools, air quality)
Task group 7 Stage 2&3 Calibration
22: Establish buying and selling utility equation parameters
23: Establish size terms for imports and exports in external zones
24: Establish technology allocation utility equation parameters
25: Establish location allocation utility equation sensitivity parameters
26: Establish location allocation utility equation zonal constants
Task group 8 Stage 3 Calibration
27: Set-up semi-automated calibration process
PECAS Tasks-Document
What‘s Next?: 2010+ PECAS Work Plan
• Zoning Data Development
• Evaluation of Model Output over a thirty-
year planning horizon
– Sensitivity testing for the parameters developed
in Stage II (during 2009)
– Calibration of floorspace synthesizer, space
transition costs and constants, and dispersion
parameters as needed
• Limited testing with varied TDM networks
2010+ PECAS Work Plan
• Review/Analysis of New or Updated Data to
modify parameters as needed
– IMPLAN, CoStar, Means, ACS PUMS, Other
• Directory Structure Improvements/Rsync
– Refinement of coded scripts >> enhanced
integration of the AA and SD modules through
time (year by year), full integration of the ARC
transport model (ultimately, tour-based)
– GUI Interface work (possibly)/ GIS linkages
On the Shoulders of\In Tandem with…
• Portland and Oregon
• Baltimore
• Montgomery
• California
– San Diego (SANDAG)
– LA (SCAG)
– SF (ABAG)
– Statewide (CALTRANS)
• Others?
The Goals of 2010+ Work?…
• Use PECAS directly as ARC‘s small area
allocation model in the NEXT conformity
modeling process (next RTP/RDP)
• Use PECAS scenario modeling ability to
shape and inform policy development,
rather than just to test ‗existing‘, already
formulated policy
Current and Future Reference…