United States
             Environmental Protection
            Office Of The Administrator
November 1990
A Study Of
Freeway Capacity Increases
In The San Francisco Bay Area
And Greater Sacramento
                                       Printed on Recycled Paper

A Study of Freeway Capacity  increases in the San Francisco Bay
                Area and Greater Sacramento Area
                                               Tom Addison
                                               NNEMS Fellow
                                               U.S. EPA,  Region IX
             US Environmental Protection Agency              Air Programs  Branch
             Region 5 Library (PL-12J)                    September 28,  1990
             77 West Jackson Blvd., 12th Floor
             Chicago, IL 60604-3590


This report was furnished to the U.S. Environmental Protection
Agency by  the student identified  on the cover page,  under a National
Network for Environmental  Management Studies  fellowship.

The  contents are essentially  as  received  from  the  author.  The
opinions, findings,  and conclusions  expressed  are  those  of the author
and  not necessarily  those of the  U.S. Environmental  Protection
Agency.  Mention,  if any, of company, process, or product names is
not to be considered as an endorsement by the U.S.  Environmental
Protection  Agency.


This study evaluates the adequacy of freeway capacity increases' air quality
analyses and the accuracy of their predictions of future traffic volumes.  New-
alignment freeways, freeway lane additions, and freeway interchange additions or
expansions in the greater Sacramento and greater San Francisco Bay Areas,
planned in the last 10 years, were all included in the study. The CEQA and/or
NEPA documents for these projects gave the needed air quality analyses and
traffic predictions.
    The study revealed simplistic air quality analyses.  Only 22% of the projects
analyzed mesoscale emissions of CO, NO0 and hydrocarbons; the rest  of the
projects had only microscale analysis of CO emissions or no quantitative emissions
analysis whatsoever.  Despite a shortage of detailed traffic count  data  in the Bay
Area, the study showed a pattern of traffic projections that significantly
underpredicted the observed actual volumes.  Thus, these freeway capacity
increases appear not to have had the predicted regional air quality benefits, but
instead have likely worsened air quality.


This work was supported by a grant from the National Network  for
Environmental Management Studies program of the U.S.
Environmental Protection Agency.  Numerous officials of differenl
governmental organizations as well as academic institutions
contributed to this study by sharing their insights, data, and
time.  I would particularly like to thank Julia Barrow, Mark
Brucker, Jennifer Dill, and Frances Wicher of the U.S. EPA's
Region IX Air Programs Branch, who all found the time and
patience to meet regularly with me and provide invaluable advice.

     Additionally, staff from the following organizations born
allowed me access to the files and information I needed ~or rr.is
study, as well as generously volunteering their time and years c;
experience and insights:
-California Department of Transportation (Caltrans) District  4,
San Francisco. Environmental,  Highway Operations, Transportation
Planning, and Traffic Branches.
-Caltrans District 3, Marysville.  Environmental and Traffic
Census B Branches.
-Caltrans Headquarters, Sacramento.  Offices of Environmental
Analysis and Traffic.
-Federal Highway Administration, Region IX, San Francisco.

     Despite the assistance provided by the above individuals'as
well as dozens elsewhere, any errors in this study are solely
attributable to the author.  Furthermore,  the views and opinions
expressed herein are not those of the U.S.  EPA,  but are rather
those of the author alone.


                      EXECUTIVE SUMMARY

     The purpose of this research is to evaluate, for projects

that increase freeway capacity, the adequacy of their air quality

analyses and the accuracy of their traffic predictions.  The

study examines three types of capacity-increasing measures on

limited-access, divided highways  ("freeways") in the greater San

Francisco Bay Area and greater Sacramento regions:  1)  new-

alignment freeways, 2) expansion of existing freeways by adding

"mixed-flow" and/or high-occupancy vehicle (HOV, or diamond)

lanes, and 3) adding or expanding interchanges on existing

freeways.  Such projects are required under the California

Environmental Quality Act (CEQA)  and/or the National

Environmental Policy Act (NEPA) to undergo environmental review

prior to their construction.

     By limiting the study to all such projects for which the

final environmental document is less than 10 years old (and

adding a few additional projects that were one or two years

older),  there exist 27 such freeway projects.  I looked at the

level of detail of the air quality analysis for all these in Part

I of the study.   In Part II I compared, for those projects on

which construction had been completed,  the projected future

traffic volumes with actual volumes as determined by traffic


     The large majority (63%)  of the 27 capacity-increasing

projects analyzed only carbon monoxide (CO)  air quality in ~ne

immediate freeway vicinity.  22% of the projects had more

detailed air quality analysis, which typically also involved

mesoscale analysis of outputs of GO, hydrocarbons (HC),  and

oxides of nitrogen (NOJ .   15% of the projects had no

quantitative studies of air quality effects.

     The second half of the study,  comparing traffic projections

to observed volumes,  revealed serious shortages of traffic count

information, as well as extremely simplistic projections in the

original environmental documentation.   Because of ~hese and "~er

problems with the data, a precise quantitative assessment of ~he

accuracy of traffic volume forecasting was not possible.

However, in 5 of the 6 cases in this portion of the study,

traffic projections underpredicted the later observed volumes.

     This study does not determine the reason for the

underprediction.  Induced trips or latent demand, unexpected ,

growth, and other factors are all probable partial causes.

Whatever the reasons for repeated underprediction,  the predicted

air quality benefits that were used to justify the projects'

construction may never have materialized.  In fact,  these

projects likely had overall negative impacts on air quality in

the short run.  Their long run effect on air quality, while still

undocumented, may be even more negative,  since travel and

population growth are likely to continue to exceed the forecasts

made in these projects' environmental documents.

     Given the past underprediction of traffic volumes,  future

freeway projects analyzed using standard traffic forecasting

techniques that advertise regional air quality benefits should be

viewed cautiously by air pollution officials.  The environmental

planning process would benefit greatly from improved traffic

forecasting.  Improving traffic count data is a critical step

toward achieving this goal.  Additionally, the air quality

analyses of all these projects should model the consequences of

the freeway expansion not only on the emissions of CO, but on the

emissions of HC, NOX, and PM-10  (small particulates) as well.


     Despite the toughest air pollution regulations in the U.S.,

most Californians are still forced to breathe unhealthy air.   The

total annual costs of this pollution have been estimated by many

economists to be in the billions of dollars.   Countless studies

have documented that this sorry state of affairs is largely a

result of motor vehicle use.

     Increases to freeway capacity have the potential to

significantly affect air quality.   The environmental planning

process under the National Environmental Policy Act (NEPA)  and/cr

the California Environmental Quality Act (CEQA)  is the only pcinn

at which the air quality effects of capacity increases are

examined on an individual project, as opposed to a regional,

basis.  Thus, it is imperative that the air quality analyses in

this planning process be accurate, as well as adequate.

     I undertook a two-part evaluation of the adequacy of the/

environmental planning process for projects that increase freeway

capacity.  First, I evaluated the project's air quality analysis,

and then compared the traffic volumes projected for the improved

freeway to the actual traffic volumes observed after it was

completed.  I present the background information,  the findings,

and a discussion of the findings for each of these two halves of

the study in Parts I and II below.

                              PART I.

A. Background and Methodology

     The analysis included the following three types  of projects

in the greater San Francisco  Bay Area and the greater Sacramento

regions:   1) new freeways; 2) expansions of existing  freeways  by

adding mixed flow (i.e., no restrictions on use) and/or HOV  (use

limited to vehicles with a minimum number of passengers) lanes;

and 3) projects that added new freeway interchanges or expanded

existing interchanges.  I included only projects for  which the

final environmental document  had been completed in the last  10 to

12 years.  Today these projects are in various stages of

development.  Some projects are completed, some are under

construction, and some are still in the planning stage.

     The National Environmental Policy Act (NEPA), and the

California Environmental Quality Act (CEQA),  NEPA's state

corollary, require that the environmental consequences of freeway

expansion projects be evaluated prior to the project's

initiation.  Both laws require similar environmental documents

for these projects.   First,  a federal Environmental Assessment

(EA)  or a state Initial Study (IS)  must be prepared.  If the

environmental consequences of the planned work are deemed

relatively insignificant,  the final federal  document is a Finding

of No Significant Impact (FONSI)  while the final state document

is a Negative Declaration (ND).   If the consequences are deemed

to be more major,  then the federal  and state  documents prepared

are an Environmental  Impact  Statement (EIS)  or Report (EIR),


     Some of the roadwork I looked at was subject to both laws,

whereas other projects were only covered by one.  In the 1950's

and 1960's when the U.S. interstate program was under full

construction, the bulk of the funding for such projeer3 was

federal.  Under the Reagan-era "new Federalism" of the 1980's,

the federal money was very limited.

     For projects that are built without any federal funds, ana

that are not connected ro federal or federal aid highways,  only

CEQA applies.  In the Bay Area,  the number of projeers subjecr

only to CEQA is rising dramatically.  The majority of planned

capacity increases are being financed by local or countywide

initiatives that raise sales taxes for highway dollars.  This

shift in freeway funding is of significance to the EPA, for

virtually all projects subject only to CEQA regulation are not

routinely reviewed by Region IX.  Also, the EIR's prepared

directly by or for county or local governments are increasing.

With Caltrans no longer being the sole preparer, there will

likely be an increasing range in the quality of the EIR's.

     There were 21 projects in the Bay Area and 6 projects in the

greater Sacramento region that met the previously stated

criteria.  These are listed in Table 1.

     In reviewing the environmental documents for each of these

27 projects, the first part of this study looked at the level of

detail of the air quality analysis of each project.  In addition

TABLE 1:  The Title, Type, and Date of the Projects'"  Environmental  Documents

   Abbreviations: I/C = interchange  EA = Environmental Assessment IS = Initial Study  ND =
   Negative Declaration FONSI = Findings of No Significant Impact  DEIS/FEIS = Draft/Final
   environmental impact statement  FEIR = Final Environmental Impact Report

   Greater Bay Area projects:

   1.    EA and  IS for  two new I/C's at Stonendge Drive on J-680 and Hacienda Drive  on 1-580
        in Pleasanton, modify existing I/C on 580, and build auxiliary lanes on 580 and 680, 12/ST.

   2.    EA/IS for Widening from 4 to 6 Lanes and Construction of 2 Sound Barrier Walls on I-
        880 in Alameda and Santa Clara Counties, 6/88.

   3.    DEIS: 1-80 and 1-180. Operational Improvements in Alameda and Contra Costa Counties,

   4.    FEIS and 4ff) Statement.  Route 101 in Santa Clara County, 0.6 miles '.ami: nf C.jcnranc
        Road in  Morgan Hill to 0.7 miles Nonn of Route 
                              TABLE 1 (continued)

      19.  ND/FONSI. Proposed Widening of Route 152 in Santa Clara County, 4/87.

      20.  EA. Proposed Ramp Connection for Route 238 West to Route 17/I-MO South, in and near
          San Leandro, Alameda County, 3/85.                                           i

      21.  ND/FONSI. Ramp, Road, Overcrossing and Signals Construction, Modification, IVidenmq,
          and Installation on Route 237 in Santa Clara County at Fair Oaks Avenue, 2/33.            '

      Greater Sacramento region projects:

      22.  Environmental Reevaluation. Route 99 [rural highway convened to 4-lane freeway], 12/83.
          [FEIS, 8/75].

      23.  FEIS. Roseville Bypass (Route 65], 9/84.

      24.  ND/FONSI. Route 99. Build 2 lanes in the meaian in ^acramcntu OKtr.'.v rcr^ccn Macx
          Road ovcrcrosstng and Sacramento Blvd. ovcrcrossurj, :!i/X7.

      25.  FEIR. Silva Valley Parkway I/C with 1-50, 2/90.

      26.  IS. Laguna Blvd./Route 99 l/C Reconstruction, 2/88.

      27.  FONSL North Natomas Freeway Improvements,  1/90.
to  a final EIS/R  or FONSI/ND, most of the  projects  had  at least

one supplementary technical  report.   These reports  present  the

results  of the air quality models used to  predict the effect of

the increased freeway  capacity  on the emission of pollutants.

Additionally,  these reports  often contained the traffic forecasts

used in  Part  II of this study.

B.  Findings and Discussion

      All of the 27 air quality  analyses in this report  concluded

that the project  would improve  air quality (although not

necessarily by a  significant margin),  or would result in no


significant change in air quality.  For 17 of the  27 projects

(63%), the air quality analysis consisted of a microscale  study

of only CO (carbon monoxide) emissions.  Analyzing the effects of

only  one pollutant often was justified by the inaccurate

conclusion that CO serves as an "indication of the full range of

pollutants"1.   The effects  of a project on  the  full range  of air

pollutants, however, can not be estimated by CO emissions.   In

general, increasing the average travel speed on a  freeway  fro:?, a

congested, stop-and-go condition to a steady flow decreases  tr.e

emissions of both CO and total HC  (hydrocarbons), but increases

the emissions of NOX  (oxides of nitrogen).   Furthermore, the

impacts of CO are localized, but the formation of ozone from HC

and NOX affects the  larger  air  basin.

      Only 6 of the 27 projects  (22%) had both a microscale

analysis of CO emissions and a mesoscale analysis of emissions of
CO, HC, and NOK.   Five of these 6 projects  were  of  enough

significance to require an EIS. Two projects requiring an EIS did

only  a microscale CO analysis.  One of the 5 mesoscale analyses

projected emissions of SO,  and  particulate  matter,  and divided HC

predictions into total and nonmethylated hydrocarbons.  Another

included a prediction of project-caused lead emissions.

      Four of the 27 projects (15%)  had no quantitative analysis

of air quality.  One such project justified this minimalist

approach as follows:  "Based on previous analysis [of different
     1  (EA,  Construct  4-Lane  Freeway from  Mini Dr. to  Sage St.
Overcrossing...  in Vallejo. Solano Countyr  11/85).

projects]... no significant impact... is anticipated"2.   Anotr.er

justification was that because the project was predicted to

decrease total vehicles miles traveled  (VMT) as well as

congestion, the project will benefit air quality, and thus

quantitative analysis is not required3.   Of  these 4  projects with

no quantitative analysis, 3 were interchange projects, while  the

fourth was a freeway-to-freeway ramp project.  Given that both

the Bay Area and Sacramento are violating federal and state

standards for ozone and CO, adequate air quality analyses for all

these projects would have predicted the effect ^n er.icsicr.s  of

CO, and the ozone building blocks HC and NO,.   Additionally,

Sacramento is violating the federal standard for small

particulate matter (PM-10)\   Only  one  of the  27 projects

modeled what its consequences would be for either PM-10 or TSP.

     A shortcoming of these models is that,  for interchange

additions or expansions, they tend not to model the consequences

of additional traffic on the mainline, but simply look at the

emissions of the cars on the ramp.  This is a significant

oversight.  If the mainline is congested or nearly so, these

entering vehicles can bog down the mainline enough so that total

emissions of CO and HC may increase dramatically.  Thus the
     2  (EA/IS, Reconstruction  of the 80/Alamo  Drive Interchange
(I/O in Vacaville. Solano County. 10/86).

     3  (EA/IS,  Two  new  I/C's at Stoneridge Dr. on 680 and Hacienda
Dr. on 580 in Pleasonton.  modify  existing I/C. and build auxiliary
lanes..., 12/87).

     4  PM-10  replaced  total  suspended   particulates  (TSP)  as a
criteria pollutant in 1987.


addition of exrra vehicles to a freeway near saturation nor  only

negatively affects air quality through the emissions from the  new

traffic, but also by increasing emissions from all the vehicles

on the freeway.  Although ramp metering might mitigate this

problem, many interchanges and ramps are still being buil~

without ramp metering, and the air quality analyses are typically

done for a scenario without metering.

     None of the projects gave a qualitative, let alone

quantitative, analysis of the effects of the increased capacity

on carbon dioxide emissions.  This is a serious snorrco-ir.g, fcr

CO2 is  the  primary  "greenhouse gas,"  which wnen  emitted

contributes to the problem of global warming.  Carbon dioxide  is

not directly harmful to individuals, but rather in sufficient

quantities damages our planet's atmosphere.

     While both the Reagan and Bush administrations have shown

great reluctance to limit CO, emissions, these emissions will
have to be reduced.  In California, over half the carbon

emissions are from our transportation network.5  C02 emissions are

tied directly to fuel consumption.  Thus,  a vehicle idling in

stop-and-go,  congested traffic is maximizing CO2 output.

     Another shortcoming concerned mitigation of project-caused

air quality declines.  One air quality analysis stated that  "air

quality will be monitored to determine the need for ramp metering
     8  The Impacts  of Global  Warming  on California.  California
Energy Commission, p. c-1. August,  1989.

to avoid significant air quality impacts."0   No  timetable of -,;nen

or where such monitoring would be done was provided, however.

Nor were critical pollutant levels specified which would trigger

the installation of ramp metering.  Without a specified

implementation plan, it is unclear whether mitigation measures

will be actually carried out.  Therefore, it is questionable

whether the measures should be given much weight when evaluating

the project impacts.  A recent CEQA amendment now requires

monitoring plans for mitigation measures in final CEQA documents,

which may improve the situation.
                             PART II.

A. Background

     A wide number of varying factors, such as average fuel  /

economy, percentage trucks, and pollution control standards, are

used in the air quality analysis for a capacity-increasing

highway project.  The most critical component of these models are

traffic volumes for both the build and no build scenarios. Thus a

critical first step to check the accuracy of the air quality

impact analysis for a project is to compare actual traffic to the

predicted traffic used in the emissions model.  The second major

part of the study did just that and compared the projected
     a  (ND, Widening 101... in Santa Clara County from the Lawrence
Expressway to the San Mateo County Line. 10/86).

traffic volumes to the actual volumes as measured by counts on

completed projects.

     Before presenting the results of these comparisons, I first

discuss problems in the modeling process as well as limitations

to the data used in this part of the study.

     Traffic modeling is a complex process, but its essence is

that land use, growth forecasts, and socioeconomic data are used

to predict numbers, origins, and destinations of future trips.

Various models then assign these trips to nodes ftransit, private

vehicles, carpools, etc.), times, and roadways, using travel

time, expenses, and other factors.  Certainly any attempt to

predict the future is an inexact art, and predicting traffic

volumes is especially tricky, given the enormous range of

variables that potentially affect travel.  These factors include

gas, transit, parking, and toll prices, changes in regional

growth patterns, new travel options, earthguakes, etc.

     Although these models are highly sophisticated, they contain

two serious flaws.  First, they treat land use strictly as an

explanatory variable for highway use.  Although land use

certainly helps explain observed highway use, it is also true

that freeway location and congestion levels influence the land

use decisions.

     Historically, land use decisions in this country have been

made at the local level.  Regional or state planning attempts at

more centralized land use planning are typically controversial

and are often challenged successfully in the courts or through


voter initiatives.  Caltrans and FHWA are thus extremely

reluctant to appear to be making land use decisions.   Officials

at these agencies maintain that roads are expanded to meet rhe

demand provided by present and future land uses,  as decided by

local governments.  Their position is that these projects are

often undertaken to fulfill existing needs for greater freeway

capacity, so any influence on land use is minimal.

     While it may be acknowledged that highways influence land

use through market forces, highway agencies maintain ~har ~ne

responsibility for changes in land use that arise lie with the

local governments who set policy through zoning and other

mechanisms.  An apolitical examination reveals that where and

when roads are expanded or built has the potential for tremendous

impacts on regional growth and development.  Several senior

Caltrans officials agreed off-the-record that this is the case,

but agency policy refuses to acknowledge this.

     This debate, over whether growth causes the highways or

highways cause the growth, is often polarizing and acrimonious,

with Caltrans and environmental groups exchanging heated rhetoric

both in the press and in the courts.  Rarely does either side

cite controlled or quantitative studies to reinforce their

position, although there is a body of literature on this subject.

(See Appendix A).  Certainly the issue does not lend itself

readily to quantitative study.  Difficulties include the

abundance of confounding variables, and the need for controls

combined with the problem of finding areas similar in all aspects


 but  for  freeway  improvements.  After  reviewing the  literature,  it

 appears  that  the effects  of roadwork  on growth are  variable  and

 site specific.   While growth is limited without  an  adequate

 transportation network, such a network is not of itself

 sufficient to ensure that growth will follow.  Other  factors,

 such as  sewage lines and  nonrestrictive zoning,  are also


      The second  theoretical flaw in traffic models  is that the

 models do not include feedback loops  to racord the  effect of

 increased capacity on the public's demand for or use  of the

 facility.  Demand, and thus the trips generated,  are  assumed to

 be the same both before and after the capacity increases are

 completed.  Most models in use today do allow and account for

 diversion of  trips to other routes or modes, but not  for new

 trips.   Thus, the models  show that increased freeway capacity may

 increase traffic volumes, but only by diverting  (capturing)   '

 vehicles from, e.g., local arterials,  or capturing  trips from

 existing transit riders.

      In highly congested  areas such as the urban regions of

 California,  many individuals are either foregoing automobile

 trips altogether or have  altered their destinations and/or travel

 times to avoid peak hour  congestion.   Called "discouraged

drivers," they create a "latent demand"  for improved road

conditions.   These phenomena are not addressed by traffic

projection models currently in use.   When the capacity on a

freeway is increased,  very likely some of these discouraged


drivers will return to their vehicles on the expanded roadway,

leading to new "induced trips."

     Even if transportation agencies were to acknowledge the

existence of pre-construction latent demand and post-construction

induced trips, the new trips would be difficult to quantify.  One

hypothetical way to measure induced trips would be to observe

traffic volumes on the freeway or ramp immediately prior to

construction commencing, and then to record the volumes after

construction.  Any increase in ..traffic could be attributed

directly to induced trips.  This information could then be used

to develop models to predict induced trips.

     The problem with this approach is that other variables

besides induced trips could explain the difference between these

two volumes.  For example, unexpected regional growth including

increased housing and commercial, industrial,  or retail

development would increase traffic, while improved transit or

traffic signal coordination on parallel arterials would decrease


     This problem would be minimal if the planning and

construction periods were very short, and the traffic was counted

immediately before and shortly after construction.  Most of the

freeway capacity increases studied, however, took a year or often

much longer to complete, allowing greater amplitude in the

confounding variables.  Also, as will be discussed in detail

later, traffic counts are done very infrequently in the Bay Area.

     Another problem in quantifying induced trips is in


accurately measuring the types of responses to road expansions:

shifts in mode, travel time, route,  and destination.  Route

shifts, for example from arterials to the improved freeway, prove

difficult to assess since few arterials are regularly counted.

In the Bay Area, few towns other than Berkeley and San Jose have

regular counts of their arterials, although this non-freeway

network is included in the traffic models.   Mode shifts, e.g.,

from vanpocls to single occupant vehicles,  would also confuse the

issue, in this example by having some of the traffic increases

due not to induced trips.

     One theoretically valid way to quantify induced trips is to

conduct interviews with drivers.  Interviews could take place on

the completed roads that have a tollbridge or plaza;

realistically, this seems unlikely because of the time required.

A large cohort study might also yield useful data.  Regional t

travelers could be questioned on their behaviors before and after

the roadwork.  Problems with this approach include the high tine

costs of gathering the data and the possibility of unreliable

data, due to respondents' untruthfulness or forgetfulness.

     The primary rationale for the construction of the vast

majority of the 27 freeway projects examined was decreased

congestion.  The benefits attributed to reduced congestion are

better service to the drivers and improved air quality.  Two

problems underlie this reasoning.  First the assumption that

lessened congestion equals improved air quality is problematic.

Although CO and hydrocarbon emissions are reduced as congestion


decreases, NOX  emissions are  increased.  Ozone, the major  snog

ingredient, is produced by the sunlight-aided reaction of

reactive organic gases (from hydrocarbons)  with oxides of

nitrogen.  While both smog ingredients have a number of sources

other than vehicles, and high levels of NO, may curb ozone

levels, it is simplistic to say that decreasing one ingredient at

the expense of the other improves air quality.

     Secondly,  as mentioned previously, capacity increases do not

always result in decreased congestion, especially in the  long

run. Reiterating, roadway improvements have the ability to spur

regional growth and thus attract more trips,  and the latent

demand for better roads creates post-completion induced trips.

In the short run, before development can occur and before people

change their driving habits to take advantage of the uncongested

road, emissions of CO and hydrocarbons may very well decrease.

However, in the long run,  as induced trips and/or regional growth

occur, these emissions may go up.  The roadway may end up

congested, but now with more vehicles and thus more emissions

than it had previously.

     Despite the problems mentioned previously with quantifying

induced trips,  it is still valuable to compare traffic

projections to actual counts.  I could only find two instances

where this has ever been done in the U.S.,  and both are small,

limited studies.  The Transportation Studies Center, a research

branch of the Department of Transportation, has agreed to provide

FHWA's Office of Planning with such a study.   The research will


tentatively look at 7 or 8 major highway projects from around the

country, and will compare actual costs to predicted costs in

addition to traffic volumes.   The results from this modestly-

sized study (FHWA HPN-23) will not be available for perhaps a

year, however.

B. Limitations of Data

     Comparing traffic projections to volumes was difficult.

Due to lengthy construction time, unresolved environmental

problems, or shortage of funds, 21 of the 27 projects I examined

were not yet completed.  One of the projects actually raaae no

prediction of future traffic volumes7.

     Often the traffic projections in the EA/IS or EIS/R were not

predictions of volumes (numbers of cars on the road), but rather

of minutes of delay.  Underlying this delay data were the volume

predictions, but locating the volume numbers was not

straightforward.  Sometimes they could be found in one or more of

the technical appendices, such as the Noise/Air/Energy reports.

For other projects, I obtained the needed data from the original

computer model printouts.

     To accurately model vehicle emissions at any point, at the

barest minimum, projections of the following are required for

each of the freeway's directions of travel:  the average annual

daily traffic (AADT; given in total vehicles per day), and the
     7  (Environmental Document, Ramp... on 237 in Santa Clara County
at Fair Oaks Avenue. 2/83).

a.m. and p.m. peak hour volumes.  This latter figure is usually

defined for urban areas as the number of vehicles passing the

point in the tenth busiest hour of the year3.

     Sometimes the traffic predictions were extremely crude.  For

example, some projects did not make predictions for each travel

direction on the freeway, but rather lumped both directions into

ADT (average daily traffic) and peak hour projections9.

Sometimes the ADT figure was stated to be the AADT; other tines

it was unclear if the ADT was the AADT or the ADT for -he busiest

month.  Other projects had no peak hour information, cur only ADT

projections10.  Other projects gave peak hour projections for the

a.m. only, ignoring the p.m. peak11.

     The major problems I encountered, however,  were a result of

Caltrans1 District 4's (Bay Area)  troubled traffic count program.

Over the last decade the program has been understaffed,

underfunded, and its information undervalued.  While recently

this trend has been partially reversed, the office is still low

on hardware and skilled staff.  California's traffic census
     8 Personal communications with Emory Stoker, Research Analyst,
Traffic  Engineering Branch,  Caltrans Headquarters,  Sacramento;
August 1990.

     9  (FEIS,  Roseville Bypass. 9/84).

     10 (ND/FONSI, Proposed Widening of 152 in Santa Clara County,

     11  (FEIS,  Route 101  in Santa  Clara  County;  fnew alignment
freeway from! Cochrane Rd....to 82. 7/78;  ND,  Widening 101  ...in
Santa Clara County  from the  Lawrence  Expressway  to the San Mateo
County Line. 10/86).

program calls for at least a third of the state highways and

interstates to be counted each year.  Thus the allowed maximum

time between counts on a roadway is three years.  In District 4,

during the last decade these requirements were rarely met.  Cfzen

traffic on a road went uncounted for 5 or more years.

     The lack of up-to-date traffic counts created problems when

trying to match a prediction to a count.  Predictions are always

made for a year at least 20 years in the future, and on some

projects for intermediate years also, often at 5 or "_ "D year

intervals.  Growth (or decline) in freeway traffic volumes is not

always linear, but is often exponential or logarithmic.  Thus, in

the typical case when the traffic was counted in a year for which

there was no projection, simply interpolating the accuracy of the

prediction based on straight-line traffic increases between the

present and the predicted year(s)  is somewhat inaccurate.

     Furthermore, the census program calls for the highway under

study to be counted for a period of at least a week; in each of

the quarters of the year.  Unfortunately, some of the District 4

traffic counts were less than one week, and some were as short as

3 1/2 days.  On most urban freeways and ramps, traffic varies

substantially throughout the week, typically with highest volumes

on Friday and other weekdays, and a marked drop in Saturday and

especially Sunday traffic.  In all cases where counts were made

for less than a full week, the number of days is noted in the

following Tables of results.   The ADT figures I calculated from

these short counts were not straight averages of the daily


traffic totals, but were adjusted using knowledge of weekly

variations on adjacent ramps and freeway sections to provide a

best estimate of an accurate ADT volume.  Similarly, for counts

of more than 7 days, the ADT's given are not a straight average

of daily traffic totals, but are averages of the averaged weekday


     Traffic volumes fluctuate not only with the day of the week,

but also by the season.  While seasonal fluctuation is generally

less on urban as opposed to rural roads, in California with our

emphasis on recreation, urban seasonal variations are still

significant.  These variations are usually not recorded in zhe

Bay Area, though, since the counts are made only once in the year

of the count.

     Another major obstacle to comparing predictions with counts

was that even if the predictions were fairly extensive,

frequently count information was available for only a small

portion of the project.  Basing a decision on the accuracy of an

entire project's traffic predictions on the validity of a small

portion's (e.g., one interchange) prediction is risky business.

As seen in the findings discussed below, on many of the projects

some predictions seemed reasonable while others were inaccurate.

On one project, the only portions to be counted were those for

which no predictions had been made12.

     Another problem in the data arises from the fact that
     12 (DEIS, 92 Gap Closure and 92/101 Interchange Completion  Fin
San Mateo County1. 2/79).

traffic on  freeway  ramps  is counted with  rubber air  hose

counters, instead of wire loops buried  in the roadway.  The  loops

record passing vehicles.   However the air hoses count,  axles,  not

vehicles, and record 2 axles as 1 vehicle. Therefore each  truck,

which has up to  5 axles,  incorrectly triggers the  ramp hose

counter, into thinking that 2.5 vehicles have passed.   There  is  no

separate or additional tally of truck traffic on ramps.  Thus,

the ramp volumes in the data actually overstate the  number of

vehicles.   Of course, if  the percentage of ramp traffic that:  is

trucks is very low, this  problem is minimized.

     However, I  chose not to use some arbitrary truck percentage

to adjust the ramp volumes down, because  of other  problems that

cause these figures to be too low.  Measuring only one week out

of the year and  recording the highest observed volumes as  the

peak hour volumes is inaccurate.  This is because  the actual peak

hour volume is the volume recorded if the highway  were to  be  ,

continuously monitored for an entire year and the  tenth highest

volume was  selected.  If  the volumes were randomly distributed,

the highest volume in a week of counting would be  only 28% of the

actual peak hour volume13.

     However, peak hour volumes are not randomly distributed,
     13 Actual peak hour =  10th highest hour
                          8670 hrs. each year
and 240 hours  in  10  days.   So observed peak  hour  is x of actual
peak hour, where:     10    =     x           x=.28
                    8670         240

because freeways have a finite carrying capacity.  While the peak

hour volumes I use as actual volumes are certainly more than 28%

of the real peak hour figures, they are still probably

underestimates.  Thus this error tends to counteract the effect:

of not adjusting ramp counts downward to account for the presence

of trucks on the ramp.

C. Findings

     Because of ~he reasons described above, it was possible re

include only 5 of the 27 projects in Part II of this szuay.  Four

were located in the Bay Area, and one was in the greater

Sacramento region.  Figures 1 and 2 show the location of the

proj ects.

     The first project was on 1-680 in eastern Contra Costa

County.  In 1985, a new interchange at Bollinger Canyon Road was

added to the freeway, and two existing interchanges, at Crow ,

Canyon and Sycamore Valley Roads, were expanded.  The FONSI/ND

prepared in 1983 predicted traffic levels on all the ramps for

the year 2005.   (See Figure 3 for diagrams of all the ramps).  In

late April and early May of 1986, Caltrans crews counted traffic

volumes on all the ramps, with their data collection varying from

10 to 3 1/2 days for different ramps.  The 1986 ADT, and a.m. and

p.m. peak hour volume counts are compared with the  2005

projections in Table 2.

     For these three interchanges, there was a total of 18 ramps.

For 11 of these  ramps, the 1986 volumes were significantly less


 Figure 1: Greater Bay Area Capacity Increases
            included in Part II
                      v*£'  I/1'580 [Hoffman Freeway]
                      ' -°    ° •*

                             .             -   • .  ,
                         iW^-—a  , -• \ .\ •U.I,*^. •.   1    '
                         fOo.-.^-, , , --. --_ ,•-.,,' 'iUm,', -.
                           •'-' ! ' OaitlcnaJ  ^ -—i-\ \ •";-^-• -.
                          -^i^t^---'1- ••/~X-'--'N- \   --^-^
                         ii "'"-1*11  	
   S«fL,; ~|»J-§;'^r* i :ir^^r W~fe^ I-680 Interchanges {
   ;,%«.,:r^ ^Y"'''^^^ II- Leanor° - 15S lv-f""' J^*^ '" y^43
   wN-j-H^-d   -r:i^m-,LKi
  ZfL^'x&iST? &%*''
11    J  ---.... mr^B^ '-^-^ «*-  ^^ MI i" 	.r -_1  -J_

\  Route 101- new alignment freeway

  / >^»-w<»ft.//cr"jrf.. ,---- •—»". \ \«. v.

            Figure 2: Sacramento Area Capacity Increase
                          included in Part II
       _^ii	!——=•  Roseville Bypass [Route 65]
                        i \ V ^V s
.? /'  !•'  U
••,. '.-«„..'  I1, s "  -^ ,.•••'

            Figure 3:  Ramp Configurations for 1-680 Interchanges
 SB off ramp'
 SB on loop
 SB on
SB off ramp
SB on loop
SB off loop,
SB on ramp
                              on ramp
                              on loop
                             off ramp
  Bellinger Canyon Interchange
           Crow Canyon Interchange
Sycamore Valley
                                                           >f>iB on ramp
                                                           NB off loop
                                                           ,NB off ramp
               SB off ramp
               SB on ramp
               SB off loop
                           Sycamore Valley Interchange

Table 2: Predicted and Observed Traffic Volumes on 3 1-680 Interchanges
Note: Predicted volumes are for the year 2005, and observed volumes are from counts
in April and May of 1986.  Ramp configurations are shown in Figure 3.  North and
southbound are abbreviated NB and SB, respectively. Interchange is abbreviated I/C.

                      ADT            a.m. peak hour       p.m. peak hour
Ramp         Observed   Predicted  Observed   Predicted   Observed   Predicted
                  1986   2005           1986   2005           1986   2005
(Bellinger Cyn. I/C:)

NB off ramp       5,890   18,950         1,435   2,285           714   1,505
NB on  loop'       2,186   1,850           488   130             189   240
NB on  ramp2      4,045   12,325          239   650             S95   1,515
SB off  ramp       5,458   15,175         1.097   2.U10           'i22   :,025
SB on loop2       3.296   15,500          373   1.250          I.U56   l.cSO
SB on ramp1       2,275   2,320           322   250             233   260
(Crow Canyon I/C:)
NB on loop 9,038 15,350 840 1,315
NB off ramp2 3,686 20,175 577 2,270
NB on ramp2 7,054 15,625 432 1,760
SB off ramp2
plus off loop 16,880 29,525 2,344 2,640
SB on loop2 3,105 13,175 245 1,200
SB on ramp1 5,976 6,775 491 535
(Sycamore Valley Rd. I/C:)
NBofframp1J 3,180 2,520 209 250
NBoffloopu 3,682 3,990 387 245
NB on ramp2 9,803 18,060 1,135 1,420
SB off ramp1-2 4,534 3,890 367 995
SB off loop2 6,313 9,770 358 995
SB on rampu 7,226 4,510 672 715

1,074 1,755
521 1,765
1,078 1,365

1,383 3,265
394 1,435
770 820

304 275
353 440
880 1,985
415 430
717 1,075
672 495
1 indicates a ramp with a significant underprediction for one or more parameters.
2 indicates a ramp where the observed ADT is an estimate based
week of traffic counting.
on less than a full


 than the 2005 projections, typically ranging from a third ~o

two-thirds of the 2005 projections.  Because of the 19-year

discrepancy between the counts and the forecast, this range of

observed traffic levels seems reasonable.  For at least one of

the three parameters of ADT or a.m. or p.m. peak hour, traffic en

7 of the ramps in 1986 was higher than that predicted for 2005.

Another of the ramps had 1986 volumes that were as high as 94% of

the year 2005 predictions.  These ramps are indicated in Table 2,

and are distributed across each of the three interchanges.

Caltrans seriously underestimated the future traffic volurr.es for

these ramps, since traffic modeling shows yearly increases in

traffic levels until the ramps or freeways are saturated, and

then constant levels.

     In 1984, 12 miles of new-alignment freeway opened in Santa

Clara County from Cochrane Road in Morgan Hill to Route 82 in San

Jose.  The new freeway, part of Route 101, replaced a length ^f

signalized highway known as Monterey Road and linked existing

freeway sections to the north and south.   The 1978 FEIS predicted

1995 AADT and a.m. peak hour mainline traffic volumes at 2 points

along the new freeway.  Two complete sets of these predictions

were made for two different population projections.  The

"losouth" predictions assumed a moderate rate of growth in

southern Santa Clara county,  while the "grosouth" predictions

assumed accelerated, substantial growth.   Unfortunately, no

predictions were made for ramp or p.m. peak hour volumes, and the

AADT figures are 2-way volumes (lumping both travel directions


into the given volume).

     Both the north and southbound directions of the new freeway

between the Bernal and Cochrane Roads interchanges (the southern

half of the project)  were counted for 6 1/2 days in August of

1985.  For the northern half of the new freeway, only the

southbound direction was counted, in both April of 1985 and

October of 1984 for 7 days each time.

     Table 3 lists the traffic forecasts and the actual counts.

Assuming that traffic volumes will not decline, Caltrans

substantially under-predicted both AADT and peak hour traffic on

this project.  Virtually all the predictions for 1995 using

either growth alternative were exceeded a decade early.

     Another project on Route 101 further north in Santa Clara

County widened the existing freeway from 6 to 8 lanes through the

towns of San Mateo, Mountain View, and Palo Alto.  Completed in

December of 1988, this project had peak hour predictions for 1995

and 2010 for all ramps and sections of the mainline.   These

predictions were made by DKS Associates'4.  Unfortunately, only

one of the numerous interchanges along this section of freeway

had been counted since the project was completed, and none of the

freeway mainline had been counted since it was widened.

     The interchange that had been counted, in April of 1989, is

the Lawrence Expressway/Route 101 interchange at the southern

terminus of the project.  The ramp configuration and designations
     14 (July 1987 Route 101 in Santa Clara County: Bernal Road to
the San Mateo County Line.  Corridor  Study and Operations Analysis.
Final Report. DKS Associates).


Table 3:  Predicted and Observed Traffic Volumes on Route 101
(new alignment freeway from Cochrane Road to Route 82)
Note:  All predicted volumes are for 1995.  North and southbound are abbreviated
NB and SB. "GroSouth" assumes high, accelerated population growth in southern
Santa Clara County; "LoSouth" assumes steady, continuing population growth.

Mainline south of Metcalf Rd. overcrossing (between Bernal and Cochrane

"LoSouth" traffic predictions:
Two-way ADT:         NB a.m. peak hour:       SB a.m. peak hour:
    45,000             2,420                    1.620
"GroSouth" traffic predictions:
Two-way ADT:         NB a.m. peaK hour:       S3 a.m. peak hour:
    56,000             2,970                    VJ80

Observed traffic volumes (based on a 6 1/2-day count 8/85):
Two-way ADT:         NB a.m. peak hour:       SB a.m. peak hour:
    68,201             2,722                    2,623
Mainline south of Blossom Hill Rd. (between Blossom Hill and Bernal

"LoSouth" traffic predictions:
Two-way ADT:                                       SB a.m. peak hour:
    54,200                                           1,950
"GroSouth" traffic predictions:
Two-way ADT:                                       SB a.m. peak hour:
    67,200                                           2,420

Observed traffic volumes (SB counts only; no NB counts):
(10/84)    SB ADT: 44,142                     SB a.m. peak: 2,980
(4/85)     SB ADT: 36,576                     SB a.m. peak: 3,034
SB average ADT: 40,359                 average SB a.m. peak: 3,007

Two-way ADT:

'(calculated assuming SB ADT=NB ADT; actual two-way ADT may be greater
since for mainline segment to the north, SB ADT was less  than NB ADT)

        Figure 4:  Lawrence Expressway/Route 101 Interchange
                        Ramp Configuration
                              Route 101
Off ramp from 101 SB
On loop to 101 SD
On ramp to 101 S
Off loop from 101 S
           On ramp to 101 NB
            ff loop from 101 NB
              loop to 101 NB
           Off ramp from 101 NB

 of  this full  cloverleaf are  shown in  Figure  4.

      Of the  8  ramps,  4  had 1989 a.m. peak hour  volumes higher

than the 1995  predictions, and 2 of these were  even  higher  than

the  2010 predictions.   The comparison  is shown  in Table 4.   While

it was only  possible  to judge the validity of the forecasts for

one  of many  intersections and nowhere  on the  mainline,  and  no

forecasts were made for ADT volumes, fully half of the

comparisons  possible  show substantial  underprediction.
      Table 4:  Predicted and Observed Traffic Volumes for the Route
      101/Lawrence Expressway Interchange
      Note: Predicted volumes are for 1995 and 2010, and observed volumes are
      from April 1989 counts. Ramp configuration is shown in Figure 4. North
      and southbound are abbreviated NB and SB, respectively.

                                aun. peak hour:
      Ramp:              1989 (Observed) 1995 Prediction  2010 Prediction

      On loop to 101 SB from Xway    334        539          651
      Off ramp from 101 SB to Xway1   701        693          771
      Off loop from 101 SB to Xway1    344        328          371
      On ramp to 101 SB from Xway    798       1,050         1,000
      On ramp to 101 NB from Xway   392        590          619
      Off loop from 101 NB to Xwa/    1,025        842          1XX)
      On loop to 101 NB from Xway    532        649          o91
      Off ramp from 101 NB to Xway1   722        674          720

      1 indicates a ramp where traffic was significantly underpredicted.
      The fourth and  final Bay  Area project in  this segment  of the

study was the  construction of  a new  freeway  from Route 80 to the

Richmond/San Rafael  bridge.  The 6-lane freeway, with an HOV lane

in both directions for much of its length, replaced  a road

network that consisted of both a signalized  highway  and a limited


access 4-lane expressway (Hoffman Boulevard).   Original plans

were for the freeway to be labelled Route 17 or 1-180, but today

it is designated 1-580, although it is not yet complete.  The

road is open to traffic, but a few interchanges are nor yet -n

their final configuration or are still closed.  Caltrans

officials believe that current traffic volumes are below the

volumes the road will carry when it is completed.

     Detailed predictions for 1995 ramp and mainline ADT and a.ni.

and p.m. peak hour volumes were made in ~he August 1973 P.evisea

Traffic Projections for the 130 (17) Corridor... -n tne Ti~v ;r

Richmond.  Caltrans counted traffic on most of the existing on

and off ramps in late April and May of 1990, but has not counted

mainline volumes on the freeway.  Indeed, counting the mainline

will prove difficult, for in the rush to open the new lanes after

the Loma Prieta earthquake of October 1989, no loops were


     However, because the bridge at the western end of the

project is a toll bridge, daily traffic volumes are always

recorded there.  Thus ADT is readily calculable, but only in the

direction of the toll, which is westbound.  Knowing the total

vehicles at the project's western terminus and adding and

subtracting all vehicles entering and leaving the mainline could

theoretically yield mainline ADT volumes at any point along the

project.  This was not possible, however, since Caltrans had not

counted the exit immediately east of the bridge, among others.

     Table 5 compares the 1995 predictions to the 1990 counts.


  Table 5:  Predicted and Observed Traffic Volumes on 1-580 [the Hoffman Freeway]

  Note: Predicted volumes are for 1995, and assume the freeway is completed. Observed volumes are
  from counts in late April and early May of 1990, when the freeway was not yet completed. West ana
  eastbound are abbreviated WB and EB, respectively.

                              ADT:           ajn. peak hour:     p.m. peak hour:
  Ramp                     Observed Predicted   Observed Predicted   Observed Predicted
                           1990    1995       1990     1995      1990    1995
  WB on ramp from Central Ave.1  4,458    2,940      378      260       511     140
  EB off ramp to Central Ave.    2^80    2,625      193      210       236     345
  WBofframptoBayviewAve.1   3,564    5,400      200      490       711     320
  WB on ramp from Bayview Ave.  1,904    5,870      180      495       182-   370
  EB on ramp from Bayview Ave.1  2,039    5,400      407      300       183     325
  WB off ramp to Erlandson St.   2,288    7,000      187      400       235     600
  EB on loop from Eriandson St.  1.801    7.000      123      500       222     360
  WB off ramp to 23rd St.1       5.506    7,860      316      iSO       599     -30
  EB on ramp from 23rii St.'     2J.22    3,950      372      100       226     ;1;>
  EB off ramp to 23rd St.        1.649    4,670      i28      :?5       i','1     -7~>
  EH on loop from 23rd St.1      3,937    4,320      884      3
Caltrans District 10 traffic count staff to carry relatively

minor levels of traffic, and thus the ADT mainline predictions

just east of Western Drive can be used as roughly equivalent to

the ADT predictions at the toll plaza.  Table 5 also compares the

toll plaza ADT count based on the late April and May traffic to

the mainline prediction east of Western Drive.  While District 10

traffic count staff suspect that west and eastbound ADT volumes

are not equal on 1-580, since the toll is only collected in the

westbound direction, there is no measurement of eastbound

traffic.  Thus I could only compare tne observed vestscund ADT

volume at this point to the westbound prediction to the east.

This prediction seems reasonable, for the 1990 traffic is

considerably below the level projected for 1995.

     Of the 6 projects in the greater Sacramento area, it was

possible to include only one, the Roseville Bypass, in this

second part of the study.  In September of 1987, a new-alignment,

4-lane limited access expressway opened from the existing Route

65 to 1-80, allowing through traffic to bypass the signalized

highway through downtown Roseville (see Figure 2).   This bypass

is essentially a new-alignment freeway.  Predictions of 2-way ADT

and peak hour volumes on the bypass were made in the 1984 FEIS

for traffic levels in the year 1987 and at 3-year intervals

thereafter.  Traffic on the bypass is counted continuously, and

Table 6 compares the predictions to the counts.

     The actual volumes are less than the predictions for this

project.  Roseville was predicted to develop substantially as the


                Table 6: Predicted and Actual Traffic Volumes on
                        the Roseville Bypass (Route 65)
       Note: Actual volumes are rounded to the nearest hundred. Actual ADT
       volumes are AADT figures; predictions are not assumed to be AADT figures.
       Actual volumes were obtained for a continuous count of the freeway for
       the entire year specified.
       Two-way mainline ADT volumes:
       1988 Actual    1989 Actual   1987 Prediction   1990 Prediction
        13,000      14,800      11,700        21,200

       Two-way peak hour volumes:
       1988 Actual    1989 Actual   1987 Prediction   1990 Prediction
        1,300      1,500        1,200        2,200
computer and  semiconductor industries  moved to  the area.   One

factor that may explain  the lower than expected volumes  is that

as  a  result of changes in  these industries much of the predicted

development has not occurred.

D.  Discussion of Part II Findings

      Despite  the numerous  problems and uncertainties discussed at

length earlier with the data this study uses, it is clear that

the planning  for freeway capacity increases has frequently

underestimated the traffic that actually uses the new or  improved

roads or interchanges.  While  the magnitude of  the

underprediction has varied,  in 5 of the 6 projects that were

analyzed, the forecasted traffic volumes or some portion  thereof

were  exceeded,  as much as  a decade ahead of schedule. Thus it

seems that there has been  a consistent underprediction of traffic

use for a variety of projects,  including those  planned quite

recently as well as more than  a decade ago.


     Looking in detail at the specific factors that could explain

where the forecast models erred is beyond the scope of this

study.  Certainly one problem is the lengthy time between project

planning and construction. Because of the complex environmental

process, the vagaries of construction, political opposition, or

inertia, planning and constructing a major freeway project can

take up to a decade or more.  This tends to decrease the validity

of the forecast models for several reasons.

     First, given infrequent traffic counts, the counts on -he

existing road network, wnich are an essential input tc the models

predicting future traffic, often predate the final environmental

document by as much as 4 years.  If traffic has increased during

this time, which almost invariably is the situation in

California, the models will tend to underpredict future traffic.

Second, lengthy time between planning and construction may mean

that the effects on traffic of significant increases or      •

locational shifts in regional growth in the meantime are not


     An "explanation" by Caltrans for the observed

underpredictions might simply be that regional growth outstripped

the projections of the various local and regional planning

bodies, and blame for this is simply not attributable to the

transportation providers.  However, increased regional growth is

often a function, at least in part, of freeway capacity


     Ultimately the reason for underprediction is irrelevant when


weighing the consequences of this trend on regional air quality.

Because congestion levels and tailpipe emissions are partially

determined by traffic volumes, underprediction of volumes means

that these projects' air quality analyses were inaccurate and

overly optimistic.  No matter the cause, consistent

underprediction of future traffic means that the supposed air

quality benefits of freeway work: have been consistently



     This study found significant flaws in the air quality

analyses done for freeway capacity increases.  The level of

detail of the analyses was often inadequate and the traffic

forecasts underlying the analyses showed a pattern of

underprediction of the improved roads' actual use.  Freeway

projects that were allowed on the belief that they were going^to

help solve our pollution problems have instead probably made them


     Checking on the accuracy of the traffic predictions is

essential to preventing further deterioration in air quality.

After years of federal and state reductions in highway project

funds, Proposition 111, in addition to county transportation tax

initiatives throughout the state, will provide a huge influx of

money into cash-starved construction programs.  Many projects

which have been planned for years will now have their NEPA/CEQA

documents prepared, or will be put out for bids.  If traffic


predictions turn out to significantly understate the actual

traffic levels on these projects, Californians will suffer by

breathing air that will be more noxious than predicted in the

projects' air quality analyses.

     Furthermore, given the drastically tighter standards of the

California Clean Air Act,  as well as the forthcoming Clean Air

Act amendments, it is now more important than ever to scrutinize

these types of highway projects closely to see that they don't

pull our urban areas into a smog-choked future fron wnicn we

could only escape at enormous social cost.

     To improve the planning process for these types of projects,

I suggest the following steps be undertaken.  Most urgently,

Caltrans should improve its traffic counting program, especially

in the Bay Area and other areas of the state where this program

is weak.  Spending over a billion dollars a year on construction

and so little on traffic counts is poor public policy, for the'

counts form the basis on which all highway planning rests.

     The technology exists, and is employed in the Sacramento

area, to continuously count freeway traffic.  For roughly $4,000

for each location, Sacramento has permanently installed the

hardware needed in conjunction with buried roadway loops to

continuously monitor freeway mainline and ramp volumes.  Using

loops to monitor ramp traffic, as is done in Sacramento,

eliminates the inaccuracy of air hose counts caused by vehicles

with more than 2 axles.  Also, new loops have become available

which are round in shape rather than square, and appear to be


considerably more reliable and less prone to  failure.   Broken

loops should be replaced, and all new projects should have  loops

installed as a matter of course.

     Predictions for traffic use on capacity  increasing projects

should be standardized to include estimates of AADT and a.m. and

p.m. peak hour volumes, for each roadway direction, rather  than

minutes of delay, combined 2-way volumes, or  other information.

Even more importantly, Caltrans should check  the accuracy of

their traffic predictions by comparing them to actual rcadvay

counts to continually improve predictions on  future projecrs.

     The analysis of the effect of the project on emissions

should as a minimum quantify the emissions on CO (both  in the

immediate project vicinity as well as regionally) , HC,  NO.,  CO.,,

and PM-10.  A practical monitoring and compliance plan  should be

required to be a part of every project that alludes to  mitigation

measures.  Interchange projects should model not only the    >

emissions of the cars actually on the ramps or in the

interchange,  but also the emissions of the vehicles on  the

mainline affected by the interchange traffic.

                           Appendix A

There is a fairly large body of research that examines the effect

of highways on regional economies and land use.  With the large

amounts of federal money provided for the interstate network in

the 1950's and early 1960's,  there was an accompanying flurry of

academic research, as well as studies sponsored by the DOT and

FHWA, on the effects of these highways.  Much of this work, falls

into one or more of the following categories:

     -The effecr of a highway bypass on the economy of "he

bypassed town or village;

     -The regional economic effects of urban beltways;

     -Land use patterns at freeway interchanges in both rural and

suburban areas;

     -Highway impacts on both actual and perceived property

values;                                                      ,

     -Land use changes resulting from arterial and highway

improvements, for both rural and for urban areas;

     -Demographic and community changes resulting from new

alignments and highway improvements.

     These studies have used a very diverse assortment of data

sources, including aerial photography,  number of and prices for

home sales, interviews with local officials and homeowners,

census tract information, gross sales and manufacturing data, and

county zoning maps.

     Some of the more interesting or relevant studies are


summarized below.  The call number, in parentheses, is used by

the Institute of Transportation Studies Library, at UC Berkeley,

where most of these are available.
Griggs, A.0.  Review of Some Effects of Manor Roads on Urban
Communities. '83.  Chapter 3, "Land Use Changes".
Includes summary of various studies done on the effect of house
prices.  Typically these show a slight decrease in value for
houses very close to the freeway  (a result of increased noise)
and increases for those homes nearby that benefit from the
improved mobility.  Range is -6 to +10%.  (NS 83-531).

Kingham, I.R. "Suburban Hwys. & Roads as Instruments of Land Use
Change," Trans. Research Record 565, '76.
Highway engineers see their task  as catching up on the provision
of road capacity to meet travel demand. "Suburban highways are a
result of land development and do not influence land use change."
All he did to conclude this, though, was interviews; a study with
limited use.

Buffington, J.L., et al. "Non-User Impacts of Different Hwy.
Designs as Measured by Land Use and Land Value Changes.  Research
Report 225-2.  Tx. Trans. Inst., Tx. A & M University. '78.
Found (for Texas) that urban areas were less affected by highway
improvements than suburban and rural areas,  because of the lack
of undeveloped properties to develop.

Adkins, W.G. & A.W. Tieken. "Economic Impacts of Expressways in
San Antonio, Tx. Trans. Inst., Bull.#11, '58.
Because of the lack of undeveloped/vacant land in the region,
there were few land use changes when the new expressways were

Adkins, W.G. "Effects of the Dallas Central Expressway on Land
Values and Land Use," Tx. Trans. Inst., Bull.#6, '57.
This freeway, built through a previous slum, created major land
use changes.  Much new commercial, industrial, and residential
development occurred.  Most of these changes occurred abutting
the highway or not far from it, though.

Duke, R. "The Effects of a Depressed Expressway— a Detroit Case
Study," The Appraisal Journal,'58.
Ford Expressway influences limited to roughly 300 metres on both
sides of it.

Palmquist,R.B.  "The Impact of Hwy. Improvements on Property
Values in WA State," for WA DOT.  '81.
Multiple regression study shows appreciation for areas with the


new highway is 15-17% higher than for those without.  Even within
600' of the road where noise is an issue,  appreciation due to
accessibility is generally greater than the noise depreciation.
A sophisticated study: he looked at over 9000 sale prices, as
well as interviewing residents.  However,  the increases in
appreciation occurred only where the highway could be used for
commuting.  (TA1001.5.P7).

Yu,J.C.& Allison,J.L. "A Methodology for forecasting  Beltroute
Corridor Land Use Impacts and its Application to Utah 1-215. Dpt.
of Civil Eng., U. of Utah,  '85.
Essentially the goal of this study is to allow towns to plan to
encourage "appropriate" (as defined by each community)  growth
along the road corridor.  The authors believe that beltways are
developed differently than regular limited-access expressways,
because they carry a different set of passengers.  They establish
a complex predictive model to be used by town planners across the
country, who must input data specific to existing land uses in
the region of their proposed beltway.  They test their model on
1-215 outside of Salt Lake.  Conclusions:  "[Beltroutes] are
particularly capable of altering, on a large scale,  the attitude
potential land users have for land within the region through
which the route will be located as well as for land within
reasonable distance of the beltroute."  "The perpendicular extent
of the  [beltroute] corridor is a function of the homogeny of land
use as well as homogeny of characteristics of the land itself.
The extent [of] such a homogeny, both perpendicular and parallel
to the beltroute, yields an indication of the potential area that
a particular land use may eventually occupy."  "A beltroute can
facilitate and even precipitate new land uses within the urban
area..."  (TA1001.5.P7).                                     •

Payne-Maxie Consultants, Blayney-Dyett Urban & Regional Planners.
"The Land Use and Development Impacts of Beltways: Case Studies,"
for DOT,  '80.
A huge study trying to pin down the effects of urban
circumferential highways for 8 regions.  Most of the roads
studied were built in the '60's or '70's.   It lacks any
statistical techniques to attempt to accurately link changes to
the road, but rather tries to do so anecdotally/qualitatively.
200 pages later, the reader has a picture of how areas develop,
but not how the roads affected their development.  (HE370.2.P2).

Barton-Aschman Assoc., for Illinois DOT. Highway and Land-use
Relationships in Interchange Areas. Springfield, VA. '68
A fairly typical example from numerous studies I saw on land use
at interchanges.  (There is a related set of research on the
economics of interchange location, typically in terms of sales
volumes, based on such factors as ADT on the highway).  This
study concludes "new hwy.  facilities have a strong tendency to
generate new uses of land that are often,  themselves, generators


of large traffic volumes"  [e.g., shopping malls].   (HE370.2.32).

Babcock,W.F.& Khasnabis,S. "An Analysis of the Impact of  Freeways
on Urban Land Developments in NC...," NC State Raleigh,  '74.
"Historically, the estimation of traffic for the freeway  &  the
intersecting roads has been accomplished by standard traffic
projection techniques.  Generally, these have been based  upon
existing land development plans and adopted transportation  plans.
In many cases, such predictions have not been realistic,  because
of unanticipated traffic that was generated by new land uses
brought about by the existence of the freeway."  This study  tried
to determine what development had occurred because of the  freeway
by looking at aerial photographs over time and interviewing  city
planners.  (4505 Microfiche).

Burkhardt,J.S.  Socio-economic Reactions to Hwy. Developr-.enx:.
'83.  Anaiyz-es effects of freeways typically built througn  urban,
and usually poor, areas in the 1960's & 1970's.  He looked  az
demographic, land use, housing market, etc. changes.
While such freeway building has essentially stopped roday because
such projects are no longer socially acceptable, he has some
interesting findings.  Measurable impacts were limited to a  5 to
10 block swathe adjacent to highway,  and the impacts were not
necessarily negative for the neighborhood.  Big conclusion:  the
general patterns of these freeways on the adjacent areas are
definitely secondary to site-by-site variations.  (TA1001.5P7).

Cosby,P.J.& J.L.Buffington OR Herndon,C.W.& Buffington.   "Land
Use Impacts of improving... Collins St. in a Developed Area  of
Arlington, TX/ Gessner Rd. in a Developing Area of Houston/  etc.
Tx. Trans. Inst., Tx. A&M U., Ntl. Tech. Info. Service. '79  to
These are 6 very similar studies which use the same research
methodology.   All look at improvements to arterial [non-
expressway] roads in different urban and suburban areas of Texas.
The studies set up 6 categories of land use, and guantify the
changes in land use occurring after the road upgrades.  Typical
conclusions;  "although the improvement of Collins St. helped
create an area more attractive for development, the impact on
land use [for this developed area] was not extensive.  a). Most
of the development in this area occurred— before the road
improvement began and was most likely not influenced by the  road
change....c)  The road improvement is  viewed as a positive
influence,  because if the street had  not been widened the
resulting congestion would: have been  a deterrent to development.
 OR: [for a developing area],  1)Commercial and multi-family
residential developments that were put on unimproved land were
located in the Gessner area partly because of the improved
access.  2). The improvement was also  important in the changes
from single family to commercial and  multi-family uses.   General
criticisms;  These studies are fairly simplistic.  By lumping all


land use into 6 categories, they do not distinguish between
density, quality, etc. of the areas before and after
improvements.  Furthermore, they limited analysis to what struck
me as an overly-narrow band along the improved roadways.  There
is no attempt to verify causality of changes, but only the
assumption that the changes are a direct result of the road.
Finally, none of the studies looked at limited-access highways.

Rollins,J.B. et al. Effects of Roadway Improvements on Adjacent
Land Use;  An Aggregative Analysis and the Feasibility of using
Urban Development Models.  Tx. Trans. Inst. in coop, with FHWA.
Research Report 225-22. Study 2-8-77-225.  '82
This study looked at 18 arterial improvements in non-rural Texas,
including the ones above.  Typical improvements were arterial
lane increases, adding medians, turning lanes, etc., and most
were done in the '70's.  This was easily the most statistically
sophisticated study I found.  Techniques included ordinary least
squares multiple regression and a simultaneous equations ^cdel
with two-stage and three-stage least squares. The ir.odei allowed
the changes in the percent of each of the 6 land use categories
to be explained by numerous other independent variables besides
just the road improvements.  Conclusions: net overall land
development is not significantly changed due to the roadway
improvements.  But, roadway improvements do affect the
development rates of specific types of land uses.  Residential
and public development are associated with ADT growth, and thus
road improvements.  (N.S.82-116).

Economic and Social Effects of Highway Improvements, Section IV,
"Land & Prop. Values & Land Usage in relation to Dort Hwy.    t
improvements. U.Mich., '61.
Older study, but some interesting findings.  "The Dort Highway  [a
Flint, MI bypass built in the late 1950's], like most other major
arteries, has been a powerful force molding & developing the area
which it serves."  "The highway, by providing accessibility,
makes it possible to subdivide large tracts for the more
intensive uses demanded as a result of increasing economic
activity and growth..."

Mountain West, Socioeconoroic and Land Value Impact of Urban
Freeways in Arizona, for AZDOT in coop, with FHWA.  (FHWA #AZ87-
282). '87.
Very thorough, careful study of the impact of freeway
construction on land use changes and property values in the
greater Phoenix region.  To study land uses, aerial photography,
zoning changes, census data, and planning documents
were used.  Property sales and valuation data and owner
interviews were also used.  This study used control areas
(lacking the freeways but otherwise similar) to link observed
changes specifically to the freeways.  Major findings:  "The
strongest and most obvious conclusion about the historic


 socioeconomic  impact  of  freeways  in metro  Phoenix  is  that
 freeways  are a necessary but  not  sufficient cause  for development
 to  occur.   Other  factors are  equally as  important,  including
 municipal planning and zoning,  land availability,  existing
 utilities and  infrastructure, and other  transportation modes—
 railroads and  arterials...  (etc).  Freeways merely create a
 condition that improves  the market opportunity  for change....
 Development around freeways can be controlled by strong urban
 land use planning.  However,  it is clear that income-generating
 properties— non-residential  uses and apartments— have strong
 locational  preferences for freeway corridors... [The]  intensity
 of  freeway  corridor development depends  on a combination of
 macroeconomic  demand  conditions and the  supply  of  developable
 land... Beyond these  broad statements, the specific kinds of land
 uses and their locations are  very much dependent on the
 peculiarities  of  place— existing land uses, existing zoning,
 etc...  Land values in proposed freeway  corridors  have increased
 due to road alignment announcements... It is clear that freeways
 have stimulated non-residential growth.1'
 This grossly oversimplifies very  detailed findings.

 Briggs,R. "The Impact of Interstate Hwy  System  on  Non-metro.
 Growth," DOT,  Office  of  Univ. Research.  Ntl. Tech.  Info.  Service,
 Statistically  sophisticated look  at the  big national  picture;
 included suburban, exurban, and rural growth across the country.
 The county  was the unit  of analysis.  Conclusions:  "The results
 of the research showed that, while counties with Interstates...
 have higher average growth rates,  even after confounding
 factors...  are controlled, the  presence  of a limited  access
 highway is  far from an assurance  of development for an  individual
 county... The  Interstate  system was less able to explain  the
 spatial pattern of development  than non-transportation  factors.
 (PB81-212987 fiche).
     Few studies investigate the effect of highway improvements

on induced trips/ADT/congestion, and thus air quality.  All too

often this topic is discussed in general qualitative terms with

apparently very little quantitative work having been done to

demonstrate actual effects.  The following studies touch on this

issue at least peripherally.

John Paterson Urban Systems. Feasibility of Assessing Effects of
Road Improvements on Trip Making and Urban Public Transport, for
the [Australian] Commonwealth Bureau of Roads, '71.
Essentially, this whole document is an attempt by a consultant to
secure a contract to write a computer model to forecast the
effect of highway improvements on induced trips.   He wanted about
Aust.  $25,000 in 1973 money to do it, and predicted that it would
take 3 people most of a year to write the model.   Apparently he
didn't get the contract.  This document/proposal is very general
and of limited use.  (HE370.2.J6).

Ziering,E. et al.  "Energy Impacts of Transportation System
Improvements," Trans. Research Record 870.
This study basically applies modeling and work done by numerous
other researchers into a new model.  "Unlike many earlier energy
impact estimation procedures, this methodology explicitly
considers induced and diverted travel resulting from a
transportation  improvement and the effect of znis travel ^n ~ne
level of transportation services."  Key to calculating rne
induced traffic levels are a set of travel-demand elasticities
developed by Charles River Associates for the CA Energy
Commission in 1982.  "[The model] produced results that were
frequently counterintuitive... and contrary to commonly accepted
conclusions [ie, road improvements conserve fuel by reducing
congestion] concerning the energy [and air quality] impacts of
projects."  Furthermore, "Highway widening or bypass projects can
either increase or decrease... consumption... Ramp-metering
projects yield energy savings when implemented under congested
conditions... In most cases, ramp delays reduce the amount of
induced new travel."

"The Vehicle-Miles of Travel-Urban Highway Supply Relationship,"
in Ntl. Coop. Hwy. Research  Program's Research Results Digest
#127, 12/80.
Summary of NCHRP project 8-19 by Cambridge Systematics et al.
which generated a computer model that relates highway supply to,
among other things, air quality.  This model, which apparently is
huge and consumes large amounts of computer time, was run on 2
Bay Area highways:  lane additions to Rte.24  from the Caldecott
Tunnel east to Concord, and  construction of Rte. 24 west from the
tunnel.  The model says both these projects have been net air
quality improvements for all pollutants except for NOx.