GUIDELINE FOR PM10 MONITORING AND DATA REPORTING
                    May 1985
      U.S. Environmental  Protection Agency
  Office of Air Quality Planning and Standards
     Monitoring and Data Analysis Division
         Monitoring and Reports Branch
 Research Triangle Park,  North Carolina  27711

-------
GUIDELINE FOR PM10 MONITORING AND DATA REPORTING
                    May 1985
      U.S.  Environmental  Protection  Agency
  Office of Air Quality Planning  and Standards
     Monitoring and Data  Analysis  Division
         Monitoring and Reports  Branch
 Research Triangle  Park,  North Carolina  27711

-------
                                  FOREWORD

     Many individuals were involved in developing the monitoring  strategy
to support the PM^g Ambient Air Quality Standards, as well  as the material
contained in this document.  For further information on specific  subject
areas, please contact the individuals listed below:
Subject Area

Sampler Allocation and
  Di stribution

Selective Sampling Schedule

SIP Development


Sampler Siting

Qua! ity Assurance
SAROAD/AIRS Data
  Hand! i ng System
Contact
Stan Sleva
Neil Frank

Joseph Sableski
Roger Powell

Neil Berg

John Puzak
Frank Me Elroy

Jake Summers
Phone Number
919 Area Code

  541-5651
  541-5558

  541-5697
  541-5697

  541-5651

  541-2188
  541-2622

  541-5694
FTS Number
 629-5651
 629-5558

 629-5697
 629-5697

 629-5651

 629-2188
 629-2622

 629-5694
Monitoring Methodology
Larry Purdue
  541-2665
 629-2665

-------
                             TABLE OF CONTENTS


                                                                    Page

   I.     Introduction                                                 1
  II.     PMio Sampler Allocation, Procurement,  and                    3
         Di stribution

 III.     Selective Sampling Schedule                                  7


  IV.     State Implementation Plan (SIP)                              14


   V.     PM]_Q Sampler Siting                                         16


  VI.     Quality Assurance                                           19


 VII.     Data Reporting                                              22


VIII.     Other Considerations                                        24
APPENDIX   The Selection of Sampling Frequency  for PMj_o Monitoring   27
           Prior to Promulgation of the PMio Standards
REFERENCES                                                          36

-------
              GUIDELINE FOR PMiQ MONITORING AND DATA REPORTING





                             I.   INTRODUCTION






    The major purposes of this guideline are to provide  interim  assistance



and information for the implementation of PM^g  sampling  networks to  support



PM]_o analyses and for the acquisition of data to be  used  in  the  development



of PM]_o State Implementation Plans (SIP's).  The guidance will apply for  the



interim period prior to promulgation of the NAAQS and SIP requirements, and



should be implemented immediately because of the scarcity of PMiQ ambient



air quality data and the potentially short-time period for SIP development



in areas of high probability of nonattainment of the PMio standard.



     Further, there are currently two separate   Federal  Register actions



pertaining to PM]_o anc' air Quality monitoring and data reporting.  On March



20, 1984, EPA proposed revisions to (a)  the particulate  matter standards1



and the reference method in 40 CFR Part 50^, (b) the monitoring  and reporting



regulations in PartSS^, and (c) the reference  equivalent method require-



ments  in Part 534.  The time period for public  comment on these  proposals



is still open and will remain so until 60 days  after the  proposal of
Part 51 SIP requirements.   In addition,  proposed generic changes to  Part 58



were proposed on March 8,  1985 .   These  latter changes  affect  PM10 as well



as other criteria pollutants.



     To significantly increase the available PM^o data  base, EPA has



purchased and distributed 662 PM^o samplers  for use  by  State and local



agencies to expedite establishment of PM^o monitoring networks.



     Due to the rather lengthy time period between the  Part 58 proposal and



projected promulgation date, numerous questions have arisen regarding the

-------
best approach for PM^o monitoring during  this  interim  period.  This guideline
has been prepared to respond to these questions,  and thus  it  is structured
in a question and answer format and  organized  by  subject area.  EPA suggests
a careful review of the proposed Part 58  (March 20, 1984 and March 8, 1985)
monitoring regulation amendments and welcomes  comments on  them.  In this
interim period, EPA encourages States to  begin monitoring  following, to the
maximum extent possible, the March 20,  1984, and  March 8,  1985, proposed
requirements for probe siting, station location,  quality assurance, and
sampling frequency.
     As stated earlier, the intent of this guideline is to set-forth criteria
and related rationale for establishing  networks for collecting data prior to
the time that final PM^o rulemaking  occurs.
     Strict adherence to the criteria contained in this guidance will not
constitute the only permissible way  to  collect PM^g that can be used for
SIP purposes.  Local adaptations of  this  guidance may be appropriate;
however, it is our recommendation that any such departures be carefully
considered, well justified by technical  and/or cost factors, and be
documented on a case by case basis.

-------
                  II.  PM10 SAMPLER ALLOCATION, PROCUREMENT,
                               AND DISTRIBUTION
Ql   What was the rationale behind the PM^o sampler distribution made

     by EPA?

Al   The rationale is described in the August 29, 1984 memo from Richard G.

     Rhoads, Director, Monitoring and Data Analysis Division,  to the Regional

     Office Environmental Services Division Directors.  To summarize, the focus

     is on acquiring PM]_o data in those areas which had a high probability of

     their TSP SIP's being inadequate to attain the PM]_o NAAQS based on  the

     best available data at that time.  These areas are, in descending order

     of importance; (a)  any area (county, urbanized area, city or town)  which

     has a probability of nonattainment equal to or greater than .95;  (b)  any

     urbanized area greater than 100,000 population which has  a probability

     of nonattainment equal  to or greater than .20 and less than .95;  (c)  any

     urbanized area greater than 250,000 population and less than .20 proba-

     bility of nonattainment;  ( d) any urbanized area equal  to  or greater than

     100,000 population  and less than 250,000 population with  a probability

     of nonattainment equal  to or greater than .05 and less than .20; and (e)

     any remaining area  (county, city, town)  with a population less than

     100,000 with a probability of nonattainment equal  to or greater than .41

     and less than .95.  Areas with populations less than 100,000 and with

     probabilities less than 0.41 will be considered prime candidates for

     FY-86 Section 105 Grant monies for PM]_o  samplers.


     The distribution of dichotomous samplers to urbanized areas was made

     considering factors other than PM^o nonattainment decisions.  These

-------
     factors include acid deposition,  visibility protection, and the likelihood



     of PMio SIP control  strategy  development.  In these situations, the



     ability to separately measure fine  and coarse PM^Q particulate matter



     may be important.
Q2   Where within the geographical  areas  identified in Al, should the



     samplers be located?



A2   Each area that has been identified  in  Al as an urbanized area, city,



     town, or a rural  part of a  county, had a specific NAMS or SLAMS TSP



     sampler from which the probability of  nonattainment was derived.  The



     PM]_o samplers should  be located  at those existing sites.  However, the



     local agency may have additional  information concerning sources, popu-



     lation or emission changes, or other factors which may suggest that



     another location within the area may be more suitable.





Q3   Should a PM]_g size selective inlet  (SSI) sampler be collocated with a



     dichotomous sampler?



A3   Not necessarily.  Since the monitoring objectives for the two samplers



     may be different, it  would  be  a  coincidence if one location could



     adequately fulfill both sets of  monitoring objectives.  For instance,



     the SSI might be used to demonstrate PM^o attainment at the expected



     maximum concentration location while the dichotomous samplers might be



     located in areas where visibility degradation and a vista with targets



     is important.  In another instance, the dichotomous sampler may be more



     suitable in a source  apportionment analysis where the fine (2.5 microns)



     versus coarse (10 microns)  data  could be important in determining a more



     cost effective control  strategy.

-------
Q4   With limited resources,  what is  the  priority  in which the PM]_g samplers



     should be brought on  line?



A4   Since most of those areas with a 95  percent or greater probability of



     nonattainment could be  required  to submit a full control  strategy SIP



     within 9 months after promulgation of the NAAQS, they should be brought



     on line first.  Next  would  be those  locations which consider the urban-



     ized area size and probability of nonattainment, and finally smaller



     cities, towns, or rural  county areas based on their probability of non-



     attainment.  In some  cases, the  control  agency may choose in lieu of



     PM]_g monitoring, to develop a SIP based  on a design value which may be



     the maximum one measured, rather than the second maximum, or even an



     extrapolated PM^g or  TSP value.   Decisions to forego PM^g monitoring in



     such areas would allow monitoring in areas of higher uncertainty, i.e.,



     those areas with probabilities between  .20 and .95.





Q5   What guidance can be  given  regarding the purchase of PM^g samplers?



A5   The question of sampler  procurement  by  State and local  agencies is



     addressed in terms of three time frames:  Case 1, between now and pro-



     mulgation of the PM]_g NAAQS; Case 2, after promulgation but before



     designation of any specific instruments  as reference or equivalent



     methods; and Case 3,  after  federal reference or equivalent method



     designation.





     For Case 1, any method  which is  based on the  proposed Federal Reference



     Method principles, is designed to produce a 50 percent cut point of lOum,



     and includes the availability of a manufacturer's warranty that the sampler



     will  meet or be upgraded to meet final FRM specifications is acceptable.



                                      5

-------
For Case 2,  as discussed in EPA's  SIP  guideline6, approval of ambient



PM^Q samplers for SIP purposes  will be made by EMSL, therefore, the



prospective purchaser should consult with  the Environmental Monitoring



and Systems  Laboratory (EMSL) Methods  Standardization Branch, for the



latest information on the status of those  monitors which have applied



for, or are  in the process of being approved as reference or equivalent



methods.





For Case 3,  only approved reference or equivalent methods may be



purchased for use as a NAMS or  SLAMS or  for other SIP monitoring purposes



It is likely that selected instruments purchased prior to promulgation



will be grandfathered for use for  up to  five years.  An appropriate



data adjustment factor may, if  necessary,  be developed for use with



the grandfathered samplers.

-------
                      III.   SELECTIVE  SAMPLING  SCHEDULE





     The issue of selective sampling  schedules  has caused by far, the most



concern in the implementation of the  PM^Q  monitoring  program.  Consequently,



a fairly lengthy discussion has been  prepared to discuss this issue and has



been included as an appendix to this  document.  This  section highlights a



number of questions concerning sampling  frequencies for PM^Q samplers.
Q6   What benefit is there in sampling  every day instead of continuing to



     sample every sixth day as is the current  practice for TSP?



A6   There are three major benefits  of  everyday sampling:  a) reduces the



     risk of miscl assifying the attainment/ nonattainment status of an area,



     b) provides a more accurate design value, and c) allows for attainment



     decisions to be made with less  than three years of data.





     Any area, where a 1 in 6 sampling  schedule is selected, would immediately



     indicate nonattainment as soon  as  the  first exceedance of the NAAQS is



     measured.  This one measurement would  translate into six expected exceed-



     ances as described in Part 50 Appendix K.?  However, it is possible



     that this one measured exceedance  was  the maximum concentration and the



     only exceedance for the area in the year, which could result in a



     mi sclassification of the area.





     The accuracy of SIP design values  are  also dependent on sampling



     frequency.   More frequent measurements provide for a better chance of



     actually measuring the second maximum  value (design value).  In general,



     the maximum concentration levels measured on a 1 i n 6 day sampling

-------
     schedule are substantially  lower  than  the  true maximum value.  Conse-



     quently, to  establish  an  accurate design value, from a 1 in 6 day sampling



     schedule, may require  the use  of  an  adjustment factor.  Typically, the



     measured values may be increased by  40 percent, more or less, depending



     on the variability of  the data base  used.





     Attainment decisions may  be made with  less than three years of data.   As



     proposed in Appendix K of Part 507 ,  a  single year of PM1Q data collected



     on a daily sampling schedule with at least a 75 percent capture rate



     would suffice.  Similarly,  2 years of  data with a 50 percent capture



     rate would also be sufficient  to determine the attainment status of an



     area.
Q7   Would it be better to obtain  some PMio  data  from three different



     locations rather than complete  data  (everyday sampling, i.e., three



     samplers at one station)  from one location?



A7   The response to this  question depends on the level  of the probability



     of nonattainment of the  PM^o  NAAQS and  the time period until promulga-



     tion of the PM^Q standards.   These factors are considered and discussed



     in detail in the Appendix for several situations and, therefore, are not



     repeated here.   A response, however, is given for the case of an area



     which already has a 95 percent  probability of exceeding the PM^Q NAAQS



     based on existing TSP data.   In this situation, if three separate sites



     were used on a 1 i n 6 day sampling schedule, there would be three separate



     chances  of determining a  NAAQS  violation (by recording a single exceedance)



     and no chance to unequivocally  pass  the attainment test.  If however,

-------
     the three samplers were used at one location,  presumably  the  location
     with the highest expected PMio levels,  then  a  determination of NAAQS
     attainment could be made after 1 year of everyday monitoring.

Q8   Assuming the availability of three PM]_o samplers per  site, are there
     any options available for relaxing the  weekend burden of  an everyday
     sampl ing schedul e?
A8   EPA is in the process of evaluating several  alternative sampling
     schemes which may alleviate this problem.  These alternatives are
     discussed below.  During the interim period  until promulgation, options
     b, c, d, and e are acceptable.   However,  the proposals are still  being
     statistically evaluated for their impact on  estimation of annual
     averages, expected exceedances, or data capture requirements, and
     therefore may be changed at promulgation.
     a.  Eliminate all  sampling on one of the  weekend days, i.e.,  Sunday,
         is an option.  However, it is not acceptable because  of the undeter-
         mined bias that would be introduced to the annual average and
         subsequent problems if the annual  average  turned  out  to be the
         controlling standard.   Also, the effect  on estimated  exceedances
         needs to be evaluated.
     b.  Adjust the sampling period from midnight-to-midnight  to noon-to-noon,
         This option would eliminate weekend site visits.  It  is not known
         if this alternative would introduce a bias into  the data  but this
         scheme would offer the most complete  data  capture.
     c.  Systematically alternate the non-sampl ing  day between Saturday and
         Sunday.   The bias to the annual  mean created by  this  sampling

-------
         scheme could be accounted for by analyzing enough data to determine
         a weekday/weekend difference.  The weighted average for each sam-
         pling regime (Monday through Friday regime and Saturday,  Sunday
         regime) would be added to create an annual mean.   The expected
         exceedances of the 24-hour standard would also be derived for each
         regime.  An exceedance on a weekend could count as two with  no
         additional  missing data since only 50 percent of  the  total weekend
         days were sampled.  Exceedances observed during the Monday through
         Friday regime would then be combined with the weekend exceedances.
     d.  Systematically alternate the non-sampling day between Friday,
         Saturday, Sunday, and Monday.  The main difference between this
         scheme and that described in (c) would be that an exceedance of  the
         24-hour standard on one of the Friday through Monday  4-day weekend
         regime days would be worth 1.33 exceedances rather than 2.   A
         factor of 1.33 or 4/3 is used since only 75 percent or three of  the
         total 4-day weekend regime days were sampled.  Conceptually, the
         annual mean would also be calculated on-a weighted average.  Further
         guidance on these approaches will be developed.
     e.  Operate a 4th sampler at the every day sampling site  and eliminate
         weekend vi sits.

Q9   Is sampling only during certain seasons of the year allowed (seasonal
     samp! i ng)?
     Seasonal sampling and the use of different sampling frequencies  depending
     on the  season of the year (increased sampling frequencies during the
     season(s) of greatest expected impact and reduced frequencies for  seasons

                                      10

-------
     of low potential)  are generally  not  allowed.   The  reason for disallowing
     seasonal  sampling  and varying  sampling  frequencies is that in most situa-
     tions there is insufficient PM^o  air quality  data  to make an informed
     eval uation.

     In a few cases, it may be possible to justify deviations from uniform
     sampling schedules.   In these  situations, at  least one complete year of
     data is required.   The number  of samples necessary for a complete year
     of data is determined by the area's  nonattai nment probability, i.e., an
     area with a >_ .95  probability  would  require 365 samples.  However, if at
     least 75 percent of these samples were  obtained with adequate coverage
     in each of the four calendar quarters,  this would be satisfactory.  All
     relevant air quality and emission data  must be presented to the Regional
     Administrator who  will  have final approval over any State's proposed
     deviations from normal  (uniform)  sampling schedules.

Q10  A large area with  a probability  of exceeding  the PM]_Q NAAQS of .95 or
     greater currently  has three or four  PM^Q sampling sites, all  with high
     concentrations.  One of these  sites  is  judged as the area of maximum
     concentration and  conducts everyday  PM]_Q sampling.  The others sample
     every sixth day.   If there is  only one  exceedance of the 24-hour standard
     in the area during the first year of monitoring and it occurs at the
     site monitoring everyday, it is  counted as one exceedance and the area
     is in attainment.   If, however,  the  exceedance is measured at a site
     which samples every sixth day  the expected exceedances are six and the
     NAAQS are considered to be violated. How will this situation be equit-
     ably hand!ed?
                                      11

-------
A10  In this example the assumption  is made  that  the  four  sites are within



     the boundaries of some  spatially defined area.   Guidance on defining the



     spatial extent of a nonattainment area  (or an  area having a high proba-



     bility of nonattainment)  is  given in  the probability  guideline.8 There



     is a provision in the  proposed  Part 58  Regulation9 to change the location



     of the monitor on the  selective sampling schedule based on the annual



     SLAMS review.   In addition,  the proposed SIP guideline^ specifies that



     the first observed exceedance will  not  be adjusted for missing data and



     be counted as  only one  exceedance if  everyday  sampling is initiated.



     Only one site  would be  required to  intensify sampling frequency.   If a



     single exceedance was  observed  at another site,  it would also be counted



     as only one exceedance  and not be adjusted for missing data.  The SIP



     guidelink specifies rules for  estimating attainment  status regarding



     the first observed exceedance,  subsequent exceedances and previous years



     of monitoring  data.  In any  event,  a  review  of the monitoring data should



     be made at the end of the year  along  with other  pertinent information,



     i.e., population density, emission  patterns, etc. to  determine if a



     change in the  maximum  concentration site and required sampling frequency



     is warranted.





Qll  If an agency is sampling every  sixth  day and an  exceedance is observed,



     is a determination of  NAAQS  violation automatic?



All  During the first year  of monitoring,  the first exceedance detected by a



     sampler operating every sixth day will  only  count as  one, provided



     sampling at the site commences  on an  everyday  schedule for the next



     year within 90 days after the end of  the calendar quarter in which the





                                      12

-------
exceedance occurred.   This assumes that a  second  exceedance  was not



registered at the site before  everyday  sampling was  initiated.  Further



guidance may be obtained from  the SIP  guideline.6
                                 13

-------
                    IV.  STATE IMPLEMENTATION PLAN (SIP)






     This subject is also addressed more fully in  the  Appendix, thus only a



few questions will be addressed here.
Q12  How do the requirements for PM^Q SIP development affect  PM^o  sampling?



A12  PMio SIP requirements would vary dependent upon  the  category  to which



     EPA assigns a particular area.  There will  be three  types  of  categories.



     The assignment to a category would be based initially on PMio nonattain-



     ment probabilities and then adjusted based on additional information



     which could affect categorization.  Group 1 would contain  areas with >95



     9 months of NAAQS promulgation;  Group II  would contain areas  with >_ 20



     to < 95 percent probability in which "committal"  SIP's would  be due



     within 9 months; and Group III would contain the  remaining areas for



     which existing TSP SIP's would be considered adequate for  PM^o-  States



     should consider placing PM^g samplers in  Group I  or  II areas  for reasons



     such as:



     a.  To obtain a design value for the area based  on ambient PM^o data,



     b.  To pass the attainment test so that no SIP will  be due,



     c.  To obtain ambient PM^o data to use in calculating an updated (more



         accurate) nonattainment probability,  or



     d.  To determine PM]_o background for the  area.



Prior to promulgation of the PM]_o NAAQS, States may wish  to move samplers



around among monitoring stations to accomplish as many data objectives as



possible.  See Section 4 of the Appendix for additional information.
                                      14

-------
Q13  How much PMio data is needed to determine  attainment of the PM^o NAAQS.
A13  Appendix K to 40 CFR Part 507  proposal  contains  the PM10 NAAQS attainment



     criteria as follows:



     a.  If three years of data are available,  then 12  observations per



         calendar quarter would be  sufficient.   The data must be representative



         of normal conditions, but  may be derived  from  l-in-6, l-in-2, or



         everyday sampling schedules.



     b.  If 3 years of data are not available,  then 2 years of representative



         everyday sampling data are needed and  with a 50 percent data capture.



     c.  If only 1 year of data is  available, the  1 year of representative



         everyday sampling data is  necessary, and  with  a 75 percent data capture





Q14  When will an area be officially categorized as Group  I,  II, or III?



A14  Area categorizations will be available  at  the time EPA promulgates the



          NAAQS.
Q15  What data base will  be used for this categorization?



A15  Several months before promulgation (as  accurately as this can be deter-



     mined) preliminary groupings will  be done  by  the States and Regional



     Offices using the most recent 3 years of ambient TSP and/ or PM^o data.



     These data will  be put into one-half calendar year segments and the most



     current six continuous segments will  be used.
                                      15

-------
                          V.   PMio  SAMPLER SITING





Q16  The TSP sites used to determine the probability of  nonattainment for



     PMio were classified as  neighborhood scale  with respect to roads and



     point sources if they were NAMS,  and no  closer than middle scale with



     respect to roads and point sources  if they  were SLAMS.  Now that micro



     scale PMio sites are allowed,  will  they  be  required?



A16  The reasons for allowing micro/middle scale PMio  sites is to provide



     additional locations from which sampling sites could be chosen rather



     than a mechanism for requiring all  category (a) maximum concentration



     sites to be microscale sites.   The  exclusion, in  the past, of microscale



     NAMS sites for TSP sampling was for the  purpose of minimizing the impact



     of larger particles coming from roadways on the collected sample.  For



     PM^o samplers, the impact of the larger  particles is eliminated by the



     design of the sampler.  Therefore,  it is by no means a certainty that a



     roadway microscale site  will yield  higher annual  or 24-hour levels than



     a neighborhood scale site in the midst of point sources with no heavily



     traveled roadway nearby.  If however, it is judged that a roadway corridor



     or street canyon type exposure would represent the maximum concentration



     area for PMio samplers,  then such a site should be used.
Q17  With multiple PMio and/or TSP samplers  at the  same  site, what are the



     requirements for separation distances between  monitors?



A17  If possible, a separation distance  of 2 meters should be maintained.  It



     is recognized that this may be physically impossible if the site is on



     the roof of a small  building or trailer and  there are three PMio samplers,
                                      16

-------
     a NAMS TSP sampler, and collocated  lead  samplers.  At a minimum, the



     samplers which are operating  concurrently,  i.e., the PM^Q and the NAMS



     TSP sampler or the collocated samplers must be a minimum of 2 meters



     apart,  It is also recommended that concurrently operating samplers



     have their exhausts directed  away from each other.  Additionally, any



     two particulate samplers operating  simultaneously for meeting precision



     requirements of 40 CFR 58 Appendix  A10 purposes, should not be more than



     4 meters apart.  For the case of PM]_o sampling everyday (three samplers),



     if a 2-meter separation distance between  PM^o samplers is not possible,



     then the greatest practical distance will be acceptable.  Further,  for



     collocated samplers the same  relative distance from roadways, parapets,



     or penthouse obstructions should be maintained and prevailing wind



     directions should be considered.
Q18  If a NAMS TSP site is selected  as  a  site  for PM^Q (which will  be



     operating on an everyday or  every  other day sampling schedule), how



     frequent must the hi-vol  be  run in order  to comply with the requirements



     in Appendix C11 which states that  the TSP  hi-vol sampler must be operated



     concurrently with the PM^o  sampler for  a  1-year  period?



A18  Once every 6 days is required.   It may  be  advantageous to run the hi-vol



     samplers more frequently, however, to cross check the PM^g results or to



     develop a site specific PM^Q/TSP relationship  for possible probability of



     nonattai nment calculations.





Q19  Under what conditions may a  high volume (TSP)  sampler be substituted



     for a PMio sampler?
                                      17

-------
A19  As specified in Appendix C to the proposed  Part  58 changes,11 a high
     volume (TSP) sampler may be used  in  place of  (designated as) a PM^o
     sampler if:
     a.  The sampling is for the purpose  of showing compliance with the
         NAAQS.
     b-  All 24-hour TSP concentration levels and  annual average TSP
         levels remain below the respective concentration levels of the
         PMio NAAQS.
     c.  The sampling site is not a NAMS  site.

Q20  The regulation states that as soon as  the TSP sampler (referred to in
     question 19) measures a single value that is  higher than the PM^Q 24-hour
     standard or has an annual  average greater than the PM^o annual  standard,
     it is necessary to replace it with a PM]_Q sampler.  How soon is soon?
A20  For the 24-hour standard,  the TSP sampler should be replaced with a PM]_o
     sampler before the end of the calendar quarter following the quarter in
     which the exceedance occurred.  This would  allow a minimum of 91 and a
     maximum of 181 days to procure and install  the required PM]_o sampler.
     For the annual  standard, the PM^o sampler should be operating by June 30
     of the year following the exceedance.   This would also allow up to 181
     days.
                                      18

-------
                            VI.  QUALITY ASSURANCE

     The questions and answers included in this  section  also  reflect  the
March 8, 1985 proposed changes to Part 58.12

Q21  What are the proposed precision assessment  requirements  for  PM^o samplers?
A21  The proposed reporting organization precision assessment requirements for
     PM]_o sampling are the same as for TSP sampling:   two  collocated  PM^Q
     samplers per reporting organization at each of the  two  highest concen-
     tration sites, operated concurrently at least once  per week.

Q22  What are the acccuracy assessment requirements for  PM]_Q  samplers in
     the March 1985 proposed Part 58?12
A22  The proposed accuracy assessment requirements for PM^o samplers  are also
     the same as for TSP samplers:  a fl ow audit of the  sampler's regular
     flow rate once per year (or more if there are fewer than four sites in
     the monitoring network), using a flow standard different than the one
     used for calibration.

Q23  How should precision and accuracy data be submitted under the proposed
     new reporting requirements?
A23  New forms have been developed for entering  the results of individual
     precision and accuracy tests.  If desired,  these  forms may be filled out
     directly at the monitoring site during the  precision  or  accuracy test to
     avoid the need for transcribing the data to another form later.   These
     forms are to be submitted quarterly to the  Regional Office.  Alternatively,
     we encourage the submission of the precision and  accuracy test results
     electronically via a terminal input program that  will be made available.
                                      19

-------
Q24  What precision and accuracy data  should  the  annual  report contain?



A24  Since in the March 1985 proposed  Part 5812,  it  states that EPA will cal-



     culate all  integrated precision and  accuracy estimates for all reporting



     organizations from the individual  information  submitted quarterly, no



     additional  precision or accuracy  information need be submitted annually -





QZ5  What are the new requirements for the number of collocated sampling



     sites for manual  methods?



A25  Under the new requirements (March 1985 proposed generic revisions to



     Part 58)125 the number of  collocated sites required per reporting organ-



     ization is determined separately  for each method by the number of sampling



     sites in the sampling network:  one  collocated  site would be required



     in a network of one to five sites, two collocated sites in a network of



     6-20 sites, and three collocated  sites in a  network of over 20 sites.



     These proposed requirements would also apply to lead sampling networks,



     for which collocated sampling replaces the previous requirements for



     analysis of duplicate filter strips.  The number of collocated sites for



     lead and TSP networks is determined  separately, even though common



     filters are analyzed for both TSP and lead.  However, in such cases, a



     single collocated site could serve both  networks.
Q26  What type of filter media should  be  used  in PM^o samplers?



A26  The proposed reference method  requires the filters to pass a collection



     efficiency test, handl ing integrity  test, and an alkalinity test.13  The



     alkalinity test eliminates glass  fiber filters for PM^g sampling.



     Quartz filters, however,  meet  the specifications and are currently being
                                      20

-------
used in size selective inlet (SSI) PMio  samplers.  Teflon filters are



recommended for use  in dichotomous PM]_o  samplers.





For TSP sampling,  States  should continue to use glass fiber filters.



The primary reason is for historical  data continuity-  Secondary



reasons are (a) cost of glass fiber  filters are about 40 percent of the



cost of quartz filters; and, (b)  quartz  filters are fragile and if they



are to be used with  hi-vols, the  hi-vols must be retrofitted to make



them cassette compatable.
                                 21

-------
                             VII.  DATA REPORTING

Q27  Can PM10 data be reported to SAROAD?
A27  PMio data from the 662 EPA supplied samplers must be  reported  to  SAROAD
     as part of the routine quarterly data submission.  PM^o  data ^rom  State
     and locally owned samplers which are to be  used  for SIP  purposes must be
     reported as well.  In addition, the Agency  requests that all PMiQ  data
     be reported to SAROAD.  The data must be coded on SAROAD Form  2 (Daily
     Data Form) as discussed in Volume II, AEROS User's Manual,  Section
     3.4.3.14  All data must pass routine SAROAD edits as  described in
     Section 7.1.2.

Q28  What specific codes must be used to report  PM^g  data?
A28  Interim codes were initially assigned in 1979-80 to permit  the storage
     of data for particulate matter data less than 15 microns.15  Interim
     codes were assigned in 1984 for PM]_Q.  These interim codes  utilized one
     pollutant code (81101) and different method codes to  designate collection
     mechanism and particle size.  These codes are listed  in  Volume V, AEROS
     Manual  of Codes, Section 4.5.0, Parameter Method File.15

Q29  When will  other SAROAD codes be assigned?
A29  Revised SAROAD codes will  be assigned for PM^o samplers  by  June 1985.
     The pollutant code for PM^o wl11  be 81102,  and a unique  method code will
     be assigned for each type sampler in use or being marketed.  For  the
     dichotomous sampler, the pollutant code will  be  81103  for 10-2.5 micron
     size range and the pollutant code will  be 81104  for <2.5 micron size
     range.   For a given dichotomous sampler, the same method code  will be

                                      22

-------
     assigned for all  size fractions.  The codes will  be  distributed  to



     Regional Office SAROAD,  NAMS, and QA  Contacts  when available.





Q30  How will the sampling frequency be handled?



A30  SAROAD does not currently provide the capability  to  store the sampling



     frequency.  SAROAD is being replaced  with the  Aerometric  Information



     Retrieval Systems (AIRS) in early FY-87.   Because SAROAD  is going to be



     replaced, no modifications are being  made to it.  The capability to



     store sampling frequency is currently being implemented on AIRS.



     However, agencies should provide EPA  with the  sampling  frequency for



     each PM]_Q sampler as part of its PM^Q network  description.
Q31  If a TSP sampler is used as a substitute PM^g  sampler as specified in



     Appendix C, how should the data be reported?



A31  Since it is a TSP sampler which is being used  only  to show compliance



     with the NAAQS for particulate matter (PMio) tne  data should continue to



     be reported like any other TSP sampler.
Q32  Should previously collected PM]_Q and PMis  data  be  reported  to  SAROAD?



A32  Yes, the more PM^Q or PM]_5 data available  to  characterize an area, the more



     accurate the classification.
                                      23

-------
                         VIII.  OTHER CONSIDERATIONS





Q33  Appendix K  indicates that PM1(j data from exceptional  events may be



     weighted or discounted in the computation of exceedances  or averages.



     What is the definition of a rare or unusual  event  for  PM^o sampling?



A33  The definition and criteria for these exceptional  events  are to be



     covered in a guideline document entitled "Guideline  on the Identification



     and Use of Air Quality Data Affected by Exceptional  Events."16  This



     guidance is currently in draft form.  A final  version  is  not expected



     until late 1985.






Q34  Should a PM^g sampler be used to replace a TSP sampler currently included



     in the National Particulate Network (NPN)?  This  network  was previously



     called the National Filter Analysis Network (NFAN) and National Air



     Surveillance Network (NASN).



A34  No, metals and benzo( a)pyrene (BaP) analyses will  continue to be performed



     on samples of particulate matter collected with the  high  volume air



     sampler on glass fiber filters.





Q35  How long should PM^o quartz filters be kept after  analysis or weighing?



A35  No specific time periods are specified in the regulations; however, it



     is sufficient to continue the same practices as currently exist



     with the agencies'  TSP glass fiber filters.





Q36  How frequently should PM^Q sampling be conducted  for Pollutant Standard




     Index (PSI) purposes?



A36  The regulation (Appendix Gr, Part 58)17 specifies  that  the data used to
                                        24

-------
     prepare the daily index report must be based on data  collected  from  the
     regular schedule sampling days.17  That is,  an  everyday  schedule requires
     daily PSI reporting while a 1 in 6 day schedule would only  require PSI
     reporting every sixth day.   EPA recommends,  therefore, that  initially
     the PM^o sampling frequency for PSI purposes be dictated by  the site's
     probability of nonattainment, i.e., greater  or  equal  to  .95,  sample  and
     report everyday; greater or equal  to .20 but less  than  .95,  sample and
     report every other day; and less than .20,  sample  and report every sixth
     day.  If after 1 year of monitoring at any of the  above  frequencies, the
     PSI value has not exceeded 50, the PSI reporting is left to  the discretion
     of the reporting agency.

Q37  Can PM]_o sampling for emergency episodes deviate from the reference
     method, similar to the way TSP was handled  in the  past?
A37  Yes, sampling over short time intervals and  staggered samples are
     acceptable alternates.  These topics are discussed in the draft docu-
     ment "Guideline for PM1Q Episode Monitoring  Methods."1

Q38  Should emergency episode data be submitted  to EPA?
A38  Yes.
Q39  Will additional PM^o samplers and support equipment such  as microbalances,
     calibration devices, and filter cassettes be furnished  to State and
     local  agencies?
A39  At this time, the Agency has no plans to directly  procure additional
     PMj_0 samplers and related equipment.   The Agency requested an increase of
     approximately $400K in FY-86 Section  105 grants to support the additional
                                      25

-------
     samplers and microbal ances by grantees.   Approximately 80 samplers



and 18 microbal ances could be purchased with  these  funds, but the majority



of the additional  equipment necessary to complete the  PM^g networks will



have to be obtained through normal  funding procedures.
                                 26

-------
 ^'PENDIX
     THE  SELECTION OF SAMPLING FREQUENCY FOR PM10 MONITORING PRIOR TO
                    PROMULGATION OF THE PM STANDARDS
                            1.  BACKGROUND

     The proposed Part 58 monitoring regulations for participate matter

specify minimum sampling frequency requirements for the first year and

subsequent years of PM1Q monitoring.  These are described in Section 58.13.9

EPA has allocated PM]_g instrumentation throughout the Nation, so that State

and local agencies can initiate monitoring prior to formal  promulgation.

These instruments have been allocated according to the estimated probabil-

ity of nonattainment using the probabilty guideline8 and TSP data, and

the corresponding Section 58.13 first year requirements.  The most stringent

PM1Q NAAQS (150 ug/m3, 24-hour and 50 ug/m3, annual) were assumed.  The

proposed regulations specify that the monitoring frequency for the site

with maximum concentration within an air quality monitoring area shall  be:

  o every day for areas with estimated probability >0.95 (Group I)

  o every-other day for areas with estimated probability 0.2-<0.95 (Group II)

  o every sixth day for areas with estimated probability <0.2 (Group III)

The Group I,  Group II and Group III classifications are used in the

proposed PM10 SIP development policy.  These PM10 nonattainment probabil-.

ities will  be redetermined based on the latest PM data just prior to

promulgation  of the revised NAAQS.

     The allocation of EPA equipment was based on a preliminary estimate

of the national  network of NAMS and SLAMS monitors.   Estimated probability

of nonattainment (i.e., the group category)  and other factors, such as

population size were also used in the allocation process-   Multiple

inst'   '<;ts were allocated at the maximum site per area to  permit the


                                    27

-------
appropriate number of samplers per site (operated  with  timers)  needed  to

sample at proposed frequencies and to minimize  site  visits:  three  instru-

ments were allocated for Group I NAMS and SLAMS, two instruments  were

allocated for Group II NAMS and SLAMS.  Single  instruments were allocated

for Group III NAMS and for other areas needing  more  than  one site.  Although

the instrument allocation will permit agencies  to  monitor at the  proposed

post-promulgation frequencies, how should these instruments be  utilized

prior to promulgation?  Should multiple instruments  be  placed at  a  single

location for frequent sampling or should the instruments  be spread  out to

cover more locations?  The following discussion addresses this question

from two viewpoints: (1)  how could available PMio  data  affect the estimate

of PMiQ nonattainment probability; and (2)  how  could available PM^ data

affect categorizing an area for purposes of PM^g SIP development? A summary

of recommended pre-promulgation sampling frequencies is found in  Table 1.



                                  TABLE 1

     Pre-Promul gation Sampling Frequency (for the  worst site per  area)


               Nonattai nment
     Group      Probabil ity	Sampling  Frequency

I
II
III

> 0.95
0.2 - < 0.95
< 0.2
1 i n 1 1 i n 2
recommended
recommended
optional'5
Iin6
-
optional9
recommended3
      a for attainment areas estimated  to  be  <  0.8  standard
      b for attainment areas estimated  to  be  >  0.8  standard
                                     28

-------
                  2.  PRE-PROMULGATION USE OF PM10 DATA



     The Probability Guideline specifies procedures for estimating  the



probability of not attaining the PMio standards.  According to this guideline,



a site  is assigned a probability of zero if it satisfies Appendix K to the



proposed Part 50 requirements for attainment determination with the use  of



available PM1Q data.7 This situation occurs if:




        o One year of PM]_o data ""s available with



         - every day sampling and 75 percent data capture



         - £ 1 exceedance and annual mean £ level  of standard



        o Two years of PM^o data are available with



         - everyday sampling and 50 percent average data capture



         - <^ 1 estimated exceedance and annual  mean £ level  of standard



        o Three or more years of PM^o data are available with



         - 12 observations per quarter



         - _< 1 estimated exceedance and annual  mean <_ level  of standard



     If PM]_g data are not sufficient to declare attainment,  then the proba-



bility  of nonattainment shall be calculated using both TSP and PM^o data,



following the methods in the probability guideline.8  In situations



where plans are to use TSP data for one or more years and PM^o data for  the



remainder of the 3-year assessment period, agencies should collocate TSP



and PM^g samplers in order to provide a site specific relationship  between




the two pollutants.





           3.  EFFECT OF TSP DERIVED NONATTAINMENT PROBABILITY



     For the purposes of this discussion, assume that 3 years of TSP data



with 60 samples per year were available.  Furthermore, each  year had similar






                                    29

-------
air quality, and each contained TSP concentrations above the level  of the



PMjo standards.



     If PM]_g is sampled 1  in 6 days, then each year of WIQ  data  simply



replaces a year of TSP data.  If any PMig exceedances are observed,  the



daily exceedance probabilities used in the computational  formulas  (3)  &  (4)



are defined as 1.0.  Thus, if any PM]_g exceedances occur, the  probability



of nonattainment automatically becomes 1.0.  On the other hand, if  a  year



of PM]_g data had no exceedances, then the 3-year nonattainment probability



will be made smaller.  The magnitude of the decrease depends on the  "replaced"



number of high TSP concentrations and their corresponding estimated daily



PM]_g exceedance probabilities.  See Section 5 of the probability guideline



for clarification and further details.



     If PM]_g is sampled more frequently, then one or more PM]_g exceedances



may not indicate violation of the PM^g NAAQS.  With every-other day  sampling,



one PM]_g exceedance may be permitted; with everyday sampling,  up to 3  PM^g



exceedances may be permitted.  In general, as long as the estimated  number



of PM]_g exceedances in a single year are less than or equal  to 3, then the



attainment test over a  3-year period has not automatically  been failed.



If PM^o exceedances are observed, but attainment or NAAQS violation has not



been determined, the probability of nonattainment is estimated using  prior



TSP data coupled with the  collected PMj_o data.   Recall  that  attainment



could be demonstrated with 1-3 years of PM^g data as discussed above.  The



probability guideline considers the impact of additional  Pf^g  exceedances



that would be expected with more frequent PM10 sampling.   Considering the



typical  case of prior years with 1  in 6 day TSP sampling,  the  nonattainment



probability calculations using the combined TSP and PM^g  data  would be the




                                     30

-------
same whether one exceedance is observed with  180  days  of  PM]_Q data or two
exceedances with 360 days of PMio data.   To this  extent,  the PM]_Q sampling
rate does not affect the computations of PM^o nonattainment probability.
     In all cases, especially if PM^o exceedances  have not been observed,
nonattainment probabilities which were based  on TSP  data  can be rederived
using site specific PMio/TSP information.  If PMio data are paired with TSP
data (from co-located samplers), then a site-specific  PMio to TSP ratio
distribution could be developed.  Frequent  PMio sampling  would be more
desirable for this purpose.  Using the revised distributions, the TSP
derived nonattainment probability can be recalculated.  This approach
should produce a more reliable probability  estimate.
3.1  Potential Group I Areas
     For these areas, more frequent PMio sampling  reduces the risk of
miscalculating the nonattai nment probability.  If  the  NAAQS is truely being
attained, then a single observed exceedances  with  1  in 6  day sampling could
maintain the Group I classification.  The ultimate probability will be
lower if no exceedances are observed with 1 in 6  day PMio sampling.  It
will be equivalently lower if one exceedance  is observed  with every-other
day sampling, or 2-3 exceedances are observed with every  day sampling.  If
the 24-hour NAAQS is truely being violated, then  exceedances are expected.
Again, more frequent sampling reduces the risk of miscalculation of the
nonattai nment probability.  With 1  in 6  day sampling, the true exceedance
day may not be sampled and the probability  of nonattainment might be
inappropriately reduced.
3.2  Potential Group II Areas
     For these areas similar arguments apply- More  frequent sampling
                                    31

-------
reduces the risk of nonattainment probability miscalculation.



3.3  Potential Group III Areas



     For these areas, attainment status is assumed.   Nonattainment probability



miscalculation with 1  in 6 day sampling is possible  but  not very likely.





                    4.  EFFECT ON SIP REQUIREMENTS



4.1  Group I Areas



     The maximum site in this area is expected to  be in  violation of the PM}Q



NAAQS and must have a SIP submitted,  with a full control  strategy, 9 months



after NAAQS promulgation.  In most cases, the 24-hour standard  is expected



to be the controlling standard.   This is based on  the probability calcula-



tions.  In order to estimate a good design value  for such an area, at least



365 observations are desired.9 Even if the annual  standard is controlling,



a  small data base may cause the 24-hour standard  to  appear to be controlling,



resulting in the wrong control  strategy and potential  over estimation of



control requirements.



     If the 24-hour standard is expected to be controlling, everyday



sampling at the expected high concentration site should  be initiated as soon



as possible.  This shall provide 1 year for data collection and 9 months for



data analysis and control  strategy development.   If  less  frequent monitoring



is performed, the design value must be estimated by  extrapolation from



observed measurements with potential  over or under estimation of control



requirements.  (The 24-hour design value is defined  as the concentration



with one expected exceedance per year, as discussed  in the PM^g SIP guide-



line.^ With incomplete sampling, the  design value  estimate is uncertain due



to variability among daily values and to possible  extrapolation beyond




observed PMio values).

-------
     If the annual standard were known to be controlling,  1  in  6  day
sampling could be sufficient to estimate the annual  design value.   Due to
variability in the estimation of 24-hour design values with incomplete
(i.e. 1  in 6 day) sampling, the 24-hour standard may appear to  be controlling.
Until a PMio data history is established, the misidenti fication of  the con-
trolling standard can be a problem.  Therefore, more complete data  is
necessary.
4.2  Group II Areas
     A sampling site in this area is likely to be close  to the  standard( s)
or not be in attainment.  Data for Group II areas will  first be used to
establish the adequacy of the existing particulte matter SIP and  next, to
develop a PM]_o control  strategy, if necessary.  A "committal" SIP for the
area must be submitted within 9 months after NAAQS promulgation.  Eighteen
months from SIP approval the State must declare the  SIP  to be adequate or
must submit a control strategy within the next 6 months  (24 months  total
from SIP approval).
     In light of these time periods, over 3 years from the present  will
elapse before States will  have to establish attainment or  determine that
the SIP needs modification for Group II areas, plus  an additional 6 months
for control  strategy development.  Every other day sampling  is  recommended
for the first year(s) of sampling to determine attainment  of the standards
or to improve design value estimation in potential  Group II  areas.
     If attainment is anticipated, then 3 years of data  with 1  in 6 day
sampling could be sufficient to declare attainment as  per  Appendix  K,
Section 2.4.7 However, If the site is in reality between 80  and 100 percent
of the standard (as estimated by its design value),  the  chance  of determining
                                    33

-------
that the NAAQS are being violated (i.e.,  measuring  a 24-hour  exceedance) is



estimated to be greater than 5 percent.19  Thus,  this 1  in 6  day monitoring



strategy could lead to this erroneous conclusion  and to  the resulting need



for a PM]_o control strategy.  In addition, the strategy  could result in the



need to intensify sampling to everyday following  the first measured exceed-



ance as per the proposed Section 58.13.9



     Every other day sampling, therefore, is  recommended for  the first year



of sampling and for subsequent years if the site  is anticipated to be



within 20 percent of the 24-hour standard (corresponding to the Selective



Sampling requirements of the proposed Section 58.13)-9   The site could be



sampled 1  in 6 days, however, if the State is prepared to increase sampling



to every day.  While the two instruments  needed for every other day sampling



at one location can be used to sample 1  in 6  days at two separate locations,



the increase to everyday sampling would require three instruments which may



mean the termination of a third sampling  site.   This possibility must be



weighed against the likelihood that a single  exceedance  would be observed.



If for example, the probability of nonattainment  from TSP data were closer



to 0.2U, the agency could take that chance while  if the  probability were



> 0.50, the agency might not be willing to do so.



     If attainment status is uncertain (e.g.  the  probability  of



nonattai nment is _>0.50), then exceedances of  the  standard are  expected and



design value development will be needed.   In  this case,  enough data should



be collected during the available 3-year  period to  determine  a good design



value.  This requirement indicates that at least  every other  day sampling



should be used.  Although NAAQS violation is  anticipated, the high
                                    34

-------
site may not necessarily be the site with the maximum  nonattainment proba-
bility.  This is because the relationship between  PMjo and TSP may be some-
what different from site to site.   Therefore, 1  in 6 day sampling, compared
to more frequent sampling,  would permit the  agency to  monitor at additional
locations,  thereby providing the opportunity to  explore the spatial varia-
bility of PMio and to find  "higher" concentration  sites.  However, with 1
in 6 day sampling, estimation of design values over a  3-year period will be
uncertain and control requirements may be over or  under estimated (recall
that at least 365 observations are desired.) If exceedances are not observed
during the first year(s) (perhaps, the probability estimate may have been
too high), then attainment  could be established.  As indicated earlier;
however, if the site is estimated to be between  80 to  100 percent of the
standard, at least 1 in 2 day sampling should be maintained, to minimize
the chance of erroneously concluding a NAAQS violation.
4.3  Group III Areas
     This site is expected  to be in attainment of  the  PMio NAAQS and
would not be required to sample more often than  1  in 6 days.
                                    35

-------
                                   REFERENCES


  1-   Federal  Register,  Part 50, 49:10430-10435.  March 20, 1984.

  2.   Federal  Register,  Part 50, Appendix J, 49:10430-10433.  March 20, 1984.

  3.   Federal  Register,  Part 58, 49:10441-10454.  March 20, 1984.

  4.   Federal  Register,  Part 53, 49:10454-10462.  March 20, 1984.

  5.   Federal  Register,  Part 58, 50:9541-9555.   March 8, 1985.

  6-   PMio SIP Development Guideline.  U.S. Environmental  Protection Agency,
      Office of Air Quality Planning and Standards, Research Triangle Park
      NC.  August 1984 (Draft).

  7.   Federal  Register,  Part 50, Appendix K, 49:10433-10435. March 20,  1984.

  8.   Pace, T. G. and N. H. Frank,  "Procedures  for Estimating Probability of
      Nonattainment of a PMio NAAQS Using Total Suspended  Particulate or
      Inhalable Particulate Data,"  U.S. Environmental  Protection Agency,
      Office of Air Quality Planning and Standards,  Research Triangle Park,
      NC.  February 1984.

  9.   Federal  Register,  Section 58.13,  49:10441-10442.   March 20,  1984.

10.   Federal  Register,  Part 58, Appendix A, Section 3.2.1,  44:27575.  May
      10, 1979.

11.   Federal  Register,  Part 58, Appendix C, 49:10445.   March 20,  1984.

12.   Federal  Register,  Part 58, Appendix A, 50:9541-9551.   March 8,  1985.

13.   Federal  Register,  Part 50, Appendix J, Section 7,  49:10431-10432.
      March 20, 1984.

14.   AEROS Manual Series, Volume II:   AEROS User's  Manual  (Third  Edition),
      Section  3.4.3.   U.S. Environmental  Protection  Agency,  Office of Air
      Quality  Planning and Standards,  Research  Triangle  Park, NC.  Publication
      No. EPA-450/2-76-029b.   July 1984.

15.  AEROS Manual Series, Volume V: AEROS Manual  of Codes  (Third Edition)
      Section 4.5.0.   U.S. Environmental  Protection  Agency,  Office of Air
     Quality Planning and Standards,  Research  Triangle  Park, NC.  Publication
     No. EPA-450/2-76-005b.   December  1984.

-------
16.  Guideline on the Identification and Use of Air Quality Data Affected by
     rxceptional  Events.   U.S.  Environmental  Protection Agency,  Office  of Air
     Qua! ity Planning and Standards, Research Triangle Park, NC.  October 1984
     (Draft).

17.  Federal  Register,  Part 58,  Appendix G,  44:27598-27604.   May 10,  1979.

18.  Pel ton,  D. J.,  "Guideline  for  Particulate Episode Monitoring  Methods,"
     GEOMET Technologies, Inc.,  Rockville,  MD.  Prepared for U.S.  Environ-
     mental  Protection  Agency,  Office of Air Quality Planning  and  Standards,
     Research  Triangle  Park, NC.   EPA Contract No.  68-02-3584.   February
     1983 (Draft).

19.  Frank, N.  H.,  Sleva, S. F.,  and N.  J.  Berg, Jr.,  "Revising  the National
     Anbient Air  Quality  Standards  for Particulate  Matter--A Selective
     Sampling  Monitoring  Strategy,"   U.S.  Environmental  Protection Agency,
     Office of Air  Quality Planning  and  Standards,  Research  Triangle  Park, NC.
     Paper No. 84-109.3.   Presented  at APCA Annual  Meeting.   June  1984.
                                      37

-------