IMPROVE Cost Savings in a Flat Funded World - PowerPoint Presentation

Slide1
Slide2

IMPROVE Cost Savings in a Flat Funded World

Scott Copeland

10/16/14Slide3

We have done this before

June 2006 –

Expected FY2007 EPA budget cuts prompted an evaluation of which IMPROVE sites should be removed.June-December 2006

–

Marc Pitchford led a process to rank sites in the order to be shut down if necessary.

Several analyses done by Schichtel,

Poirot

and others.

Site List produced.

Comments solicited from states and FLMs were compiled.Slide4

Basic Reality

An purely objective method for site selection and ranking does not exist!

Need to use objective criteria and information to guide a subjective processSelect sites to keep under all circumstances (keepers)

Select sites suitable for decommissioning and rank them

Schichtel 6/2006Slide5

Sites with high fraction errors heavily weighted towards keeping

Site with low fraction errors Because haze is dominated in the east by sulfate, which is the most spatially uniform component, more of the eastern sites are redundant

Also show parts of AZ & MT as having redundant sites

Aerosol Bext Fractional Error

Schichtel 6/2006Slide6

Poirot

6/2006Slide7
Slide8

Overview of Comments

General comments received from 18 states, 5 RPOs, 4 EPA Regions, numerous FLMs

its premature (with regard to the RHR process) to shut down any of the 110 sites – SIPs not yet complete; need to ensure progress by trends tracking; some sites with only a few complete years of data; don’t know the fate of other protocol sites that would be caretakers

reducing the number of sites effectively diminishes the number of visibility-protected areas since the RHR uses monitoring data to define the pace of progress and document its performance

IMPROVE Steering Committee is not the appropriate body to make decisions since they can’t balance it against other air program needs

other approaches to reduce cost should be considered, instead of shutting down sites

the methodology of using current data to make decisions about redundancy is flawed for a 60-year trends program where emissions will undoubtedly change significantly

concerns that depending on a state or tribal protocol site for RHR tracking is vulnerable to changing priorities of the sponsor

No written comments were received supporting the reduction of IMPROVE monitoring networkSlide9

IMPROVE Response to Comments

Issues being considered

(brief responses in red)

Should we proceed with the priority listing of sites for decommissioning?

Yes, by categorizing sites instead of a single priority ordered list.

Are we the appropriate organization to do this?

Yes.

Is this the best time to do it? If not, then when?

Categorization now, final selection after the budget is available.

Should we pursue other ways to reduce cost (e.g. 1 day in 6 instead of 1 day in 3 sampling) instead of reducing sites? Not at this time.

Should we modify the current list of sites and if so how? Yes.Do we want to redo a data-based assessment to identify redundancy using other parameters or a different approach?

No, except for minor changes.

Should we work from the current list making changes based on comments received?

Yes, except for minor changes.

Should we change the reassignment of class I areas to remaining monitoring sites based on comments received?

Yes, in some cases.

Should we explicitly indicate our judgment about the degree of representation a site has for the class I areas assigned to it?

Yes, this is the thrust of our response.

Should we consider other ways to reduce cost in addition to reducing the number of sites?

Rejected at this time to preserve the utility of data at remaining sites for RHR tracking, source attribution, model testing, etc.

most sites only operating 4 years out of each 5

most sites only weighing the samples until years end when we choose the extreme mass events to analyze

one day in six instead of one day in 3Slide10

From

Assessment of the Potential Data Redundancy between

Nonmonitored CIAs and Their Representative Sites, Schichtel, 2007:

“These

results suggest that at least some of the

nonmonitored

CIAs have sufficiently different concentrations from the representative CIA that they warrant the addition of IMPROVE monitoring sites within or closer to the CIA

.”Slide11

We have done this before v 2

October 2012 –

Flat funding necessitated cost savings.October 2012-February 2013 –

I led a series of budget work group calls.

Several analyses done by

me,

Hand, Moore,

Poirot

and others.

Cost savings measures were implemented which avoided shutting down sites.Summary report produced.Those cost savings measures are still in place.Slide12

Excerpts from Moore, 2012

Data Requirements for Tracking Progress Under the Regional Haze Rule

Step

Responsible Agency

Reference *

Comment

Choice of

Deciview

as metric

n/a

40 CFR 51.308 (d)(1)

GTP Sec. 1.6

Identifies deciview as tracking metric, but does not define it mathematically.

Identify best/worst days

IMPROVE program

GTP Sec. 2.2, Step 8

GTP Sec. 4.2

The "best" and "worst" days are here defined as the cleanest (lowest) 20% and dirtiest (highest) 20% daily

deciview

values, per complete year (as determined in GTP Step 7).

Provides the calculation for determining how to identify the 20% best/worst days.

Calculation of the 5-yr deciview metric

IMPROVE program

GTP Sec. 2.2, Step 9

GTP Sec. 2.2, Step 10

GTP Sec. 4.3

Calculation of annual average deciviews for best/worst days (as determined in GTP Step 8).

Calculation of 5-yr average deciviews from the annual best/worst average deciviews for complete years (as determined in GTP Step 9). It is these 5-yr average

deciview

values for best/worst days which are to be used for setting the baseline and tracking progress.

Repeat of above discussion. Also defines 5-yr progress periods as "2005-2009, 2010-2014, etc."Slide13

Ancillary Uses of IMPROVE Monitoring Data – Potentially Affected by Funding Reductions

 A. Development and refinement of sampling and analytical methods

1-15

B. Assessment of sampling, analytical and data processing artifacts, errors and uncertainties

16-31

C. Comparison, evaluation and synthesis of methods and data across measurement networks

32-39

D. Characterization of aerosol formation mechanisms, composition and morphology

40-57E. Improved understanding of aerosol vs. optical relationships 58-89F. Assessment of long-term temporal and spatial patterns and trends 90-107

G. Regional and historical background for short-term intensive field and/or tracer studies 108-130H. Performance evaluation and refinement of regional and global air quality models

131-153

I. Input data for multivariate mathematical and/or back trajectory receptor models

154-195

J. Evaluation of “natural” source impacts (smoke, dust, sea salt, etc.) and regional air quality events

196-215

K. Assessment of sources, composition, optical & radiative properties of carbonaceous aerosols

216-233

L. Assessment of

transboundary

and intercontinental aerosol transport influences

234-245

M. Comparisons to and synthesis with remote sensing, modeling and surface observation data

246-251

N. Inverse modeling / development, confirmation and refinement of emissions estimates 252-257O. Assessment of human health and/or environmental impacts of specific aerosol species 258-267P. Assessment of sources of potentially toxic trace elements 268-274Q. Evaluation of single source impacts and control strategies 275-278

Excerpts from

Poirot

, 2012Slide14

Jenny Hand 2013Slide15

Cost

Cutting

Measures adopted

Projected Effective Savings

Stop ARS support of newsletter and reduce steering committee meeting support to note taking.

$ 50,000

Stop collection and analysis of backup quartz filters at 7 newest sites. Continue collection at original 6.

$ 34,000

Reduce number of collocated sites to 3 of each module rather than 6

$ 43,000

Skip one week of sampling network wide at Christmas

$ 38,000

Reduce funding for cooperative services agreement

$ 100,000

Reducing annual site maintenance efforts

$ 122,000

Use NPS dollars from 3 NPS protocol sites (TRCR1, DEVA1, INGA1) and MALO1 site for routine network operations

$ 63,000

Total:

$ 450,000 Slide16

We have done this before v 3

October 2013 –

Flat funding necessitated cost savings.October 2013-January 2014 –

I led a series of budget work group calls.

Several analyses done by myself and Hand.

Final call in January ended with a nearly perfect split in preference for site removal versus reducing sample

frequency, leaving consensus in doubt.Slide17

Why

the split?

Eliminating n samples per month would be harder to implement for the contractor. Data completeness criteria would need to be redefined. Introduces a discontinuity in data record. It is likely that it would be less efficient in terms of cost savings per sample lost or cost savings per relative error introduced. Will yield errors much larger than the reported iterative means at

some

sites each year.

Eliminating a site is “easy” for the contractors but focuses the pain on individual states, MJOs, EPA regions, and FLMs. Restarting sites could be problematic. It is implied that class I areas that lose a sampler would be reassigned to (RHTS?) site with highest redundancy. Slide18

“-1” indicates COHU1 or GRSM1 value below ~3*mdlSlide19

Fractional RMSE in annual haziest 20% dv

site

One per month

*

site

One per month

*

site

One per month

*

site

One per month

*

site

Site Removal

SAWT1

3%

CHIR1

2%

LOST1

1%

COHU1

1%

COHU1

2%

SENE1

2%

CRLA1

3%

LYBR1

2%

DOME1

2%

DOME1

2%

SEQU1

2%

CRMO1

2%

MACA1

1%

CACR1

1%

CACR1

3%

SHEN1

1%

DENA1

3%

MELA1

1%

JARI1

1%

JARI1

3%

SHRO1

1%

DOSO1

1%

MEVE1

2%

SAMA1

1%

SAMA1

4%

SIAN1

2%

EVER1

2%

MING1

1%

SWAN1

1%

SWAN1

4%

SIME1

1%

GAMO1

3%

MOHO1

2%

IKBA1

1%

IKBA1

5%

SIPS1

1%

GICL1

3%

MONT1

3%

MORA1

1%

MORA1

5%

SNPA1

1%

GLAC1

2%

MOOS1

2%

THRO1

2%

THRO1

5%

STAR1

2%

GRCA2

3%

MOZI1

3%

VOYA2

2%

VOYA2

5%

SULA1

3%

GRGU1

2%

NOAB1

2%

AGTI1

1%

AGTI1

7%

SYCA1

2%

GRSA1

2%

NOCA1

2%

ACAD1

2%

THSI1

2%

GRSM1

1%

OKEF1

1%

BADL1

1%

TONT1

1%

GUMO1

2%

OLYM1

1%

BALD1

2%

TRIN1

3%

HALE1

2%

PASA1

3%

BAND1

2%

TUXE1

2%

HAVO1

3%

PEFO1

2%

BIBE1

1%

Cumulative Fractional Error One per month:

195%

ULBE1

2%

HECA1

2%

PINN1

1%

BLIS1

2%

Cumulative Fractional Error Remove 11 sites:

45%

UPBU1

1%

HEGL1

1%

PORE1

1%

BOAP1

2%

VIIS1

1%

HOOV1

3%

RAFA1

1%

BOWA1

2%

WEMI1

2%

ISLE1

2%

REDW1

1%

BRCA1

2%

WHIT1

2%

JARB1

2%

ROMA1

1%

BRID1

2%

WHPA1

2%

JOSH1

1%

ROMO1

2%

BRIG1

1%

WHPE1

2%

KAIS1

2%

SACR1

1%

BRIS1

1%

WHRI1

2%

KALM1

2%

SAGA1

1%

CABI1

2%

WICA1

2%

LABE1

2%

SAGO1

1%

CANY1

2%

WIMO1

1%

LAVO1

2%

SAGU1

1%

CAPI1

2%

YELL2

2%

LIGO1

1%

SAPE1

3%

CHAS1

1%

YOSE1

2%

ZICA1

2%

*Mean of 100 IterationsSlide20

We have done this before v 4

October 2014 –

Flat funding necessitates cost savings.October 2014 +

It is not clear that more budget work group calls will yield a consensus.

My thought is to split the cost savings between removing sites and decreasing sample frequency.

Everybody

suffers

!