Environmental Justice Screening Method (EJSM)...1 January 16, 2015 Rachel Morello-Frosch, Manuel Pastor, James Sadd, Madeline Wander Environmental Justice Screening Method (EJSM) Overview,

1

January 16, 2015 Rachel Morello-Frosch, Manuel Pastor,

James Sadd, Madeline Wander

Environmental Justice

Screening Method (EJSM)

Overview, Updates & Improvements

(Part 1)

AGENDA

Part 1: Overview, Updates, and Improvements EJSM – Method origins and rationale

Metric and layer updates

Results (statewide and regional scoring) Preliminary water results

Part 2: Data Accuracy, Enhanced Geospatial Approach and Scoring, Comparison with Other Screening Methods Data accuracy and ground-truthing

Improvements in Spatial Methods for Hazard Proximity

Comparing EJSM with CES and CEVA

2

RATIONALE FOR EJSM

Researchers and regulatory scientists want to better address cumulative impacts

Consider multiple environmental and social stressors and links to environmental health disparities:

Multiple hazards where communities live, work, and play

Vulnerability due to chronic social stressors • poverty, malnutrition, discrimination,

chronic health problems

RATIONALE FOR EJSM

Community and individual-level stressors can amplify pollution/health outcome relationships • Science still playing “catch up” to community wisdom

Traditional risk assessment does not account for the combination and potential interaction of hazard exposures and socioeconomic stressors.

Underlying science to achieve this will take awhile…

3

IN THE MEANTIME…

Indicator approaches to mapping cumulative impacts can advance EJ goals in decision-making, even as the science of cumulative risk assessment improves.

Integrating place – level measures of environmental and social stressors

KEY CONSIDERATIONS

Develop cumulative impact screening methods with neighborhoods and social context as important unit of analysis Sources of impact/vulnerability

Sources of resilience

Elucidate areas for regulatory attention or exposure reduction efforts Important for highly impacted and vulnerable

communities

4

ENVIRONMENTAL JUSTICE SCREENING METHOD (EJSM)

Develop indicators of cumulative impact that:

Reflect current research on environmental and social determinants of health.

Are transparent and relevant to policy-makers, regulators, and communities

Applicable for: Land use planning

Funding allocations

Regulatory decision-making and enforcement

Community outreach/engagement

Source: David Woo

EJSM DEVELOPMENT

Method co-created with stakeholder input (scientific review committee, regulatory scientists from different agencies, decision-makers, community organizations)

Helped identify and provided feedback on indicators and data inputs

Iterative process of review and methodological improvements

Engaged diverse communities in “ground-truthing” interim results

5

EJSM DEVELOPMENT

Maps where people live

Measures the “cumulative impact” using a variety of indicators

Mapping done at the census tract level

Scoring system: each tract receives “scores” related to quintile distribution of indicators

Statewide coverage, REGIONAL scoring

FIVE CATEGORIES OF CUMULATIVE IMPACT

Proximity to hazards & sensitive land uses Point and area emissions sources

Land uses associated with sensitive populations

Health risk & exposure State and national data sources

Social & health vulnerability Based on epidemiological literature on social determinants

of health

American Community Survey/Census Data

State data sources

Climate change vulnerability Based on climate change and health literature

Heat islands, temperature, social isolation

Drinking water OEHHA data

Metrics developed with stakeholder input

6

METRIC UPDATES – CATEGORY 1

Proximity to Hazards & Sensitive Land Uses

CATEGORY 1: HAZARD PROXIMTY DATA UPDATES

Replacement of AB2588 and Chrome Platers data set:

Facilities of Interest (CARB FOI): Greenhouse Gas (GHG) Mandatory Reporting database under AB 32

Facilities with emissions over 25,000 metric tons of CO2-equivalent (CO2e)

CA Emission Inventory Development and Reporting Systems (CEIDARS) – criteria pollutant facility data and toxic air pollutant data

Facilities emitting >10 tons per year

Five regional “industry-wide” data layers: Auto paint and body shops (CARB n=3847)

Gas stations (CARB n=9770)

Permitted hazardous waste (OEEHA; n=119)

Dry cleaners (CARB n=2422) [REMOVED]

Printing and publishing (CARB n=1089) [REMOVED]

Land use area sources same as in original proposal: Rail

Ports

Airports

Refineries

Intermodal distribution facilities

Traffic volume (census block value; contributes to score if in top 10% statewide)

7

CATEGORY 1: SENSITIVE LAND USES DATA INPUTS

Sensitive land uses as defined by CARB: (Air Quality and Land Use Handbook, 2005)

Childcare facilities

Healthcare & senior housing facilities

Schools

Urban Playgrounds & Parks

Residential neighborhoods

Polygons receive a score of 1 if they contain at least one sensitive land use category

CATEGORY 1: METHODOLOGICAL IMPROVEMENTS

Improvements in Spatial Methods for Hazard Proximity/Sensitive Land Use Layer …

… Stay tuned: Specifics on this will be covered in PART 2 of this presentation.

8

CATEGORY 1: SOUTHERN CALIFORNIA

CATEGORY 1: SACRAMENTO

9

CATEGORY 1: SAN FRANCISCO BAY AREA

CATEGORY 1: SAN JOAQUIN VALLEY

10

CATEGORY 1: SAN DIEGO


Health Risk & Exposure

11

CATEGORY 2: HEALTH RISKS & EXPOSURE DATA UPDATES

RSEI (Risk Screening Environmental Indicators) (2007-2010) average toxic conc. hazard scores

PM2.5 interpolated annual avg concentration (2009-11)

Ozone - sum of the portion of the daily max 8 hour concentration over CA standard of 0.070 ppm (2009-11)

NATA Respiratory hazard - mobile & stationary sources 2005 (National Air Toxics Assessment)

Calculated from modeled air toxics concentrations

Estimated Inhalation Cancer Risk (NATA 2005) Includes diesel (not recognized by US EPA)

Pesticide applications (lbs/m2) – (2009-2011)


12



13



14


Social & Health Vulnerability

CATEGORY 3: SOCIAL & HEALTH VULNERABILITY DATA

Census Tract Level Metrics (ACS 2008-12)

% residents of color

% residents below twice national poverty level

Home ownership - % living in rented households

Housing value – median housing value

Educational attainment – % population > age 24 with less than high school education

Age of residents (% <5)

Age of residents (% >60)

Birth outcomes – % preterm or SGA infants 2001-06

Linguistic isolation - % pop. >age 4 in households where no one >age 15 speaks English well

Voter turnout - % votes cast among all registered voters averaged for 2004, 2006, 2008, 2010 general elections

Socioeconomic Vulnerability

Biological Vulnerability

Civic Engagement

Capacity

15



16



17


CATEGORY 4

Climate Change Vulnerability

18

CATEGORY 4: CLIMATE CHANGE VULNERABILITY

land cover characteristics

across comparable neighborhood racial/ethnic minority groups

0%

10%

20%

30%

40%

50%

60%

Lo

s A

ng

eles

CM

SA

0.0

% to

19.9

%

20

% to

39.9

%

40

% to

59.9

%

60

% to

79.9

%

80

% to

100

%

Sac

ram

ento

CM

SA

0.0

% to

19.9

%

20

% to

39.9

%

40

% to

59.9

%

60

% to

79.9

%

80

% to

100

%

San

Die

go

MS

A

0.0

% to

19.9

%

20

% to

39.9

%

40

% to

59.9

%

60

% to

79.9

%

80

% to

100

%

San

Fra

nci

sco

CM

SA

0.0

% to

19.9

%

20

% to

39.9

%

40

% to

59.9

%

60

% to

79.9

%

80

% to

100

%

per

cen

tag

e la

nd

co

ver

tree canopy

impervious surface

Shonkoff, Morello-Frosch et al. Climatic Change 2012.

Heat Island Metrics and Racial/Ethnic Composition

CATEGORY 4: CLIMATE CHANGE VULNERABILITY DATA

% tree canopy % impervious surface • NLCD, 2012

Projected max monthly temperature (2050-2059)

Change in projected max monthly temperature • (2050-2059) – (2000-2009)

Change in degree-days of warm nights (19°C) • ((2050-2059) – (2000-2009)) • National Center for Atmospheric Research, downscaled Community

Climate System Model, scenario B1, ensemble average & Cal ADAPT

% elderly living alone % car ownership

• American Community Survey Summary Data (ACS) 2008-2012

Heat Island Risk

[Statewide scoring]

Temperature [Statewide scoring]

Mobility / social

isolation [Regional scoring]

19



20



21


CUMULATIVE IMPACT SCORE

22

CUMULATIVE IMPACT SCORE

Total Cumulative Impact Scores at the Tract Level:

Sum the four impact and vulnerability scores = Hazard Proximity and Sensitive Land Use (1-5) +

Health Risk and Exposure (1-5) +

Social and Health Vulnerability (1-5) +

Climate Change Vulnerability (1-5)

Final Cumulative Impact Score ranges from 3-15 or from 4-20 with climate vulnerability

CI SCORE: CALIFORNIA – NO CLIMATE VULNERABILITY

Statewide Scoring Regional Scoring

23

CI SCORE: CALIFORNIA – WITH CLIMATE VULNERABILITY


CI SCORE: SOUTHERN CA – NO CLIMATE VULNERABILITY


24

CI SCORE: SOUTHERN CA – WITH CLIMATE VULNERABILITY


CI SCORE: LA REGION – NO CLIMATE VULNERABILITY


25

CI SCORE: LA REGION – WITH CLIMATE VULNERABILITY


CI SCORE: SF BAY AREA – NO CLIMATE VULNERABILITY


26

CI SCORE: SF BAY AREA – WITH CLIMATE VULNERABILITY


CI SCORE: SJ VALLEY – NO CLIMATE VULNERABILITY


27

CI SCORE: SJ VALLEY – WITH CLIMATE VULNERABILITY


CI SCORE: SAN DIEGO – NO CLIMATE VULNERABILITY


28

CI SCORE: SAN DIEGO – WITH CLIMATE VULNERABILITY


NEW CATEGORY: WATER QUALITY

29


Measure Variable , Data Used Scoring Method

Potential exposure to contaminants

time-weighted average system-level water quality, OEHHA

Sum of ratios of concentrations divided by MCL for 17* contaminants

Technical, managerial and financial capacity

‘population served’, PICME

• 4: <500 people, “very low TMF” • 3: 501-3300 “low TMF” • 2: 3300-10000 “medium TMF” • 1: >10000 “high TMF” • Townships given worst score

Physical vulnerability to outages

# and type of sources, PICME

• 5: GW only system with <=2 sources • 4: GW only system with 3-4 sources • 3: GW only system with >4 sources • 2: GW-SW combined system • 1: SW-only system= • Townships are given the worst score

Compliance burden monitoring & reporting violations and/or no water quality data in compliance period for arsenic, nitrate, perchlorate and TCR, PICME and OEHHA

• 4: M&R violation for all 4 contaminants and/or no water quality data

• 3: M&R violation for 3 contaminants and/or no water quality data



4 Drinking Water Metrics:


Drinking Water Contaminants: Contaminant selection criteria: 80% or more of community water systems had data for contaminant between 2005-2013

Arsenic

Barium

Benzene

Cadmium

Carbon Tet.

Lead

MTBE

Mercury

Nitrate

Perchlorate

PCE

Radium 226

TCE

Total Trihalomethanes

Touluene

Xylene

Total Coliform

List of 17 contaminants:

30


Lowest Unit of Analysis Before Aggregating to Tract: Community Water Systems and Townships

Following OEHHA’s approach: Two types of public community water system geographies are

scored:

• Community Water Systems with known boundaries (1,562 systems; 33.7 million people)

• Community Water Systems with estimated boundaries (1429 systems; 1.3 million people)

For areas of the state not covered by CWSs, a 6x6 mile grid of townships is used to define areas where people are likely to be drinking groundwater– e.g. from private wells or very small systems (1.5 million people)

~.6 million people are not assigned water quality because they are not within a township that has a groundwater sample


Deriving Tract-Weighted Averages

Time-weighted average concentration for each contaminant calculated for each CWS or township (OEHHA) • For Total Coliform, there is no concentration but a “0 or 1” assigned to each system

for whether it had an MCL violation

Census blocks (or portions) assigned contaminant concentration associated with CWS or township (OEHHA)

Population-weighted average is then aggregated to tract-level

Each contaminant’s tract-level, population-weighted, average is divided by the MCL (except for Total Coliform): • Tracts with missing data assigned tract-level average across the EJSM region (so that

there are no missing values that would count as a zero in our sum of ratios)

The 17 concentration/MCL ratios are added to produce: sum of ratios

Sum of ratios allocated quintile score based on regional distribution

31


Deriving Tract-level Vulnerability Score

For TMF, Physical Vulnerability and Compliance Burden: Each community water system (CWS)

Receives a score based on data

Each township Receives worst possible score

Population-weighted average is then aggregated to the tract-level Tracts with missing data, get filled in with a tract-level average across the EJSM region

Population weighted average is summed (range from 2-13)

Tracts with a sum >=8 are considered “high or very high vulnerability”


Final Composite Drinking Water Score

Score includes: Water Quality + System-Level Vulnerability

Step 1: Because of low variability, TMF, Physical Vulnerability and Compliance Burden summed into 1 “System-Level Vulnerability Score”

◦ Range 2-13

◦ Systems with >=8 are considered “vulnerable”

Step 2: Tract-level sum of contaminant concentration/MCL ratios are quintiled, by region (or state)

Step 3: Add 1 point to the quintile score if the percentage of a tract’s population drinking from a vulnerable CWS is >= 90th percentile

Step 4: Add 1 point to the quintile score if the percentage of a tract’s population drinking from “non-public, private wells” (i.e. townships) is >=90th percentile

Step 5: Force the values of 1-7 to 1-5 Final composite score ranges from 1-5

32


A note on geographies…

Following OEHHA’s approach: Scores given to community water systems

…And to township areas not served by community water systems

CALIFORNIA WATER QUALITY

Regional Scoring Statewide Scoring

33

SJ VALLEY WATER QUALITY WITH VULNERABILITY OVERLAY


SOUTHERN CA WATER QUALITY WITH VULNERABILITY OVERLAY


34

SF BAY AREA WATER QUALITY WITH VULNERABILITY OVERLAY


SACRAMENTO WATER QUALITY WITH VULNERABILITY OVERLAY


35

January 16, 2015 Rachel Morello-Frosch, Manuel Pastor,

James Sadd, Madeline Wander

Environmental Justice

Screening Method (EJSM)

Data Accuracy, Geospatial Method

Improvements, Comparisons (Part 2)

AGENDA

Part 2: Data Accuracy, Enhanced Geospatial Approach and Scoring, Comparison with Other Screening Methods

Data accuracy and ground-truthing • Metrics update for point hazards

• Facilities of interest (FOI), hazardous waste, auto paint and body shops, gas stations

• Validation and correcting locations

Improvements in Spatial Methods for Hazard Proximity • Use of parcel data to create CI polygons

• Introduction of point distance method

Comparing EJSM with CES and CEVA

36

DATA ACCURACY AND GROUND-TRUTHING

DATA INPUTS FOR HAZARD PROXIMTY LAYER

Replacement of AB2588 and Chrome Platers data set:

Facilities of Interest (CARB FOI): Greenhouse Gas (GHG) Mandatory Reporting database under AB 32

Facilities with emissions over 25,000 metric tons of CO2-equivalent (CO2e)

CA Emission Inventory Development and Reporting Systems (CEIDARS) – criteria pollutant facility data and toxic air pollutant data

Facilities emitting >10 tons per year

Five regional “industry-wide” data layers: Auto paint and body shops (CARB n=3847)

Gas stations (CARB n=9770)

Permitted hazardous waste (OEEHA; n=119)

Dry cleaners (CARB n=2422) [REMOVED]

Printing and publishing (CARB n=1089) [REMOVED]

Land use area sources same as in original proposal: Rail

Ports

Airports

Refineries

Intermodal distribution facilities

Traffic volume (census block value; contributes to score if in top 10% statewide)

37

GROUND-TRUTHING THE DATA

Validating and Correcting using Google Earth Pro

Example: So. Cal. Gas RCRA site

• Reported location downtown Los Angeles

• Actual location (reported street address) >10 miles away




• Corrected location is adjacent to residential land use

38




• Facility better represented as an area source (polygon)

• Different proximity relationship to residential land use



Example: Lawrence Livermore National Lab – Site 300

• Reported location differs from reported address

39





LOCATION CORRECTION

Hazardous Waste Facilities Layer


40


LOCATION CORRECTION

CARB FOI facilities in the San Joaquin Valley

FOI facility location as

reported by CARB

Corrected location


Sample of Location Errors - Hazardous Waste Sites (San Joaquin Valley only)

EPA_ID PROJECT NAME ADDRESS CITY Error (m)

CA2890090002 LAWRENCE LIVERMORE NATIONAL LAB - SITE 300 CORRAL HOLLOW RD TRACY 12,764

CAD990794133 FORWARD LANDFILL 9999 S AUSTIN RD STOCKTON 11,705

CA1570024504 EDWARDS AIR FORCE BASE 5 E POPSON AVE EDWARDS 1,519

CA4170024414 OCCIDENTAL OF ELK HILLS INC 28590 HIGHWAY 119 TUPMAN 1,500

CAD980813950 CRANE'S WASTE OIL INC 16095 HIGHWAY 178 WELDON 614

CAT000646117 CHEMICAL WASTE MANAGEMENT INC KETTLEMAN KETTLEMAN HILLS LDFL HWY 41 KETTLEMAN CITY 478

CAL000190816 RIVERBANK OIL TRANSFER, LLC 5300 CLAUS RD RIVERBANK 238

CAL000282598 BAKERSFIELD TRANSFER INC 1620 E BRUNDAGE LN BAKERSFIELD 231

CA2170023152 NAVAL AIR WEAPONS STATION CHINA LAKE 1 ADMINISTRATION CIR RIDGECREST 188

CAD982446882 EVERGREEN OIL INC FRESNO 4139 N VALENTINE AVE FRESNO 144

CAD066113465 SAFETY-KLEEN 3561 S MAPLE AVE FRESNO 115

CAD981429715 KEARNEY-KPF 1624 E ALPINE AVE STOCKTON 107

CAL000102751 WORLD OIL - SAN JOAQUIN LLC 14287 E MANNING AVE PARLIER 99

CAT080010606 BIG BLUE HILLS PESTICIDE CONT DISPOSAL 10 MILES NORTH OF COALINGA COALINGA 76

CAD982435026 KW PLASTICS OF CALIFORNIA 1861 SUNNYSIDE CT BAKERSFIELD 34

CAT080010283 EPC WESTSIDE DISPOSAL FACILITY 26251 HIGHWAY 33 FELLOWS 33

CAD980675276 CLEAN HARBORS BUTTONWILLOW LLC 2500 WEST LOKERN RD BUTTONWILLOW 21

41


Sample of CARB FOI Facility sites with >15 Km of error (San Joaquin Valley only)

Facility Name Address City Error (m) SHELL WESTERN E & P INC. P.O. BOX 11164 BAKERSFIELD 148,490 CHEVRON U S A INC WEST OF LOST HILLS GAS PLANT LOST HILLS 79,994 VINTAGE PRODUCTION CALIFORNIA LLC LIGHT OIL WESTERN 73,874 SENECA RESOURCES LIGHT OIL WESTERN 71,932 AERA ENERGY LLC MAIN CAMP ROAD BAKERSFIELD 66,831 PHILLIPS 66 PIPELINE LLC JUNCTION PUMP STATION, 14 COALINGA 65,786 MCKITTRICK LIMITED 4905 REWARD RD, HEAVY OIL WESTERN BAKERSFIELD 58,750 BERRY PETROLEUM COMPANY HEAVY OIL WESTERN BAKERSFIELD 48,263 KAWEAH RIVER ROCK CO. P.O. BOX 515 WOODLAKE 37,558 GRANITE CONSTRUCTION COMPANY ARVIN BAKERSFIELD 35,757 HILMAR CHEESE COMPANY 9001 NORTH LANDER AVE HILMAR 35,316 CRES INC DBA DINUBA ENERGY 6929 AVENUE 430 REEDLEY 29,644 CALIFORNIA CORRECTIONAL INST PO BOX 1031 TEHACHAPI 29,381 TTTI PANOCHE PUMP STATION SEC. 18-T 14S/R/12E FRESNO COUNTY 26,828 NAVAL AIR WEAPONS STATION GB 1 ADMINISTRATION CIRCLE CHINA LAKE 26,690 THREE BRAND CATTLE CO 34377 LERDO HWY BAKERSFIELD 23,014 EXXON MOBIL CORPORATION 18271 HWY. 33 MCKITTRICK 22,604 GOLDEN STATE VINTNERS 7409 W CENTRAL FRESNO 19,897 LIVE OAK LIMITED 7001 GRANITE ROAD BAKERSFIELD 19,815 WEST KERN WATER DISTRICT HWY 119 & CA AQUEDUCT TAFT 19,015 CHEVRON RIO BRAVO STATION ENOS LANE 2 MI SO OF STOCKDALE BAKERSFIELD 18,618 MACPHERSON OIL COMPANY HEAVY OIL CENTRAL BAKERSFIELD 16,747 NAVAL PETROLEUM RESERVE #1 ELK HILLS FIELD-GAS PLANT TUPMAN 16,239 NAVAL PETROLEUM RESERVE #1 ELK HILLS FIELD-PRDTN FACILITY TUPMAN 16,239 HAZEL H HEUSSER TRUST 41990 RADIO LN AUBERRY 16,012 CONOCO PHILLIPS PIPE LINE CO. 34960 AMADOR AVE COALINGA 15,128


Point Hazards used in Hazard Proximity Score Error rate - locational inaccuracy

Total >1000 ft >2000 ft >3000 ft >10,000 ft

CARB FOI 3157 624 (19.8%)

399 (12.6%)

313 (9.9%)

151 (4.8%)

Auto Paint and Body

3701 204 (5.5%)

140 (3.8%)

116 (3.1%)

63 (1.7%)

Gas stations 9682 10% random test <3%

Hazardous Waste

119 24 (20.2%)

13 (10.9%)

12 (10.1%)

5 (4.2%)

42


Validating and Correcting

Facilities of Interest (FOI) High location error rate

16 duplicate facilities

Some overlap with land uses we use in scoring:

• Area sources hazards used in EJSM hazard proximity scoring

o 41 Airports; 38 Refineries, 9 Gasoline Stations,

• Sensitive land uses

o 24 Colleges and Universities, 49 Hospitals, 1 Senior Residential Facility

Auto Paint and Body Shops Some of the largest locational errors

• Auto Netrix Recon Masters address in Laguna Nigel; reported location in Santa Clara

• Geocoding complicated as many are located in an off-street “stall” within a large parcel; geocoded address is at the street.

Gas Stations We did a 10% random test and found low

error rate (<3% were >1000 ft in error)

Most locations reported by host County; some located using Google Maps (n=1066)

About 1% of all records were duplicates (n=88/9770)

Hazardous Waste Facilities High location error rate

Two facilities (four sites) were duplicates

Four facilities could not be verified using GEP or web searches

Many sites are very large and not well represented by points; hazard proximity estimate may not reflect actual exposure at margins of the site.

IMPROVEMENTS IN SPATIAL METHODS FOR

HAZARD PROXIMITY (EJSM CATEGORY 1)

43

STEP 1: GIS Spatial Assessment (create CI poly layer [residential and sensitive land uses] with Census block info and calculate hazard proximity metrics)

STEP 2: SPSS Programming (data processing and generation of CI scores for tracts)

STEP 3: GIS Mapping of CI scores

EJSM ARCHITECTURE

Issues with Hazard Proximity

From the very beginning, communities expressed the need to include hazard proximity in the EJSM…

…but our original way of measuring hazard proximity was time consuming and limited, so we needed to find a simpler, faster, more flexible method.

METHODOLOGICAL IMPROVEMENTS

44

ORIGINAL BUFFER TOOL for identifying hazard proximity

Distance

Band

Weight Hazard

Count

1,000 ft. 1.0 2

2,000 ft. 0.5 8

3,000 ft. 0.1 6

Each CI poly receives a hazard proximity score, with the number of hazards

weighted using distance (“wedding cake approach”).

Distance-weighted hazard score =

( 1.0 x 2 ) + ( 0.5 x 8 ) + ( 0.1 x 6 )

1,000 ft.

band

2,000 ft.

band

3,000 ft.

band


NEW POINT DISTANCE TOOL for identifying hazard proximity

CI Poly

Centroid ID

Hazard

ID Distance

101 A 2800

101 B 1900

101 C 1700

NOTE: We only show the relationship between the CI Poly centroid and three hazards for sake of simplicity.

The Point Distance Tool measures the distance between the CI poly centroids and point hazards within a specified threshold.

The tool generates a table specifying the distance between the CI poly centroid and the point hazard.


45

CI Poly

Centroid ID Hazard ID Distance

101 A 2800

101 B 1900

101 C 1700

CI Poly

Centroid ID

Hazard

Count

0- 1000 ft

Hazard Count

1000-2000 ft

Hazard Count

2000-3000 ft

101 0 2 1

As with the Buffer Tool, the hazard count is weighted according to buffer distance.

B+C

How do we calculate the Hazard Proximity Scores for each CI Poly?

First sum the hazards that fall within 1000, 2000, and 3000 feet of each CI poly centroid

A



1000

ft

1000

ft

1000 ft

Buffer

1000 ft

Buffer

• This method works best with small, equant polygons

• Large, oddly

shaped polygons require additional processing



46

Solution: Cut large CI polys using a grid and run Point Distance for each centroid

1000 ft

1000 ft

1000 ft




CI Poly

CI Poly Centroid

3,000 ft.

3,000 ft.

… same distance doesn’t capture the same area.

Hazard

Hazard

Buffer Tool:

Point Distance Tool:

Wait! Don’t we need to account for the area of the CI poly when identifying

nearby hazards?

The Point Distance Tool measures the distance between two points, rather than the distance between a point and a polygon (like the Buffer Tool does).

47


To address this problem, we add the radius of a circle with the equivalent area to the CI poly to each of the distance bands in our SPSS programming. This way, we can identify how many hazards are within 1000, 2000, and 3000 feet of a CI poly.

CI Poly Centroid

3,000 ft.

Gap!

3,000 ft. Radius +

Hazard

Hazard CI Poly Centroid + Radius

Percent of tract population that lives in each block

Hazard Proximity Counts for each block

calculated in same way as Buffer Method

4

4

5

2

Tract

Block

Hazard Proximity Score for tract = (4 x .40) + (5 x .10) + (2 x .20) + (4 x .30) = 4

X 40%

30%

10%

20%

= 4

Hazard Proximity Score for tract


Final step: aggregating hazard counts to block to tract using population weights (in SPSS)

48

95

96

49

Difference in Cumulative Impact Score, Los Angeles County By Census Tract - Original vs. PointDistance

Difference in Cumulative

Impact Score using

Original vs.

PointDistance Method

-1

0

1

2

3

4

Score

Decreased

Score

Increased

97

COMPARING EJSM TO OTHER SCREENING METHODS

50

COMPARISON OF EJSM, CES, AND CEVA

Data Inputs and Metrics

Lots of overlap but important differences remain—examples: Only EJSM considers land use (sensitive and hazardous)

All three methods consider hazard proximity but in different ways

NATA: estimated cancer risk, respiratory hazard and/or diesel PM2.5

RSEI hazard-weighted emissions versus TRI site location

% living below federal poverty line versus 200% federal poverty line

DIFFERENCES AMONG SCREENING METHODS

Hazard Proximity Metrics – Sensitive Land Uses

Indicators EJSM CEVA CES Childcare facilities X

Healthcare facilities X X

Schools X

Urban Parks Playgrounds X Senior Residential X

51


Hazard Proximity Metrics – Polluting Facilities/Land Uses Indicators EJSM CEVA CES

CARB Facilities of Interest (FOI) (air toxics and GHG emissions facilities )

X

Industry-wide facilities (auto paint/body, gas stations)

X

Hazardous/solid waste facilities, cleanup sites X X Railroads X X Ports X X Refineries X X Intermodal Distribution Facilities X

Traffic Density X X TRI facilities X Chrome plating facilities (FOI) X X Cleanup Sites (EnviroStor) X Solid Waste (FOI) X X Groundwater threats from leaking underground storage sites and cleanups (GeoTracker)

X

Impaired Water Bodies X


Health Risk and Exposure Metrics

Indicators EJSM CEVA CES TRI or RSEI X X

National Air Toxics Assessment - Cancer Risk

(with diesel PM) X X

National Air Toxics Assessment – Respiratory

Hazard X

PM2.5 (Interpolated from CARB monitors) X X

Ozone (Interpolated from CARB monitors) X X

Diesel PM Emissions X

Pesticide use X X X

Water quality – Contaminants X X

Water quality – Source Vulnerability X

52


Social and Health Vulnerability Metrics

Indicators EJSM CEVA CES Race/ethnicity X X

Poverty level X X X Educational attainment X X X

Age (<5 and >64) X X X

Linguistic isolation X X X

Unemployment X

% Renters X Median house value X

Voter participation X

% Low Birth Weight and/or SGA X X X

Asthma hospitalization X X

Life expectancy X


Climate Vulnerability Metrics

Indicators EJSM CEVA CES Tree Canopy X

Impervious Surfaces X Projected Temperature and Temperature Changes X

Project Increase in Warm Nights X

% Elderly Living Alone X

% Car Ownership X

53

MAPS AND GEOGRAPHIC ANALYSIS

Merging of information reported at different levels of geography Spatial unit for both analysis and scores/mapping

CES (tracts) and CEVA (block groups) use same census polygons throughout

EJSM – uses smallest spatial unit available for each data type

• Land Use: tax parcels, municipal land use or zoning data, interpreted aerial imagery

• Population weighting of hazard proximity – census block population

• All eventually aggregated to census tracts for continued analysis and scoring

Resulting map pattern:

These differences affect the map pattern, so variations in pattern are, in part, controlled by the method, not just the data

Tracts and block groups vary in size regionally, so map colors maps may show misleading pattern

• Rural and sparsely populated areas are larger; can dominate the map

• EJSM uses “land use masks” to focus scores on populated areas

• How the map colors are defined also influences map pattern

oExample: CES maps show 20 colors, each with same number of tracts

oEJSM: scores follow “bell-shaped” curve, so fewer tracts with highest scores

METRICS CATEGORIES AND SCORING

Differences in: Number of indicator metrics used

How indicators are grouped together for scoring

Results in different implicit “weighting” of some metrics

Different range of scores among methods: EJSM:

Linear ranking within each category

These are summed and re-ranked.

Open-ended to accommodate additional indicators (3-15).

Preferred scoring is regional

CES:

Indicator categories multiplied to yield a continuous, open-ended score

Statewide scoring only

CES scores are grouped into percentiles (1-20), so same number of tracts for each score value

CEVA:

3x3 scoring matrix (1-9) with separate axes for impact and vulnerability

Scores have been applied to selected regions

54

CALENVIROSCREEN 2.0 VS. EJSM 3.0 IN MAPS

CES SCORES – 13 CATEGORIES (EQUAL)

55

EJSM STATEWIDE CI SCORES

1 2 3 4 5 6 7 8 9 10 11 12 13

DISTRIBUTION OF CES SCORES TO BELL CURVE

1 2 3 4 5 6 7 8 9 10 11 12 13

56

CES SCORES – 13 CATEGORIES (BELL CURVE)

EJSM STATEWIDE CI SCORES

57

CES SCORES–13 CATEGORIES (EQUAL): ABAG

EJSM STATEWIDE CI SCORES: ABAG

58

CES SCORES–13 CATEGORIES (BELL CURVE): ABAG

EJSM REGIONAL CI SCORES: ABAG

59

FEEDBACK

DISCUSSION QUESTIONS

1) Do the EJSM 3.0 results resonate with what you see on the ground?

2) What are other policy and scientific applications using such spatial screening methods that go beyond the distribution of cap-and-trade revenue?

a) How can you use this in your current and future work?

b) What’s the most useful way to put the EJSM out in the world? Should EJSM have an online presence?

3) What kinds of additional indicators and data would you like to see integrated into the method? How important or not important is the hazard proximity layer?

4) How would you like to see the water metrics integrated into the total CI score? Do you think there should be a separate water layer or should it be folded into one of the existing layers, such as Health Risk and Exposure?

Documents

Environmental Justice Screening Method (EJSM)...1 January 16, 2015 Rachel Morello-Frosch, Manuel Pastor, James Sadd, Madeline Wander Environmental Justice Screening Method (EJSM) Overview,