dredge data evaluation trc meeting march 15 2005 don yee
TRANSCRIPT
Dredge Data Evaluation
TRC Meeting March 15 2005Don Yee
Objectives Compile dredge testing data for San Francisco
Bay Compare collection and analytical methods Compare analytical results Test hypotheses on SF Bay sediment
characteristics
Compiled Data State effort to develop sediment quality
guidelines occurred concurrently, including compilation of selected dredged material testing data
Draft database released November 2004 ~40 dredging studies in SF Bay 11 monitoring studies (includes Hunter’s Point,
Alameda)
Spatial Distribution
Bay Segment Monitor DredgeCarquinez 20 16Central Bay 381 332Lower South Bay 41NA 37 73Pacific 3Rivers 11San Pablo Bay 89 6South Bay 57 20Suisun Bay 55 41
Central Bay most stations
Nondetect Issues MDLs sufficient for most trace elements…
0%
5%
10%
15%
20%
25%
30%
35%
Al
Sb
As
Ba
Be
Cd Cr
Co
Cu
Fe
Pb
Mg
Mn
Hg
Me
Hg
Mo Ni
Se
Ag Sr
Th
Sn Ti V Zn
MON
DRE
Nondetect Issues Many NDs for organics (esp. dredge data)
PAH NDs
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0% 20% 40% 60% 80% 100%
MON %ND
DR
E %
ND
PCB NDs
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0% 20% 40% 60% 80% 100%
MON %ND
DR
E %
ND
Sensitivity Issues Dredge data MDLs usually slightly higher
PAH MDLs
0
1
2
3
4
5
6
7
8
9
10
0 2 4 6 8 10
MON MDL ug/kg
DR
E M
DL
ug/k
g
PCB MDLs
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
MON MDL ug/kg
DR
E M
DL
ug/k
g
Reported Organics Subset of common PAHs reported Reported PCBs as Aroclors or congeners,
study specific “Total PCBs” differ (ND=0 case)
Dredge data most sum Aroclors or sum congeners = 0
Monitoring data avg Aro/cong = 1.7
Comparing Data Large % of organics NDs challenging for
comparison ND assumptions have large influence
particularly on sums Non-parametric statistics more resistant to
NDs, but MDL/RL differences and non reporting among different studies can bias outcome
Hypotheses: SeasonalitySeasonality in historic RMP sediment data (5cm
surface grabs) found not significant, therefore… Seasonality in dredge cores even less likely
Cores composited for analysis Relatively infrequent revisits in dredging may leave
gaps
Test: SeasonalityInclude areas dredged multiple times, perhaps covering
different times of year Relatively infrequent revisits in dredging = poor seasonal
coverage Deeper coring and compositing may dilute seasonal signal
Wilcoxon and Tukey HSD test Comparing Dec-May “wet” vs Jun-Nov “dry” Central Bay stations alone
Result: Seasonality Wet vs dry season
As, Cd, Cu, Mn, Ni significantly (p <0.05) different, with wet > dry
Hg, Se not significant Consistency among elements and types of test
Seasonality (Alternative) Oakland 38 and 42 ft projects large portion of
dry data (50 ft project split in both seasons) Deeper sediments include relatively
uncontaminated Bay muds Excluding data still showed significant wet and
dry differences (both tests, same metals)
Hypotheses: Interannual Trend Negligible interannual trends in sediment
concentrations RMP surface sediments show few/ no significant
interannual trends for most contaminants Interannual dredge trends likely confounded by
differences in sampled locations & signal diluted by compositing.
Test: Interannual TrendInclude dredging data over multiple years Subsets
Central Bay stations Group by collection year
Nonparametric (Kruskal Wallis) and parametric tests (Tukey HSD)
Result: Interannual Trend Significant (p <0.05) differences among years
for various metals. No consistency among metals for years with
significant differences Compare enough categories, and “significant”
differences will be found
Alternative: Interannual Trends Use smaller scale: Benicia Harbor only
Small well defined area, expect less variation 3-4 samples per year for 3 years
Significant differences 1996 > 1997 or 2000: As, Cd, Cr, Hg (p < 0.05) 1996 frequently nigher than other stations
Benicia Annual Differences
Hypotheses: Depth Integration Sediment contaminant concentrations at depth
are not significantly different from surface concentrations
Deep dredging will encounter clean Bay sediments and average lower contaminant concentration than surface samples
Test: Depth Integration Subset to focus on area around Oakland
Dredge data in port dominated by deepening project samples
Latitude & longitude from all Oakland dredging projects
Monitoring data for surface sediments in same box Wilcoxon and Tukey HSD test
Result: Depth Integration Concentrations for nearly all metals Oakland
monitoring > dredge data As, Cd, Cu, Pb, Hg, Ni, Se, Ag, Zn statistically
significant difference Cr only metal higher in dredge sediment Generally consistent among metals Significant differences on As, Cd, Se, Zn even
removing BPTCP, LEMP, Alameda NAS data
Oakland Depth Integration
Result: Depth Integration Part 2 Examine differences in dredge vs monitoring in
Suisun Bay. Hg statistically significantly higherin monitoring
(surface sediment) data Cr only metal higher in dredge sediment Differences hold even dropping LEMP, BPTCP
data
Suisun Depth Integration
Conclusions Dredge data analytical methods sufficient for most
trace elements Organics analyses usable only for PAHs due to
extensive ND results Dredge data may show seasonality in Central Bay
Larger number of samples provides power Similar trends among metals evidence of real effect Significance holds even removing obviously biasing data
sets (e.g. Oakland deepening projects)
Conclusions (cont’d) Segment scale dredge data show no
interannual trends (Central Bay) Significant results found in many comparisons, but
inconsistent among trace elements Significance may be artifact of numerous (year)
categories Smaller scale dredge data may show trends
Narrower sampling focus reduces lateral and vertical variability
Dredge Data Useful (with care) Awareness of collection, analytical, and
reporting conventions Selecting the right subset to reduce biases Inconsistent “significance” may not be
significant