Florida Department of Environmental Protection Florida Mercury TMDL FloridaDepartment Department of of Florida EnvironmentalProtection Protection Environmental Florida’s TMDL Program and the TMDL Development Process Jan Mandrup-Poulsen, Administrator Watershed Evaluation and TMDL Section Phone.
Download ReportTranscript Florida Department of Environmental Protection Florida Mercury TMDL FloridaDepartment Department of of Florida EnvironmentalProtection Protection Environmental Florida’s TMDL Program and the TMDL Development Process Jan Mandrup-Poulsen, Administrator Watershed Evaluation and TMDL Section Phone.
Florida Department of Environmental Protection Florida Mercury TMDL FloridaDepartment Department of of Florida EnvironmentalProtection Protection Environmental Florida’s TMDL Program and the TMDL Development Process Jan Mandrup-Poulsen, Administrator Watershed Evaluation and TMDL Section Phone (850) 245-8448 June, 2012 Summary of Presentation • Purpose of Meeting • TMDL Overview and 303(d) Requirements • Florida’s TMDL Legislation Administrative Requirements • Please Provide Written Comments By June 30th! • [email protected] Page 3 What Does TMDL Stand For? Total Maximum Daily Load • Establishes maximum amount of a pollutant that a water body can assimilate without causing exceedances of water quality standards (water quality criteria & designated uses) Page 4 TMDL Process: Overall Objective • Identify and quantify all point and nonpoint source (NPS) loadings to a water to determine the assimilative capacity of that waterbody for each pollutant impairing water quality • Often use computer modeling to estimate NPSs and establish assimilative capacity • TMDL is then allocated to all sources • Includes a Margin of Safety (MOS) • We generally plan to follow recommendations in 2001 Allocation TAC Report, which was submitted to Legislature and Governor Page 5 303(d) Listing Requirements • Section 303(d) of the Federal CWA requires states to: • • • • submit lists of waters that do not meet applicable water quality standards (designated uses) (“impaired waters”), identify pollutant causing or expected to cause impairment establish/implement TMDLs for these waters on a prioritized schedule new list and schedule required every 2 years Page 6 303(d) Listed Waters for Mercury • The 1998 303(d) list had 102 fresh and estuarine waterbodies on it (63 streams, 13 lakes, and 26 estuarine segments). • As a result of added sampling effort, Florida’s current 303(d) list has 249 stream segments, 127 lakes, 504 estuarine segments, and 221 coastal water segments (through 2011 updates). Page 7 Page 8 TMDL Development Schedule • EPA sued by EarthJustice in Florida on April 1998 and settled with detailed schedule for TMDL development • DEP was NOT a party to suit, but participated in some settlement discussions • EPA agreed to 13-year schedule, mainly based on DEP’s basin management cycle, but with EPA as backstop • Every listed water was given a specific due date (year), with “high priority” waters due 1st cycle and “low priority” waters due 2nd cycle Page 9 TMDL Development Schedule (continued) Waters that will be discussed today were on the 1998 303(d) list, and have been scheduled for TMDL development to aid the local restoration efforts. Plan to submit to EPA in September 2012 for approval, but must address new requirements Not all TMDLs have to be for Consent Decree waters Page 10 Page 11 Rulemaking Schedule • A “Notice of Rule Development” for the statewide mercury TMDL was published on April 27th, 2012. • E-mails were sent to interested parties list, and noticed on DEP website under “Official Notices” • http://sharepoint.dep.state.fl.us/PublicNotices/default.aspx • Can register for the notices on DEP website, or sign up today for our notices Page 12 Next Steps • Public Comment Period Runs to 6/30/12 • Next FAW Notice for Public Workshops on Revised Statewide Mercury TMDL Report Published on July 6th • Public Workshops Held July 23-27, 2012 • Public Comment Period Ends 8/27/12 • Notice of Proposed Rule in FAW-9/21/12 Page 13 QUESTIONS? • Other Contact Information: Jan Mandrup-Poulsen Environmental Administrator Watershed Evaluation & TMDL Section Florida Dept of Environmental Protection 850/245-8448 [email protected] Page 14 FloridaDepartment Department of of Florida EnvironmentalProtection Protection Environmental Florida’s Mercury TMDL Efforts Greg White, D.Sc. Technical Manager Mercury TMDL [email protected] (850) 245-8347 Outline of Presentation • Project Purpose • Overview of Scientific Approach • Atmospheric Dry & Wet Deposition Monitoring • Monitoring Summary • Select Monitoring Results • Aquatic System Monitoring • Modeling Overview of Results • Atmospheric modeling • Emissions updates • Atmospheric modeling: SMOKE & CMAQ Tagging Overview • Example atmospheric model outputs for Aquatic modeling • Aquatic Statistical Assessments to identify relationships • Examples from Inferential Statistical Analyses Page 16 Mercury TMDL Project Purpose • This effort to understand mercury cycling from atmospheric emissions through aquatic systems to bioaccumulation in fish tissue • Elevated levels of mercury , specifically as identified in fish tissue, are a health concern • In Florida mercury first raised as an issue due to release from SAPP Battery Plant, in the panhandle, 1983 & Elevated levels in the Everglades in 1989 • This resulted in wider sampling across the state • identified broader issue of bioaccumulation of mercury in select fish species • Other efforts have identified that that 95+% of mercury comes from atmospheric emissions Page 17 Overall Approach of TMDL Project Dry mercury deposition Quantify Mercury Loading to Florida MONITORING Wet mercury deposition Quantify Relationships between Mercury Sources and Receptors Aquatic System Monitoring Water Quality Sediment Quality Fish Tissue THg Develop Predictive Atmospheric Models to Quantity Loads MODELING Develop Aquatic Inferential Models for Lakes, Streams & Everglades Conduct TMDL Analysis: Determine needed Fish THg and TARGET SETTING &Tissue REDUCTION reductions to abate Fish Consumption (this will be presented subsequent to this project overview) impaired conditions Page 18 Mercury Complexity Photolysis causes shifts in chemistries through air and clouds Hg0 • Elemental mercury • 95+% from atmospheric emissions • May be identified in many forms: Their individual chemistries are different { Hg0 HgII HgP RGM GEM GOM O3 OHH2O2 CL-, BrCl2, Br2 HCl Elemental mercury Divalent mercury Hg2+ RGM HgP Particulate mercury Reactive Gaseous Mercury, volatilized mercury Hg2+ Gaseous Elemental Mercury, volatilized mercury Hg0 Gaseous Oxidized Mercury same as RGM SO2OHHO2 Λ (light) HgII Hg0 HgP • Mercury is described by 40 gas phase species that have more than 80 reactions & 30 aqueous species with over 100 reactions • Can be in the atmosphere for 2 years, and cycle deposition-thru-reemissions from 100s to 1,000s of years, depending upon how defined • Atmospheric modeling includes 51 vertical layers to address this complexity Page 19 Mercury Complexity • Elemental mercury [Hg(0), HgII or HgP, RGM] is transformed under certain specific environmental conditions to organic mercury (methylmercury) Hg MeHg • Methylating conditions • Reduction environments, e.g., wetlands, anoxic sediments, mesic soils • Proper bacterial populations, e.g., Sulfur Reducing Bacteria (SRB), periphyton communities, others • High DOC, low pH, S at suitable levels & composition • Only a small amount of Hg is transformed MeHg Page 20 Mercury Issue in Florida Summary (Modified from Pirrone et al. 2010) Page 21 Atmospheric mercury Page 22 Atmospheric Mercury Emissions Source: Pirrone, Nicola, S Cinnirella, Xinbin Feng, R B Finkelman, Hans R Friedli, Joy J Leaner, Robert P Mason, A B Mukherjee, G B Stracher, David G Streets, K Telmer “Global Mercury Emissions to the Atmosphere from Anthropogenic and Natural Sources.” Atmospheric Chemistry and Physics 10 (2010) : 5951-5964. Page 23 Major Florida Atmospheric Sources •Power Utilities Coal Oil /Nat. gas •Cement Production •Waste to Energy Facilities •Also mobile sources • Cars, trucks, trains • Shipping Page 24 Assessing atmospheric mercury in Florida Supersites and Intensive Sites Supersites, Intensives, and Wet Deposition Only Site Locations Page 25 Monitoring sample collections • Supersites • Tekran suite • Continuous monitoring • UMAQL’s ASPS (automated wet dep collection) • Intensives • Full wet chemistry • Surrogate surfaces Artificial turf surface Dry deposition plate Teflon through-fall bottle Page 26 Automated Sequential Precipitation Samplers (ASPS) Monitoring constituents Hg Speciation: bi-hourly elemental gaseous Hg, reactive gaseous Hg, particulate Hg Criteria gases: O3, SO2, CO, NO, NOy Fine and coarse particulate matter (PM2.5 and PM10): mass and trace metals (Mg, Al, P, S, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, La, Ce, Sm, Pb) Meteorology: wind speed, wind direction, temperature, relative humidity, barometric pressure, solar radiation, precipitation, surface wetness Precipitation samples tested for above, as well as nutrients and Hg isotopes Page 27 Monitoring chemistry analyses • Continuous monitoring by ARA, inc. • Wet Deposition chemistries • • • • Wet chemistry suite by UMAQL Filter mass by DEP Central Laboratory Isotopic analyses by Dr. J. Blum at UM ICP-MS support from US EPA • Suite: Aluminum (Al 27), Antimony (Sb 121), Arsenic (As 75), Barium (Ba 137), Beryllium (Be 9), Cadmium (Cd 111), Calcium (Ca 44), Chromium (Cr 52), Cobalt (Co 59), Copper (Cu 63), Iron (Fe 57), Lead (Pb 208), Lithium (Li 7), Magnesium (Mg 24), Manganese (Mn 55), Molybdenum (Mo 95), Nickel (Ni 60), Phosphorus (P 31), Potassium (K 39), Rubidium (Rb 85), Selenium (Se 77), Selenium (Se 78), Silicon (Si 28), Silver (Ag 107), Sodium (Na 23), Strontium (Sr 88), Thallium (Tl 205), Uranium (U 238), Vanadium (V 51), Zinc (Zn 66) Page 28 Expanded effort beyond supersites • Atmospheric mercury is known to have varying and different chemistry over marine waters as compared to over land • Marine layer atmospheric chemistry • Mixing • Halogen chemicals (Cl-, Br-, others) • Marine emissions & re-emissions Page 29 Added efforts to assess conditions on Gulf of Mexico Captain Matt White Hg Speciation: elemental gaseous Hg, reactive gaseous Hg, particulate Hg Criteria gases: O3, SO2, CO, NO, NOy Page 30 Daily mean Hg0 at Supersites (July 2009-June 2010) 7/09 8/09 9/09 10/09 11/09 12/09 1/10 2/10 3/10 4/10 5/10 6/10 Tekran Measurements Regional behaviors seen in peaks Seasonal patterns across sites elevated levels in summer months North Florida OLF North Florida South Florida South Florida TPA JKS DVE From ARA, Inc. First Light Software Page 31 Daily mean RGM at Supersites 7/09 8/09 9/09 10/09 11/09 12/09 1/10 2/10 3/10 4/10 5/10 6/10 Remember RGM ≡ Reactive Gaseous Mercury Very short-lived Highly Reactive in the atmosphere, with plants, and water systems Weak South Florida Trend Weak South Florida Trend North Florida North Florida South Florida South Florida OLF TPA JKS DVE From ARA, Inc. First Light Software Page 32 Daily mean HgP2.5 at Supersites 7/09 8/09 9/09 10/09 11/09 12/09 1/10 2/10 3/10 4/10 5/10 6/10 Remember HgP2.5 ≡ is particulate mercury, charge/ionic mercury bound to particles Similar trend across all sites, which may be indicative of global sources HgP 2.5 means very small particles, so easily transported in winds, as well as small size reactive with sunlight when in the air or deposited So these little particles enough mass to enter and stick inside leaves, needles, small pieces of organic matter OLF TPA JKS DVE From ARA, Inc. First Light Software Page 33 Speciated Hg by Quarter ‘09 Supersites RGM by Quarter Hg0 by Quarter OLF JKS TPA OLF DVE JKS TPA DVE 8.0 1.6 1.4 6.0 1.2 4.0 1 2.0 0.8 0.0 0.6 2009 Q3 2009 Q4 2010 Q1 2009 Q3 2010 Q2 HgPPM2.5 by Quarter OLF JKS TPA DVE 6 5 4 3 2 1 0 2009 Q3 2009 Q4 2010 Q1 Page 34 2010 Q2 2009 Q4 2010 Q1 2010 Q2 Wet Deposition Monitoring Davie (DVE) Hg - Monthly VWM & Wet Dep. S T R I P P I N G M E R C U R Y Effects of summer storms Summer Thunder Storms Page 37 Wet Deposition Monitoring Everglades (ENP) Hg - Monthly VWM & Wet Dep. Page 38 Hg Wet Deposition Everglades National Park (ENP) Daily measures Page 39 Intensives Davie Hg Wet Deposition July 4 – August 4, 2009 7.0 6.0 Hg wet deposition (ug/m2) 5.0 4.0 3.0 2.0 1.0 0.0 OKB (D2) FKE (D5) Page 40 DVE EHP (D1) MIA (D3) ENP KYB (D4) Examples of Monitoring data Hg Wet Deposition (ug/m2) –Intensives Jacksonville 8/2010 Pensacola 8/2010 Tampa 8/2009 Davie 8/2009 Page 41 Wet & Dry Deposition across sites 2010 Intensives 2010 Intensives 2009 Intensives 2009 Intensives Page 42 Evaluate Atmospheric Modeling (CMAQ) • Provide meteorological data to evaluate WRF • Provide data to assess mercury wet and dry deposition loadings • Mercury Emission Source Contributions • Provide data to assess emission source contributions, why such a full suite of analyses are performed • Tracer species, receptor modeling (USEPA: PMF, Unmix) • Provide measurement data to evaluate and compare with deterministic models (CMAQ) for: • Mercury Dry Deposition Loadings • Mercury Wet Deposition Loadings Page 43 Monitoring mercury in aquatic systems Page 44 Aquatic System Monitoring • Site Selection • Stratified random sampling • 3 constituents 5 x 5 x 5 array • Sample 133 lakes and 131 streams to develop the distribution curve of Hg concentrations in both types of resources for the entire state of Florida • Fish sampling: 12 fish for tissue sampling, priority of selection Largemouth Bass, supplemented by Spotted Sunfish & Spotted bass • Spotted Sunfish allow evaluation of many waters that do not provide LMB habitat, especially increasing the variability of streams that could be assessed • Water quality suite collected along with fish • Lake sediments, added during project , not collected at same time Page 45 Site selection Stratified Random Sampling • Data from FDEP Status and Monitoring Network (SMN)-Regional and statewide scaling to full population of waterbodies • Eliminate non-freshwater systems (specific conductance > 5,000 μmho/cm or Cl > 1,500 mg/L) • Log-transform variables of interest in setting cell values • Establish 5 equidistant (ln-values) sampling intervals between the 2.5 and 97.5% cumulative distribution limits. This improves the sampling characteristics at both ends of the distribution. • 5 x 5 x 5 sampling matrix yields 125 possible, unique parameter combinations = Cells • Lakes and streams assigned to appropriate cells based upon historic water quality measures, and candidate systems for sampling then selected at random. Page 46 Lake variables for site selection Variable Lakes Streams - Rivers pH pH Chlorophyll a NO3- Color Color Acid-base status Trophic state Methylation affect Level/Factor Parameter Groups or Cells pH Ln Chlorophyll a Ln Color 1 ≤ 5.130 ≤ 0.821 ≤ 2.486 2 > 5.130– 6.123 > 0.821 – 1.804 > 2.486 – 3.362 3 > 6.123 -7.115 > 1.804 – 2.788 > 3.362 – 4.239 4 > 7.115 – 8.108 > 2.788 – 3.771 > 4.239 – 5.115 5 > 8.108 > 3.771 > 5.115 Page 47 Stream variables for site selection Variable Lakes Streams - Rivers pH pH Chlorophyll a NO3- Color Color Acid-base status Trophic state Methylation Parameter Groups or Cells Level/Factor pH Ln NO3- Ln Color 1 ≤ 4.2089 ≤ -4.162 ≤ 2.66908 2 > 4.2089 -5.2794 > -4.162 to -2.803 > 2.669 - 3.729 3 > 5.2794 -6.2779 > -2.803 to -1.444 > 3.729 - 4.788 4 > 6.2779 - 7.2764 > -1.444 to -0.085 > 4.788 - 5.848 5 > 7.2796 > -0.085 > 5.848 Page 48 Field sampling for fish and WQ samples Tyvek suits and gloves part of clean Sampling requirements Page 50 Fish Tissue Summary Results Mean and Range of fish THg, TL, and Age Spotted lower THg as smaller, shorter lived, and and association lower trophic level Streams showSunfish higher show THg as expected with greater Hg loading with wetlands THg (ppm) Species Age (yr) Mean Range Mean Range ALL 2,810 0.46 0.02 - 2.9 327 (13") 104 (4") - 580 (23") 3.2 0 - 13 LK 1,650 0.40 0.02 – 1.7 337 (13”) 120 (5”) – 565 (22”) 3.2 0 – 12 STRM 1,160 0.54 0.03 – 2.9 315 (12”) 104 (4”) – 580 (23”) 3.0 0 - 13 0.26 0.01 - 0.99 125 (5") 56 (2") - 197 (8") LMB SPSU N Total Length (mm) 925 v v v v Mean Range Median LMB and SPSU THg for LKS and Streams LKS Sample size LMB 4-29 per site; mean = 13 1.25 STRMS Sample size 120 sites with LMB 1-20 per site; mean = 10 82 sites with SPSU 1-18 per site; mean = 11 THg (ppm) wet weight 1.00 0.75 0.50 0.25 0.00 LK LMB STRM LMB STRM SPSU Box: 25 and 75%, Points: 5 and 95% Page 51 Statewide distributions of THg by fish species Statewide THg (mg/kg) 2001-11 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Florida DOH guidelines for general population Florida DOH guidelines for at risk population • Distributions are part of the evidence for defining appropriate target levels of Hg loading and when coupled with the atmospheric modeling and monitoring for load assessment, as well as direct surface water loads, help determine requisite reductions from in-state sources, US sources, and global sources Page 52 Cumulative frequency of THg fish samples 100% 90% LKS STRMS 80% Cum Prob 70% 60% 50% 40% 30% 20% 10% 0% Mean Site THg (ppm) wet weight Page 53 Prob. 100 95 90 75 50 46 20 19 10 1 N THg Concentration (ppm) LK STRM 1.22 1.82 0.96 1.07 0.73 0.99 0.57 0.76 0.34 0.49 0.30 0.18 0.31 0.30 0.11 0.19 0.05 0.11 166 88 Mercury in Fish Tissue Lakes Streams Minimum = 0.03 (mg/kg) Maximum = 1.32 (mg/kg) Minimum = 0.04 (mg/kg) Maximum = 1.61 (mg/kg) Shows distribution of PCA3 (anthropogenic disturbance/urban runoff) Page 54 Water Column Total Mercury – Methylmercury Page 55 Lake Sediments Total Mercury –Methylmercury Basic relationship of THg to MeHg appears consistent Note different scales Same # samples each group Page 56 Understanding atmospheric mercury through modeling Global Chemical Model (ECHMERIT) Regional Meteorological Model (WRF) Initial and Boundary Conditions North American Domain (WRF Nested Domain) Regional Thermodynamic Local Thermodynamic and Wind Fields and Wind Fields Sparse Matrix Operator Kernel Emissions (SMOKE) Model CMAQ Grids • North America Regional Chemical Transport Model • Southeastern U.S. • Florida (CMAQ) Page 57 Global modeling - ECHMERIT Example of model output from July 10, 2009 Model inputs Global weather Global emissions Atm. Chemistry 80km cells Efforts led by: IIA-CNR in Italy Page 58 Grids for CMAQ & WRF The 12km scale is the finest resolution modeling done prior, and this prior work was done at scales smaller than Statewide scales; here DEP committed to do this at a regional scale. 36 km (~22.4 mi.) North American Grid Covers Gulf of Mexico 12 km (~7.5 mi.) & its coastal watershed Regional Grid 4 km (~2.5 mi.) Florida Grid 4km grid Pinellas & Hillsborough Counties Efforts led by: University of Michigan Air Quality Laboratory Page 59 Emissions inventory updating of 2005 NEI • Updating emissions inventories • Global led by CNR-IIA, Dr. Nicola Pirrone, Italy • North American grid enhancing data for Mexico, Canada and NEI 2005 from US EPA led by UMAQL • Southeastern USA enhanced by emission modeling provided by Southern Company • Florida updated by DEP Div. of Air Resource Management (DARM) and UMAQL • Emissions get processed through SMOKE, first tagging (Sparse Matrix Operator Kernel Emissions) Page 60 Tagging of Emissions in SMOKE and Hg in CMAQ Florida sources: Other sources: • Florida coal-fired electricity • S.E. states tagged individually: generation (FLEC) GA, AL, MS, LA, TX • Florida oil-fired electricity • All remaining US sources (US) generation (FLEO) • Other anthropogenic Hg in • Florida waste-to-energy North American Grid: Canada, incinerators (FLWE) Mexico (MISC) • Florida cement kilns (FLCM) • Soil emissions, re-emissions, incl. atmosphere-soil exchange • All remaining Florida (SOIL) anthropogenic emissions (FL) • Global background input to CMAQ (BKRD) Page 61 Additions to CMAQ for tagging • 42 species added [14 tags, 3 Hg species per tag - Hg(0), Hg(2+), Hg(P) ] in both CMAQ and in SMOKE (emissions) • Explicit emissions added in CMAQ code for tagged Hg • Explicit solutions for gas-phase, aqueous chemistry consistent with ‘parent’ species, i.e., track Hg species • Dry deposition/bidirectional , i.e., model hands re-emissions based upon atmospheric & meteorological conditions • Only emissions and initial/boundary conditions differ Hg global to tagged Hg in CMAQ • Tagging showed consistency in mass balance outputs Page 62 Linearity testing of CMAQ Tagging • Proof of reductions • linearity Reducedallows deposition 100% and 50% for tags to be calculated directly from the base model year • Emission reductions in any tagged category can be calculated arithmetically from the base model year output • Thus can apply such as MACT or MATS projected emission reductions Page 63 Evaluation of WRF Simulations • Examples: Temperature and precipitation distribution/magnitude Page 64 Example CMAQ Evaluation against measured wet deposition data Davie Near by CMAQ over estimates ENP CMAQ annual wet dep. matches Continuous Wet Deposition Sites Monitoring measures Very close match, with Pensacola slight overestimate, which CMAQ under estimates annual wet may dep. be the model handing dew as being rain Everglades Page 65 Inferential Dry Deposition • Ra (aerodynamic resistance) – a function of turbulence/atmospheric stability • Rb (quasi-laminar boundary layer resistance) – a function of the diffusivity of the gaseous chemical species being transported • Rc (canopy resistance) – a function of the chemical properties of the gaseous species and the physiochemical properties of the deposition surface Page 66 Example CMAQ Evaluation against measured dry deposition data Inferential calculation Measurements of ambient chemical concentrations of the gaseous and particulate species are combined with species-specific deposition velocities to “infer” rates of pollutant dry-deposition to the surface of interest. Dry deposition is not matching up as well as wet deposition Direct measurements of total mercury dry-depositional flux obtained using a surrogate surfaces during intensives are being compared with CMAQ for further evaluation. Page 67 CMAQ Hg deposition FL vs. non-FL sources All continuous sampling locations: Stacked histograms Page 68 CMAQ SE Regional Tags to Lakes • Southeastern states relative loads • You can identify regions with larger percent of Florida sources moving south down the peninsula aquatic Page 69 CMAQ SE Regional Tags to Streams • Southeastern states relative loads • You can identify regions with larger percent of Florida sources in the peninsula, and greaterr external contributions in the panhandle aquatic Page 70 CMAQ Global, USA, SE Tags • These maps shows the overwhelming influence of global sources to Florida aquati c aquat ic Page 71 Source-Receptor Modeling • Provides the ability to identify and apportion the impact of major source types at a given location (receptor site) with lack of source information. • Determine groups of chemical elements that vary statistically together within a given dataset, so as to identify common groupings of chemicals,, i.e., emission signatures (analogous to satellite imagery signatures) • Using public domain USEPA models PMF and UNMIX Page 72 Example source relationships Coal Combustion: Sulfur (S), Selenium (Se) Oil Combustion: Vanadium (V), Nickel (Ni) Motor Vehicles: high Organic Carbon (OC), some Elemental Carbon (EC), and few metals Incineration: S, OC, EC, and metals Crustal/Soil: Aluminum (Al), Iron (Fe), and rare earth metals (Cerium:Ce, Lanthanum: La, Samarium: Sm) Industrial: High amounts of heavy metals (Lead: Pb, Zinc: Zn, Chromium: Cr), some OC & EC. (natural emission) Trace signatures can have in excess of 18 constituents USEPA maintains a library of trace signatures Page 73 Examples of Variability for S-R Modeling Nickel associated with oil combustion Page 74 Example PCA output from PMF Dvonch et al. Use of elemental tracers to source apportion mercury in south Florida precipitation. Environmental Science & Technology, 33: 4522-4527 (1999). Page 75 CMAQ outputs for Aquatic Modeling • 133 Lakes • 130 Streams • Added 100 lake sites using SMN & Lange data Output THg, Hg(0), HgP, RGM at each location Page 76 Aquatic Modeling • Establish relationships between water quality constituents and HgT in fish tissue • Correlative analyses not deterministic modeling • Use normalized fish HgT to establish relationships • Allows to address variations LMB panhandle to ENP • Normalization of Spotted sunfish allow extension of fish tissue samples to waterbodies where LMB were limited • This allows a broader representation of surface waters, especially stream systems • Allows any reductions to be evaluated against a common end point, LMB-norm, which can be back related to regional measure or atmospheric model outputs Page 77 Example aquatic system relationships In cases were several data points • Water quality, sediment quality in lakes, andwere THg BDL, inferential relationships were made to are evaluated for relationships fully populate subsequent statistical tests such as Principal Component Analysis & Classification and Regression Trees Observed Ln-transformed Sediment Pb (mg/kg) Some relationships are readily observed directly from monitoring data Page 78 Predicted Ln-transformed Sediment Pb (mg/kg) Lake Model Sediment PCA Transformed Sulfate Alkalinity Chlorophyll Model Robustness and Validation 1. Homoscedasticity of residuals 2. Effects of residual heteroskadasticity 3. Multicollinearity between independent variables 4. Jackknife analysis for robustness of out-of-sample predictions 5. Comparison of robust regression 6. Comparison of neural net modeling Multiple Linear Regression Evaluate Outliers Cook’s D Model Residuals Generalized Linear Model Added variable component plots Model Robustness and Validation Jackknife analysis for robustness of out-of-sample predictions Multiple Linear Regression Fully Validated Model Page 79 Lake Model The sequence of importance of independent variable contributions to the overall variability in LMB Hg in the study lakes is: Alkalinity > chlorophyll a > urban runoff disturbance > THg > SO42(PCA component associated with mineral weathering & urban particulates) Page 80 Stream Model Development Water Chemistry PCA/MLR Outlier Analysis PCA/MLR Model for “Excess” Methyl Hg Multiple Linear Regression (MLR) Neural Network Modeling Partition Analysis Revised CART Combined Partition/MLR Model Fully Validated Model Page 81 Validation Stream Model Relationships The sequence of importance of independent variable contributions to the overall variability in LMB Hg in the study streams and rivers is: pH >> DO % saturation > Conductivity > TKN > SO4 2- > TP Page 82 Aquatic Model Results • The predicted distributions of LMB Hg concentrations indicate that the most sensitive freshwater aquatic resources with respect to Hg are: • Streams and small lakes, • Large rivers, follow closely, and • The Everglades • This reflects • Potentials of load accumulation by drainage systems • Associated wetland flows prior to potential sequestration • Potentials for re-emission Page 83 THg in LMB Raw measures vs. Normalized 2.20 2.00 1.80 THG (mg/kg) 1.60 1.40 TL Norm. LMB LK 1.20 THg LMB LK 1.00 TL Norm. LMB ST 0.80 THg LMB ST 0.60 0.40 0.20 0.00 0% 20% 40% 60% Page 84 80% 100% Good fish (low in Hg) are good for you Florida has good fish to eat, today, at your favorite fish monger, and with reduced emissions this list will grow! and Scallops ≡ FL spp. 12oz/wk ≡ FL spp. 4oz/wk Mackerel Florida DOH fish advisories: http://www.doh.state.fl.us/floridafishadvice/ USEPA fish advisories: http://water.epa.gov/scitech/swguidance/fishshellfish/fishadvisories/index.cfm Page 85 QUESTIONS? • Contact Information: Watershed Evaluation & TMDL Section Bureau of Watershed Restoration Division of Environmental Assessment and Restoration Department of Environmental Protection Greg White, D.Sc. MS, MLA 850/245-8347 [email protected] Page 86 Florida Department of Environmental Protection Process for Setting Statewide mercury TMDL Jan Mandrup-Poulsen Process for Setting the TMDL • Total Maximum Daily Loads are established to set targets for restoring impaired aquatic resources. • In this case, the impairment is elevated mercury in fish tissue. • The General and Sensitive Populations must be protected. Page 88 Human Health Concerns • For the Protection of Sensitive Populations (i.e., women of childbearing age and young children): • The “market basket” of fish species consumed (includes most frequently consumed fresh and marine species eaten by Floridians) must meet 0.1 mg/Kg • For the Protection of the General Population: • In Largemouth Bass (a high level predator found in fresh waters) fish tissue levels need to be reduced to 0.3 mg/Kg Page 89 Market Basket Approach • A “Market Basket” approach was used based on fish consumption patterns for Floridians. • Relied on survey data from Degner (1994) to create a variety consumption distributions for the most commonly consumed fresh and marine fish species (e.g., canned tuna, shrimp, flounder, snapper, grouper, freshwater catfish, and others). Degner, R. L., Adams, C. M., Moss, S. D., & Mack, S. K. (1994). Per Capita Fish and Shellfish Consumption in Florida. Florida Dept of Environmental Protection, Contract W, 1-110. Portier, K. M., Um, Y., Degner, R. L., Mack, S. K., & Adams, C. M. (1995). Statistical Analysis of Florida Per Capita Fish and Shellfish Consumption Data. Florida Agricultural Market Research Center, IR 95-1, 1-185. Page 90 Market Basket Approach • Ran 10,000 iterations looking at total consumption patterns, seafood specific consumption rates, and body weight for sensitive human populations. • Simulated 52 weeks of consumption. • Goal was to assess THg levels necessary to achieve a maximum contamination level of mercury at 0.1 mg/Kg. Page 91 Market Basket Approach • Assuming a 60% reduction in Florida fish and shellfish species, and achieving a concentration of 0.15 mg/Kg in non-Florida fish species results in a >99% certainty that the desired target (0.1 mg/Kg) will be met. • A 60% reduction in mercury from all sources translates into an 86% reduction in anthropogenic sources. • Similar results reported by Sunderland (2012) Page 92 Top Level Fresh Water Predator • 133 Lakes and 131 Streams sampled for fish and water chemistry in 2008-2010 • Primarily LMB, but also Sunfish (~20%), and Spotted Bass • Targeted collection of12 fish per waterbody (or about 3,168 fish) • An average THg in fish tissue (fillet) was calculated for each waterbody sampled then averaged for all 264 waterbodies, resulting a concentration of 0.74 mg/Kg. Page 93 Top Level Fresh Water Predator • Largemouth Bass are a ubiquitous top level predator found in Florida’s freshwaters. • FDEP has a L-T data base of ~ 31,000 LMB tissue results, going back to 1983. • The 90th percentile of the fish tissue concentration for the Largemouth Bass (or its surrogates) was 0.74 mg/Kg. • To protect the general population, an 85% reduction in anthropogenic mercury is needed. Page 94 Will Achieving the 0.3 ppm Fish Tissue Target Achieve the State’s Total Mercury Criteria of 12 ng/L in Ambient Waters? Fish Tissue Concentration Target (0.3 ppm) Convert to Based on BAF Achievable Ambient Methyl-Hg Concentration (ng/L) Convert to Based on Methyl-Hg to Total Hg Ratio Achievable Ambient Total Hg Concentration (1.25 ng/L) Note: 1. BAF – Bioaccumulation Factor, is a ratio between fish tissue concentration and ambient meth-Hg concentration. 2. Estimated achievable ambient meth-Hg concentration can be converted to achievable ambient total Hg concentration through the methyl- to total Hg ratio. 3. BAF and the methyl- to total Hg ratio were calculated based on fish tissue and ambient Hg data from LMB collected in the 2008 – 2010 period. 4. The 50th percentiles of BAF and methyl- to total Hg ratio were used in calculating the achievable ambient total mercury concentration. Page 95 Florida Department of Environmental Protection Mercury … Where are we now and where are we going? Trina Vielhauer Deputy Director Division of Environmental Assessment and Restoration Where are we now and where are we going? • Global, national, state and local issue • Sources in Florida must do “fair share” • State of the art science will influence regional and global efforts • Public awareness about fish consumption Page 97 Trend of Estimated US Mercury Emissions to the Atmosphere (Source: Husar and Husar, 2002) Page 98 Trend of anthropogenic mercury flow in Florida (Source Husar and Husar, 2002) Page 99 Florida reductions – waste to energy (tons) Page 100 Florida reductions – coal fired utilities • Many have installed mercury controls in the last several years • Annual emissions have decreased from approximately 3000 pounds per year to 650 pounds per year Page 101 Waste minimization efforts • Mercury thermometer exchange programs (leading in the nation in recycling mercury thermostats) • Creating a mercury amalgam management BMP brochure • Providing data on metals loading in ash and leachate from ash disposal • Recommending removal of mercury-containing lamps and devices from buildings prior to demolition • Recycling mercury from homeowners through Florida’s Household Hazardous Waste program • Participating in the national End of Life Vehicle Solutions (ELVs) program for auto mercury • Promoting recycling of mercury containing lamps and devices through regulation and education Page 102 Public awareness Page 103