1. Update of Tributary Concentration and Loading Estimates to Lake Champlain Maurie Clark, Annie Procaccini, and Jamie Van Clief

2. We had a tool, not a problem ●Weighted regression on time, discharge, and season (WRTDS) model ●Exploration and Graphics for RivEr Trends (EGRET) package in R Studio coding statistical program

3. Learning experience...

4. ●Phosphorus Loading o Eutrophic conditions exist in the lake including, South Lake, St. Albans Bay, and Missisquoi Bay segments ●TMDL o 2002 loading estimates turned down by EPA ●Outdated models… revisit your tools (Philip Halterman) Background… we found a problem

5. Tributaries of interest Tributary Total Phosphorus Mean Load 1000kg/yr Mean Concentration mg/l Winooski152.7.077 Missisquoi113.6.074 Lamoille141.7.099

6. ●Using the new statistical software to evaluate phosphorus trends in and among seasons ●their causation o aided by Geographical Information Systems (GIS) ●make future recommendations Overall Goals, How we used the model

7. ●EGRET package-Exploration and Graphics for RivEr Trends ●WRTDS- weighted regressions on time, discharge and season. o True condition estimate  understand history of nutrients o Flow normalized estimate  evaluate loading/management process less related to river flow R Studio

8. ●20 years of data ●200 samples at each sampling site o 100 samples needed for estimations ●Complete record of daily discharge Characteristics needed for analysis

9. Patuxent River, Maryland (Hirsch et al. 2010) Why the need for these models? TP Winooski River, Vermont (Us 2015)

10. WRTDS Model- 1st Approach: “True Condition Estimates” Unsampled days t 0, Q 0, unknown C 0 Sampled days t i, Q i, unknown C i How to estimate a concentration? (C 0 ) To: estimate the expected value of concentration for any given date (t 0 ) and discharge (Q 0 )

11. WRTDS Model- 1st Approach: “True Condition Estimates” Unsampled days t 0, Q 0, unknown C 0 Sampled days t i, Q i, unknown C i Sift through and give weight to the sampled days data How to estimate a concentration? (C 0 ) Store in the weighted regression Run regression. Get this equation, use with t 0 and Q 0

12. WRTDS Model- 1st Approach: “True Condition Estimates” Unsampled days t 0, Q 0, unknown C 0 Sampled days t i, Q i, unknown C i Sift through and give weight to the sampled days data How to estimate a concentration? (C 0 ) Store in the weighted regression Run regression. Get this equation, use with t 0 and Q 0 Flux 0 = C 0 x Q 0 ln(c)=β ₀ + β ₁ + β ₂ ln(Q)+ β ₃ sin(2πt) +β ₄ cos(2πt) + ɛ

13. WRTDS Model- 1st Approach: “True Condition Estimates”

14. WRTDS- 2nd Approach: “Flow-normalized estimates” To: estimate concentration and flux in a manner that removes variation that arises from stochasticity in discharge

15. WRTDS- 2nd Approach: “Flow-normalized estimates” Discharge: viewed on any given day as a random sample of the discharges that may have happened on that day For April 21st, 2015 P(Q 04/21/2015 ) = P(Q 04/21/2014 ) = P(Q 04/21/2013 ) = P(Q…) = P(Q 04/21/1972 ) To: estimate concentration and flux in a manner that removes variation that arise from stochasticity in discharge

16. WRTDS- 2nd Approach: “Flow-normalized estimates” To compute FN C 0 for April 21st, 2015 with a 31-year record of discharge:

17. WRTDS- 2nd Approach: “Flow-normalized estimates” To compute FN C 0 for April 21st, 2015 with a 31-year record of discharge: Method estimates 31 values of concentration using WRTDS model Uses the T 0 but Q 04/21/2015, Q 04/21/2014, Q.... 04/21/1972 Take the average of the 31 estimated concentration or flux values Flow-normalized concentration/flux

18. Flow-normalized concentration

19. Drainage area: 850 square miles 686 square miles 1,044 square miles Annual Flow Normalized Graphs of the Missisquoi, Lamoille and Winooski Rivers

20. Land Cover Assessment ●2001 Lake Champlain Basin Simplified Land Cover Dataset o Developed for accurate mapping of land uses, which influence phosphorus loading o 19 NLCD classes -> 8 classes o Derived: National Land Cover Dataset (NLCD) and 2001 imagery o 30m pixel cells o Overall 88% accuracy assessment ●Overlaid by watershed and stream corridor data o HUC - 8 watersheds o 800m riparian buffer ● Statistics Derived using ArcGIS

21. Watershed Land Cover

22. Stream Buffer Land Cover

23. ●2009 Digital Elevation Model (DEM) o 10m cell size National Elevation Dataset ●ArcGIS Processing o Generate a slope raster, which displays terrain steepness within the stream corridors o Assess the slope distribution using 3D analyst  5 slope classes by 15° increments Stream Corridor Gradient Assessment

25. Missisquoi River

26. Lamoille River

27. Winooski River

28. Seasonality

30. Conclusions ●This model supports prioritizing management practices to specific tributaries and seasons ●Flow Normalization is an applicable tool ●Revisiting WQS and TMDLs ●Ecological timing