wet years / dry years: adjusting nutrient loads to monitor
TRANSCRIPT
Wet Years / Dry Years: Adjusting Nutrient Loads to Monitor Progress
Toward TMDL Reductions
Florida Stormwater Association
December, 2016
Mike Wessel, Janicki Environmental
Co-authors:
Robert Burnes, Sarah Malone, Kelly Levy, Pinellas County
Tony Janicki, Ray Pribble, Steve West , Janicki Environmental, Inc.
Problem Definition
• The Florida Department of Environmental Protection (FDEP) has recently requested that Non-Point Source Discharge Elimination System (NPDES) permittees: include a mechanism by which to account for the effects of climatological variation in rainfall on pollutant loading.
MS4 Excerpt
Objective
Using existing information, identify a mechanism by which Pinellas County could:
adjust (“Normalize”) their annual pollutant loading estimates in order to comply with the new FDEP directive, and
track progress of their Stormwater Master Plan actions over time.
Pinellas County and Tampa Bay
Pinellas County
• Eighteen nutrient TMDL waterbodies that receive discharge from the Pinellas County MS4.
• Twelve load-based TMDLs and 6 concentration-based
TMDLs.
• County has prioritized the list of TMDLs for which they are responsible and developed a schedule for addressing load reductions under the TMDLs (Pinellas County 2013).
This Presentation Details the development of a procedure used to:
1. Establish the Baseline Loading Estimate.
2. Evaluate future data against this Baseline Load.
3. Evaluate progress towards pollutant reduction goals after accounting for the effects of inter-annual variability in rainfall.
Existing Information (Data)
• Gaged Rainfall
• NEXRAD Rainfall
• Pinellas Water Quality Data
• IWMRP NPS Model Nutrient and Hydrologic Loads
• Pinellas Streamflow
• Pinellas County Nutrient and Hydrologic Loads
Spatial Summary Major Basins IWRMP Basins
IWRMP Basins with NCDC Rain Gages Pinellas Basins
WBIDs within Basins
TMDL Basins with WQ Station Locations
Conceptual Model N
atu
ral L
og
TN L
oad
Hydrologic Condition
Procedure
• Generate long-term Interpolated basin-specific rainfall using NPS model interpolation subroutine.
• Generate Baseline hydrologic condition using long term interpolated rainfall index .
• Calculate hydrologic loads for 2003-2015 using NPS model.
• Multiply ambient WQ concentrations and hydrologic loads to generate empirically-based nutrient loads.
• Standardize loads to rainfall index to account for variation in rainfall.
It All Starts with Rainfall
• Long Term Gages – Whitted and Tarpon.
• Interpolated rain gage data from NPS model subroutine: 1950-2015.
• NEXRAD – 20 years: 1995-2015.
Day Month
NEXRAD GRID and Example from July 2009
Grand Median Rainfall
1995 1999
NEXRAD
1995 1999
NEXRAD – Same Scale
Comparing Gaged Rain and NEXRAD in Gaged Rain Basins
McKay Creek
Drought Indices
• Three types:
Pinellas County Precipitation Index (PCPI).
Standardized Precipitation Index (SPI).
Standardized Precipitation and Evaporation Index (SPEI).
Pinellas County Precipitation Index
mi m
m
x
imx
m
m
PCPI =
Where:
= standard deviation of the long term monthly mean values
= monthly rain at rainfall Gage X for month i = the long term monthly mean value for Gage X
Standardized Precipitation Index
McKee, T.B., N.J. Doesken, and J. Kliest 1995. Drought monitoring with multiple time scales. In Proceedings of the 9th Conference of
Applied Climatology, 15-20 January, Dallas TX. American Meteorological Society, Boston, MA. 233-236.
SPI Calculation
• Calculate probability density function using Gamma distribution for rainfall distribution.
• Calculate deviations from expected distribution for given time period and standardize (0,1).
• Calculate over various time scales to 60 months.
http://drought.unl.edu/archive/SPI_ClimDiv/2009/SPI012009mon1.gif
SPI – Strengths and Weaknesses
• KEY STRENGTHS: – Flexible – Short time scales provide early warning of drought – Comparable over different locations – Relevance to historical context aids decision making
• KEY LIMITATIONS:
– Based only on precipitation – No soil water balance component – No ET
Standardized Precipitation Evapotranspiration Index
Vincente – Serrano, S. M. , S. Begueria, and J. I. Lopez-Moreno. 2010. A multiscaler drought index sensitive to
global warming: The Standardized Precipitation Evaporation Index. Journal of Climate. V23: 1696-1718.
SPEI Calculation
• The procedure for calculating the SPEI is similar to that for the SPI.
• SPEI uses Thornthwaite estimator to calculate PET using temperature, latitude and a reference ET (Eto).
• ETo represents evaporating power of the atmosphere at a specific location and time of the year”.
• The difference between precipitation and the reference evapotranspiration (Di = Pi - PET) is then used as the input rather than precipitation (P).
• D values are fit to a probability distribution to transform the original values to standardized units that are comparable in space and time and at different SPEI time scales, following the same procedure as that for the SPI.
Tarpon Springs SPEI at Various Timescales
SPEI-1
SPEI-12
SPEI-24
SPEI-36
SPEI-48
SPEI-60
Tarpon Springs Twelve Month SPI vs SPEI
SPI-12
SPEI-12
The Baseline Load
Existing Methods
Delivery Ratio
Rain/Stream Hyd. Anomalies
Important Aspects of Two Methods
• Delivery Ratios
Apples to Apples Comparison (same model)
Adjustment performed for every single year irrespective of magnitude of difference
Model needed to derive numbers for adjustment
• Hydrologic Anomalies
Relates streamflow/rainfall and model hydrologic load using regression
Adjustment anticipated based on confidence intervals
Nutrient model not needed to predict when adjustment is likely
“Normalization” Procedure
1. Generate Interpolated long-term rainfall using NPS model interpolation procedure (“Normal”).
2. Generate Baseline hydrologic condition using long term interpolated rainfall index (SPI).
3. Calculate hydrologic loads for 2003-2015 using NPS model.
4. Multiply ambient WQ concentrations and hydrologic loads to generate empirically-based nutrient loads.
5. Standardize loads to SPI values.
Step 1. Generate Interpolated long term rainfall using NPS model interpolation procedure 1950-2015.
Step 2. Generate Basin-Specific SPI Values
Basin 6
Step 3. Calculate hydrologic loads for 2003-2015 using NPS model and 2011 Landuse
Step 4. Multiply ambient WQ concentrations and hydrologic loads to generate empirically-based nutrient loads 2003-2015.
Step 5. Standardize loads to SPI values
Baseline Load = 9695 kg/yr
Predicted Load= 9695+4874.95*SPI Adjusted Load = Observed - 4874.95*SPI
R2 0.55 - 0.88
Total Nitrogen (mg/l) Total Phosphorus (mg/l)
TMDL Basins with non-significant regression relationship between nutrient Loads and SPI
Summary • The use of the SPI (based on long-term rainfall data)
allowed for the estimation of both a Baseline Load and the development of a method to adjust annual loading estimates to the Baseline.
• The Baseline Load is established using available, ambient water quality concentrations and model-based hydrologic loads. In this way, any improvements resulting in improved water quality should be reflected in the calculated load.
• The majority of Basins with active water quality sampling stations displayed positive relationships between rainfall and pollutant loading.
Recommendations
• Those Basins without a quantitative relationship should be investigated for potential point source discharges.
• In such cases, the Baseline Load is currently defined as the annual (geometric) average load (or concentration) over the period of record between 2003 and 2015 until such time as a relationship can be established.
Future Efforts
• Exploration of the methodology described above using the shorter timeseries of NEXRAD rainfall and basin-specific NEXRAD-based hydrologic loads would be worthwhile to see if it improves the relationships, especially in those basins where the regressions achieved an R squared value was less than 0.50.
• The use of the SPEI index may improve the long term estimates of changes in effective rainfall if ET can be effectively estimated on a basis-specific level in Pinellas County.
• With more data, statistical certainty could be included in the determination of whether or not the loads have decreased over time after accounting for variation in rainfall.
SPI<-1 SPI>1
Questions?