Supporting Online Material for · Web view‡watqual3 routine is an adaption LimnoTech developed...

Supporting information: The hydrologic model as a source of nutrient loading

uncertainty in a future climate

Haley Kujawa1,2,†, Margaret Kalcic2, Jay Martin2,3, Noel Aloysius4, Anna Apostel2, Jeffrey

Kast1,2, Asmita Murumkar2, Grey Evenson2, Richard Becker5, Chelsie Boles6, Remegio

Confesor7, Awoke Dagnew8,9, Tian Guo7, Rebecca Logsdon Muenich10, Todd Redder6,

Donald Scavia9, Yu-Chen Wang9

1Environmental Science Graduate Program, The Ohio State University, Columbus,

OH USA

2Food, Agricultural and Biological Engineering, The Ohio State University,

Columbus, OH USA

3The Sustainability Institute at Ohio State, Columbus, OH US

4Department of Bioengineering, University of Missouri, Columbia, MO USA

5Department of Environmental Sciences, University of Toledo, Toledo, OH USA

6LimnoTech, Ann Arbor, MI USA

7Heidelberg University, Tiffin, OH USA

8Environmental Consulting and Technology, Inc., Ann Arbor, MI US

9School for Environment and Sustainability, University of Michigan, Ann Arbor, MI

US

10School of Sustainable Engineering and the Built Environment, Arizona State

University, Tempe, AZ USA

†Corresponding author at: 590 Woody Hayes Dr., Columbus Ohio, 43210.

Department of Food, Agricultural, and Biological Engineering, the Ohio State

University. Email address: [email protected] (Haley Kujawa)

1

Figure S1. Range of climate model predictions plotted as changes in average annual precipitation and temperature from the historical (1996-2015) to mid-century (2046-2065). The grey lines represent the CMIP5 ensemble mean; the open circles represent each GCM in the CMIP5 ensemble, and the filled circles represent the selected GCMs.

2

Table S1. Maumee River basin model set up details.

SWAT Modeling

Modeling Decision Decision Options ModelsOSU UT HU LT UM

Model/Sub-Model Algorithms

Model Version Rev. 635-modified†

X

Other version 645 664 645 627Tile Drain Routine Old

(SWAT_TDRAIN)X

New (SWAT_HKdc)

X X X X

Water Table Routine

Old X X XNew X X

In-Stream Processes On (QUAL2E) X X X X XOn, modified‡

Soil P Model Old X X XNew X X

Evapotranspiration Method

Penman-Monteith X X X X

Hargreaves XModel Inputs Land Use Data NLCD 2001 X X

NLCD 2006 XCDL 2007-2012 X X X

Elevation Model NED 10m XNED 30m X X X X

Soils Data SSURGO X X X XSTATSGO X

Climate Inputs* NOAA NCDC - precipitation and temperature

X X X X X

Simulated solar radiation, wind, relative humidity

X X X X X

Point Source Inputs*

Measured data from EPA DMR; aggregated to average monthly

X X X X X

Spatial Discretization

HRU Thresholds LU-Soil-Slope: 0/10/0

X

LU-Soil-Slope: 200 ha/800 ha/800 haLU-Soil-Slope: 5/10/0

X X X

3

Table S1 (cont.). Maumee River basin model set up details.

Spatial Discretization

No. of sub-basins Calculation after model setup

1482 97 374 203 358

Average HRU Area (ha)

Calculation after model setup

1,130 7,700 12,677 169

ModelParameterization & Measured Data

Methods for Assessing Model Performance

R2 X X X X XNSE X X X X X

PBIAS X X X XVariables Model Performance Was Assessed For

Streamflow X X X X XTotal Phosphorus

X X X X X

Dissolved Reactive Phosphorus

X X X X X

Total Nitrogen X X X X XNitrate X X XSuspended solids

X X X X

Additional Calibration Checks

Crop Yields X X X XTile Flow X X X X XField Losses XNutrient Loss via Tile Drains

X X

Calibration and validation time periods

2010-2015 – calibration

X X X X X

2005-2009 – validation

X X X X X

Spatial Extent of Calibration

At Waterville gage only

X X X X

At Waterville, Blanchard and TiffinFlow at multiple locations and water quality at Waterville gage

X


4

ModelParameterization & Measured Data

OSU UT HU LT UMMethod to Fill in Missing Data

LOADEST for everything except DRP; Obenour et al. (2014) method for DRP

X

Model is calibrated only to observed data; missing data not included in calibration

X X X X

Land Management Operations

Fertilizer Applications

Estimated from county fertilizer sales data from 2002

X

Estimated based on maintenance application from Tri-State Standards

X X X X X

Manure Applications

Aggregated inputs from USDA-ARS NHDPlus SWAT model (Daggupati et al. 2015)Estimated from Ag Census yield and Fertilizer Use data 1990-2010

X

Estimated from county-level livestock count

X X X X

Livestock count

Crop Rotations CS X X X X X(C = Corn,S = Soybean, W = Winter Wheat,

CSS X X X

H = Hay) CSW X X X


5

OSU UT HU LT UMLand Management Operations

Tillage Estimated from CTIC

X

Estimated from USDA/OSU Extension consultation

X

Estimated from CEAP report

X X X X X

Estimated according to crop planted

X X X

Subsurface drainage All Agricultural lands with somewhat poorly, poorly, or very poorly drained soils

X X X

C,S,W HRU’s with poorly or very poorly drained soilsAgricultural or HAY lands with hydrologic group C or D soils

X

Agricultural lands with less than or equal to 3% slope

Agricultural lands with <1% slope

X

*Data homogenized for this project.†SWAT versions were modified to fix a bug where soluble P was not properly moving through subsurface drains. ‡watqual3 routine is an adaption LimnoTech developed based on White et al. (2014).

SI Table 2. Model parameters.Parameter File Spatial Description Range Final or Calibrated Value

6

Level OSU UT HU LT UMParameters that turn sub-routines on or offICN .bsn Watershed Daily curve

number calculation method: 0-calculate daily CN value as a function of soil moisture; 1-calculate daily CN value as a function of plant evapotranspiration

0/1 0 DF 0 0 0

ICRK .bsn Watershed Crack flow code; 0=no crack flow in soil; 1=crack flow in soil

0/1 0 DF DF 0 1

IRTE .bsn Watershed Channel water routing method; 0=variable travel-time; 1=Muskingum

0/1 0 0 0 0 0

ISMAX .bsn Watershed Maximum depressional storage flag, 0 = static stmaxd from .sdr

0/1 0 0 0 0 1

ITDRN .bsn Watershed Tile drainage equations flag; 1=SWAT_HKdc routine using DRAINMOD; 0=SWAT_TDRAIN method.

0/1 1 0 1 1 1

SI Table 2 (cont.). Model parameters.Parameter File Spatial Description Range Final or Calibrated Value

7

Level OSU UT HU LT UM

IWQ .bsn Watershed In-stream water quality model: 0-do not simulate nutrient transformations in stream; 1-activate simulation of in-stream nutrient transformations using QUAL2E; 2-watqual2 simulation; 3-watqual3‡.

0/1 1 1 1 1 1

IWTDN .bsn Watershed Water table depth algorithms flag

0/1 1 0 1 0 1

SOL_P_MODELΔ .bsn Watershed Soil phosphorus sub-routine: 0=new model; 1=old model

0/1 1 0 1 1 0

ADJ_PKR .bsn Watershed Peak rate adjustment factor

0.5-1.5 1.9 1 0.724 0 1

ALPHA_BF .gw HRU Baseflow recession constant

0.1-0.99 0.048-0.6 0.9964 0.299 0.254 0.9

ANION_ .sol HRU Fraction of soil pore space from which anions are excluded

0-1 0.5 DF 0.5 0.5 0.33

EXCL

BC1 .swq Subbasin Biological oxidation rate of NH4 to NO2 in the reach at 20° (1/day)

0.1-1 0.925 0.55 DF 0.36 0.1

SI Table 2 (cont.). Model parameters.Parameter File Spatial Level Description Range Final or Calibrated Value

OSU UT HU LT UM

8

BC2 .swq Subbasin Rate constant for biological oxidation of NO2 to NO3 in the reach at 20º C (1/day)

0.2-2 1.926 1.1 0.2

BC3 .swq Subbasin Hydrolysis rate of organic N to NH4 in the reach at 20° (1/day)

0.2-0.4 0.31 0.21 DF 0.2 0.02

BC4 .swq Subbasin Mineralization rate of organic P to DRP in the reach at 20° (1/day)

0.01-0.7

0.012 0.01 0.005 0.01 0.01

BIOMIX .mgt HRU Biological mixing efficiency

NA 0.75 0.2 1.142 0.2-0.6 0.25

CANMX .hru HRU Maximum canopy storage (mm H2O)

NA DF DF 1.055 5.732 DF

CDN .bsn Watershed Rate coefficient for dentirification

0-3 0.017 1.4 1.4 0.3 1.4

CH_COV1 .rte Subbasin Channel cover factor 1

0-1 0 DF 0.22 0.048 0.5

CH_COV2 .rte Subbasin Channel cover factor 2

0-1 0 DF 0.354 0.048 0.5

CH_K1 .sub Subbasin Effective hydraulic conductivity (mm/hr)

0.025-25

DF DF 3.534 DF DF

CH_K2 .rte Subbasin Effective hydraulic conductivity of channel (mm/hr)

0.025-25

DF DF 23.52 0.417 DF

CH_N1 .sub Subbasin Manning’s roughness for tributary channels

0-0.15 0.014 0.014 0.131 DF 0.02


9

Level OSU UT HU LT UM

CH_N2 .rte Subbasin Manning’s roughness for the main channel

0-0.15 0.06 0.014 0.147 0.057 0.035

CN2 .mgt HRU Initial SCS moisture condition II curve number

0.75-1.25†

varies 35-98 49.74 – 96.98

30-95 DF

CNOP .mgt HRU SCS runoff curve number for moisture condition II

NA DF DF DF 75-89 DF

DDRAIN .mgt HRU Depth to subsurface tile drain (mm)

0-6000 1000* 0-2000

915* 1000* 1000*

DEP_IMP .hru HRU Depth to the impervious layer in the soil (mm)

0-6000 2300* 0-6000

1294* 2500-3500 for drained*

1500*

DRAIN_CO .sdr HRU Daily drainage coefficient (mm/day)

Oct-51 35 DF DF 12.7 20

EPCO .bsn Watershed Plant uptake compensation factor.

0.01-1.0 0.95-1.0 1 1 0.6381 1

ERORGN .hru HRU Nitrogen enrichment ratio for loading with sediment, 0 allows model to calculate value

NA DF 1.174 DF 2 DF

ERORGP .hru HRU Phosphorus enrichment ratio for loading with sediment, 0 allows model to calculate value

NA DF 1.25 DF 2 DF

ESCO .bsn, .hru

Watershed HRU

Soil evaporation compensation factor

0.01-1 0.97hru 0.95 0.997bsn 1 0.95bsn


10

Level

OSU UT HU LT UM

GDRAIN .mgt HRU Drain tile lag time (hours)

NA NA 24 2 96 NA

GW_DELAY .gw HRU Delay time for aquifer recharge (days)

NA 31-Mar

DF 5.177 31 DF

GWQMN .gw HRU Threshold water level in shallow aquifer for base flow (mm H2O)

NA 780-1000

DF 817.66 447.58 DF

GW_REVAP .gw HRU Revap coefficient

0.02-2 0.02 0.2-2 0.055 0.02 0.07

HRU_SLP .hru HRU Average slope steepness (m/m)

0.75-1.25†

DF DF DF† varies by HRU, max of 0.079

DF

IFLOD1R .res Subbasin Beginning month of non-flood season

12-Jan DF DF DF 1 or 12 DF

IFLOD2R .res Subbasin Ending month of non-flood season

12-Jan DF DF DF 1 DF

LATKSATF .sdr HRU Lateral soil hydraulic conductivity in tile-drained fields as multiple of original soil conductivity value

0.01-4 0.54 DF 0.994 2 or 4 1

NDTARGR .res Subbasin Number of days to reach target storage from current reservoir storage

NA DF DF DF 1 or 5 DF

NPERCO .bsn Watershed Nitrate percolation coefficient

0.01-1 0.36 0.2 0.284 0.5 0.9

OVN .hru HRU Manning’s “n” value for overland flow

0.008-0.5 0.029 X 1.227 0.1 to 0.3 0.1-0.2

PHOSKD .bsn Watershed Phosphorus soil partitioning coefficient

80-350 181 204.5 173.8 175 175

SI Table 2 (cont.). Model parameters.Parameter File Spatial Level Description Range Final or Calibrated Value

11

OSU UT HU LT UM

PPERCO .bsn Watershed Phosphorus percolation coefficient (m3/Mg)

10-17.5

10 10.94 11.637 5 10

PSP .bsn Watershed Phosphorus availability index

0.2-0.6 0.4 0.425 0.62 0.1 0.4

R2ADJ .hru HRU Curve number adjustment for increasing infiltration in non-draining soils

0-3 1 1 0.907 1.15 to 3

1

RE .sdr HRU Effective radius of drains (mm)

Mar-40

30 DF 37.9 10 20

REVAPMN .gw HRU Threshold water level level in shallow aquifer for revap (mm H2O)

NA 116-750

DF 947 388.62 DF

RS1 .gw HRU Local algal settling rate in the reach at 20ºC (m/day)

0.015-1.82

1 1

RS2 .swq

Subbasin Benthic source rate for DRP in the reach at 20° (mg P/m2-d)

NA 0.03 DF x1.22 0.05 0.01

RS3 .swq

Subbasin Benthic source rate for ammonium in the reach at 20° (mgNH4-N/m2/d)

NA 0.46 DF x0.927 0.5 1

RS4 .swq

Subbasin Organic N settling rate in the reach at 20° (1/day)

0.001-0.1

0.003 DF DF 0.01 0.001

SI Table 2 (cont.). Model parameters.

12

Parameter File Spatial Level

Description Range Final or Calibrated Value

OSU UT HU LT UM

RS5 .swq Subbasin Local settling rate for organic phosphorus mineralization at 20° (day-1)

0.001-0.1

0.1 DF x1.071 0.01 0.05

SDNCO .bsn Watershed

Threshold value of nutrient cycling water factor for denitrification to occur

0.75-1.4

0.97 1.1 1.287 1 1.1

SDRAIN .sdr HRU Tile drain spacing (mm)

7,600-30,000

10000* 21165 13720 10000*

SFTMP .bsn Watershed

Mean air temperature at which precipitation is equally likely to be rain as snow/freezing rain (°C)

-10 -2 1 -0.7 1 -2

SHALLST .gw HRU Initial depth of water in the shallow aquifer (mm H2O)

NA 1000 DF 500 DF

SLSUBSN .hru HRU Average slope length

0.75-1.25

DF DF x1.224 DF

SMFMN .bsn Watershed

Minimum snow melt factor (mm H2O/day-°C)

1.4-6.9 2 4.5 4.1 3 2

SMFMX .bsn Watershed

Maximum snow melt factor (mm H2O/day-°C)

1.4-6.9 2 4.5 6.6 4-Jan 2

SMTMP .bsn Watershed

Threshold temperature for snowmelt (°C)

-10 -2 0.5 2.08 0.5 -2

SOL_AWC .sol HRU Available water capacity

0.75-1.25

varis – calib. parameter

DF x0.854 DF DF

SOL_CRK .sol HRU Potential crack volume for soil profile

0-1 DF DF 0.142 DF 0.42

SOL_K .sol HRU Saturated hydraulic conductivity (mm/hr)

0.75-1.25

varis – calib. parameter

DF x0.922 DF DF

SOL_ORGP .chm HRU Initial humic organic phosphorus in soil layer (mg/kg or ppm)

50-250 DF DF 202 DF DF

SI Table 2 (cont.). Model parameters.

13

Parameter

File Spatial Level

Description Range Final or Calibrated Value

OSU UT HU LT UM

SOL_SOLP .chm HRU Initial labile P in the soil layer (mg labile P/kg soil)

5-100 0.5 DF 0.7 DF 1

SPCON .bsn Watershed Parameter drives the maximum concentration of sediment the river can route

0.0001-0.01 0.0026 #######

2.60E-04

0.004 #######

SPEXP .bsn Watershed Exponent parameter for calculating sediment reentrained in channel sediment routing

2-Jan 1.425 DF DF 1 1

SURLAG .bsn Watershed Surface runoff lag coefficient

NA 1 4 1.007 2.8723

1

TDRAIN .mgt HRU Time to drain soil to field capacity (hours)

NA NA 48 36 NA

TIMP .bsn Watershed Snow pack temperature lag

0.01-1 0.05 1 0.327 0.06 0.05

USLE_C crop.dat By land-use

Minimum value for the cover and management factor for the land cover

0.75-1.25 DF DF x1.055 DF DF

USLE_K .sol HRU USLE soil erodibility factor (0.013 metric ton m2-hr/m3- metric ton cm)

0.75-1.25 varis – calib. parameter

DF x1.118 DF DF

USLE_P .mgt HRU USLE support practice factor

0.50-1.25 varis – calib. parameter

0.16-1.6 DF 1 DF

VCRIT .bsn Watershed Critical velocity at which a river will resuspend sediments

NA 5 5 5 0 1

‡watqual3 routine is an adaption LimnoTech developed based on White et al. (2014).ΔSWAT 2012 revision 635 indicate in basins.bsn that 1 is the new soil phosphorus model; however, examination of the source code followed by confirmation from Nancy Sammons (in a post to the SWAT-user group on 2/26/2014) confirms that setting this parameter equal to 0 will run the new soil phosphorus sub-routine.

14

Table S3. Watershed management decisions based on guidelines provided to the modeling teams.No Input type Model setup

guidanceDecision

OSU UT HU LT UM

1 Phosphorus fertilizer application

Phosphorus (P) fertilizer is incorporated in approximately 60% of the cropland, and the rest is broadcast application. The CEAP report informs that in 2012 that 40% of acres had all broadcast applications with no incorporation, and 60% of acres had all applications incorporated.

Subsurface placement on 35% of cropland

Ensured that a tillage operation immediately followed MAP application for 60% of cropland.

Full subsurface application: 43.6%

MASWAT Cal015 was modified to change all FRT_SURFACE values to 0.10. Then, this modified management file was run through the R script and 40% of the land area in AGRR was changed to 0.75. The R script was applied to both P and N fertilizers for consistency.

57% of acres had all P fertilizer applications incorporated within 3 days;

Phosphorus subsurface placement means 99% of fertilizer application is below the top 1-cm of the soil surface.

Combined subsurface and broadcast: 53.6%

22% had all P fertilizer applications broadcast without incorporation;

21% had a mixture of broadcast with incorporation and without.

Purely Broadcast with no incorporation: 2.8%

15

Table S3 (cont.). Watershed management decisions based on guidelines provided to the modeling teams.No Input type Model setup

guidanceDecision

OSU UT HU LT UM

2 Cover crop The CEAP report says "fewer than 6 percent of acres were managed with cover crops" in 2012, yet the Wilson report [Burnett et al., 2015] suggests closer to 10% with increasing rates the past few years. Cover crops are applied in the fall and should not include winter wheat.

Cover crop (rye) is planted in 8.4% of the cropland, and only in corn-soybean rotations.

Cover crop (rye) is planted in 10% of total, corn-soy rotations

6% of HRUs

In the R script utilized, Pchange was set to 7.5%. Based on HRU size and random selection, 7.5% of the ag land was changed to include cover crops. The cover crop chosen was a RYE plant which was planted 7 days after the harvest of either a CORN or SOYBEAN crop. WHEAT was excluded. Heat units to maturity was set as 1540. All cover crops were killed using operation 8 on April 15th.

Cover crop (rye) is planted in 8.4% of cropland, in every winter of a CSS rotation, in which P fertilizers are fully incorporated

For our models, cover crops include cereal rye every winter without wheat.

3 Vegetative filter strip

30% of the cropland has filter strips.

29% of the cropland (corn, soybean, winter wheat and hay pasture)

Applied to 32% of croplands

30% of HRUs (still need to calculate)

NR 34% of cropland received a medium-quality filter strip. Rotations include CS and CSS (P fertilizers fully incorporated), and two CSWCSSW rotations (with P fertilizers partially or not at all incorporated).

Buffers are sized at 2% of the field drainage area with 50% being concentrated flow and 12.5% being flly channelized.

16

Table S3 (cont.). Watershed management decisions based on guidelines provided to the modeling teams.No Input

typeModel setup guidance

Decision

OSU UT HU LT UM

4 Manure application

The CEAP report informs that manure is applied in 9% of the cropland, whereas, the OSU farmer survey reports 14%. Modeling teams were provided with application rates by county, livestock numbers and references to estimate manure production.

Varying levels of application by county. Nearly 14% of the cropland receives manure. We estimated dairy-equivalent manure production based on livestock count at county level and applied all manure to HRUs within that county. We estimated elemental N and P amount in manure applied and reduced that amount in mineral fertilizer applications (in MAP – mono ammonium phosphate).

Approximately 14% of the fields had manure applied. Amount was based on county livestock counts and type, and evenly distributed over 14% of the county by area.

Application rates by county was implemented. Original N & P app rates were maintained as sum of both manure and inorganic.

For manure, the raw NuGIS data was used rather than the estimates for manure P provided by OSU. Data was trimmed down to only years 2010 to 2012. The recovered manure estimates were used and converted to elemental P from P2O5. The ‘Farm_TonsP’ was converted to elemental P from P2O5 and used to estimate mineral fertilizer.

Varying levels of application by county. All cropland within a country received manure from livestock produced in that county. This is an assumption we intend to improve soon. We estimated the manure generated in each county from the Ag Census livestock numbers and applied the SWAT default manure for dairy, beef, swine, layer, and broilers.

17

Table S3 (cont.). Watershed management decisions based on guidelines provided to the modeling teams.No Input

typeModel setup guidance

Decision

OSU UT HU LT UM

4 Manure application

The CEAP report informs that manure is applied in 9% of the cropland, whereas, the OSU farmer survey reports 14%. Modeling teams were provided with application rates by county, livestock numbers and references to estimate manure production.

(cont.) Another R script was used to add the manure to the cropland HRUs in the HABRI model (cal0030, with the fertilizer updates). This script cycled through the mgt2 table of the updated model. The first loop was based on the county. Every HRU was tagged with the county it resided in (this was based on subbasin). Total county area was calculated by summing the HRUs. The script then noted the total amount of annual fertilizer which should be applied and fraction of the county crop area which should receive manure. If the county fraction was zero, no HRUs were selected and no manure was applied in that county. This gave an estimated land area to which manure should be applied. Next the script randomized the county HRUs and selected enough HRUs to meet or exceed the area target. Phosphorus was applied using layer manure, which has a P fraction of 0.006 and an N fraction of 0.013.

18

Table S3 (cont.). Watershed management decisions based on guidelines provided to the modeling teams.No Input type Model setup

guidanceDecision

OSU UT HU LT UM

5 Subsurface drainage

70-78% of the cropland has subsurface drainage. Our estimate is based on total cropland and subsurface drainage extent reported in the CEAP report Table 1.1 and chapter 3 (4,861,000 acres of cropland, (3,400,000-3,800,000 acres treated with subsurface drainage). These values are for all Western Lake Erie watersheds.

75% of the cropland has subsurface drainage installed.

77% of the cropland has subsurface drainage

< 3% slope have drainage.

76% has drainage

71.5% of cropland has subsurface drainage.

6 Tillage The CEAP report (Figure 2.1) informs that 20-30% of cropland is managed in continuous no-tillage and 25-35% of cropland is managed in seasonal no-tillage.

No-till is practiced in 37% of the cropland.

NR 63% Rotational Till

NR Continuous no-till is practiced in 22% of cropland (in one CSS rotation and two CSWCSSW rotations);

Seasonal no-till is practiced in 21% of cropland (in one CS rotation and one CSWCSSW rotation)

36.5% Pure No till

0.5% Conventional Till

7 Point sources

Modeling teams were provided with point source information.

All point sources are included in the model. Point source data was compiled by UM.

All point sources included from UM compilation

NR Point source data included which was sent via Box account

All point sources are included in the model. Point source data was compiled by UM.

19

Table S4. Modeled water budgets and crop yields compared to edge-of-field data****. Watershed Hydrology (all are mm/year) and Crop Yields (all are bushels/acre)

Observed OSU UT HU LT UM

Total runoff 362±100* 355 366 392 403 344

Surface flow 222 204 187 214 111

Lateral flow 7 3 27 5 2

Subsurface (tile) flow** 260-515*** 165 117 208 182 267

Groundwater 15 85 20 61 94

Evapotranspiration 613 605 583 566 565

Corn Yield 153±21 135 160 114 128 126

Soybean Yield 46±4 36 58 41 42 37

Wheat Yield 69±5 64 67 65 47 71

*Data from National Center for Water Quality Research at Heidelberg University**Results show area-weighted mean and includes results only from HRUs with tile drains***Pease, L. (2016). Characterization of Agricultural Subsurface Drainage Water Quality and Controlled Drainage in the Western Lake Erie Basin. Thesis. The Ohio State University****King et al. (2018). Addressing agricultural phosphorus loss in artificially drained landscapes with 4R nutrient management practices. Journal of Soil and Water Conservation

20

Figure S2. Annual model results plotted against observed at Waterville gauge for discharge during the calibration and validation period (2005-2015).

Figure S3. Annual model results plotted against observed at Waterville gauge for total phosphorus during the calibration and validation period (2005-2015).

21

Figure S4. Annual model results plotted against observed at Waterville gauge for dissolved reactive phosphorus during the calibration and validation period (2005-2015).

Figure S5. Annual model results plotted against observed at Waterville gauge for total nitrogen during the calibration and validation period (2005-2015).

22

Figure S6. Annual model results plotted against observed at Waterville gauge for suspended solids during the calibration and validation period (2005-2015).

Table S5. Similar climate ensemble studies using a top-down approach (i.e. climate model selected first to then drive hydrologic model) and the number of climate models used

Citation Number of climate models

Citation Number of climate models

Wilby & Harris, 2006 4 Thober et al., 2018 5

Kay et al., 2008 5 Giuntoli et al., 2015 5

Wilby et al., 2006 3 Bosshard et al., 2014 10

Poulin et al., 2011 2 Prudhomme et al., 2014 5

Arnell, 1999 4 Karlsson et al., 2016 4

Velazquez et al., 2013

5 Addor et al., 2014 10-20

Bosshard et al., 2013 8 Dams et al., 2015 3

Vetter et al., 2017 5

23

Table S6. Difference in average annual discharge (%) between the historical and mid-century period for each SWAT model and GCM. Significant changes (p < 0.05) as tested by the Rank-Sum Test are highlighted in grey. Results are plotted in figure 4 in main text.

CSIRO_r10

CSIRO_r4

CSIRO_r6

CanESM

MPI-ESM

NorESM

Average Range

UT 20 -4 4 6 -20 12 3 39UM 17 -4 4 14 -17 13 5 34OSU 16 -6 2 12 -20 12 3 36LT 28 -6 4 17 -19 18 7 46HU 16 -4 4 12 -17 11 4 32

Average 19 -5 4 12 -18 13Range 12 3 2 10 3 7

Table S7. As in table S2, but for changes in surface runoff, tile flow, and evapotranspiration.Surface Runoff

CSIRO_r10 CSIRO_r4 CSIRO_r6 CanESM MPI-ESM

NorESM Average Range

UT -17 -25 -19 -32 -14 -12 -20 20UM -29 -29 -13 -27 -22 -30 -25 16OSU 3 -18 -9 -2 -21 -2 -8 24LT 8 -13 0 -1 -18 -6 -5 26HU -2 -18 -4 -5 -22 -6 -9 20Average -7 -21 -9 -13 -19 -11Range 36 16 19 31 8 28Tile discharge

CSIRO_r10 CSIRO_r4 CSIRO_r6 CanESM MPI-ESM

NorESM Average Range

UT 64 21 35 42 -18 34 30 82UM 28 2 8 21 -12 26 12 40OSU 32 5 15 27 -16 26 15 48LT 46 2 9 34 -16 42 19 62HU 30 6 9 25 -13 24 14 42Average 40 7 15 30 -15 30Range 36 19 27 21 6 17ET CSIRO_r10 CSIRO_r4 CSIRO_r6 CanESM MPI-

ESMNorESM Average Range

UT 9 7 7 13 6 7 8 8

24

UM 9 6 5 9 5 5 6 5OSU 10 9 7 11 8 6 9 5LT 7 5 5 9 1 5 5 9HU 10 7 6 11 7 6 8 5Average 9 7 6 11 5 6Range 3 4 3 4 8 2

Table S8. As in table S2 but for TP and DRPTP CSIRO_r10 CSIRO_r4 CSIRO_r6 CanESM MPI-


UT -13 -26 -17 -24 -15 -8 -17 18UM -1 -4 18 3 2 -1 3 22OSU 20 -6 8 7 -9 16 6 29LT -9 -24 -10 -5 -29 -19 -16 24HU -6 -25 -3 -7 -26 -10 -13 23

Average -2 -17 -1 -5 -15 -5Range 32 22 35 31 31 35DRP CSIRO_r10 CSIRO_r4 CSIRO_r6 CanESM MPI-


UT -10 -25 -18 -21 -23 -9 -17 16UM 34 24 33 38 16 32 29 22OSU 32 4 11 30 -4 25 16 36LT 23 -7 4 16 -20 12 5 44HU -4 -17 -3 -5 -21 -3 -9 18

Average 15 -4 5 12 -10 11Range 45 48 51 59 38 41

Table S9. As in table S2 but for total nitrogenTN CSIRO_r10 CSIRO_r4 CSIRO_r6 CanESM MPI-


25

UT 22 -12 5 14 -31 10 1 52UM 12 -5 11 4 -9 9 4 21OSU 13 -3 7 6 -8 9 4 22LT 3 -11 0 4 -26 -6 -6 31HU -4 -14 -6 -3 -16 -8 -8 13Average 9 -9 4 5 -18 3Range 25 11 17 17 23 18

Figure S7. Monthly changes (absolute value, cms) in average monthly discharge at the Maumee at Waterville location from the historical (1996-2015) to the mid-century (2046-2065).

26

Figure S8. Monthly changes (absolute value, mm) in average monthly surface runoff averaged across the basin from the historical (1996-2015) to the mid-century (2046-2065).

27

Figure S9. Monthly changes (absolute value, mm) in average monthly tile drainage averaged across the basin from the historical (1996-2015) to the mid-century (2046-2065).

28

Figure S10. Monthly changes (absolute value, mm) in average monthly evapotranspiration averaged across the basin from the historical (1996-2015) to the mid-century (2046-2065).

29

Figure S11. Monthly changes (absolute value, tons) in average monthly total phosphorus load at the Maumee at Waterville location from the historical (1996-2015) to the mid-century (2046-2065).

30

Figure S12. Monthly changes (absolute value, tons) in average monthly dissolved reactive phosphorus load at the Maumee at Waterville location from the historical (1996-2015) to the mid-century (2046-2065).

31

Figure S13. Monthly changes (absolute value, tons) in average monthly total nitrogen load at the Maumee at Waterville location from the historical (1996-2015) to the mid-century (2046-2065).

Text S1. Statistical transformations of dataAll tests described in section 2.6. were conducted on the data as percent change from the historical to mid-century, except for the Wilcoxon Rank-Sum test which was done as a test comparing all years in the historical and mid-century period. Predictions for discharge and DRP failed to meet the assumption of normality via the Shapiro-Wilk test (p < 0.05) (Sharpiro & Wilk, 1965), and so all data were transformed to meet normality assumptions using the rank normalization method. The ANOVA was run for both transformed and untransformed data. A smaller fraction of the ANOVA results for each variable (e.g. discharge, total nitrogen, etc.) failed to meet heteroscedasticity (6/7 variables with original data vs. 4/7 transformed data) and a larger number met normality of residuals (0/7 vs. 4/7) assumptions. However, the relative ANOVA results (i.e. ranking of uncertainty source) did not change substantially between transformed and untransformed data and therefore we chose to discuss results on the original data, as was done in Giuntoli et al. (2015).

32

Figure S14. Historical annual average observed discharge (blue) vs. historical annual average modeled discharge (red) for the UM model

33

Figure S15. Historical annual average observed discharge (blue) vs. historical annual average modeled discharge (red) for the OSU model

34

Figure S16. Historical annual average observed discharge (blue) vs. historical annual average modeled discharge (red) for the UT model

35

Figure S17. Historical annual average observed discharge (blue) vs. historical annual average modeled discharge (red) for the LT model

36

Figure S18. Historical annual average observed discharge (blue) vs. historical annual average modeled discharge (red) for the HU model

37

Figure S19. Historical annual average observed DRP (blue) vs. historical annual average modeled DRP (red) for the UM model

38

Figure S20. Historical annual average observed DRP (blue) vs. historical annual average modeled DRP (red) for the OSU model

39

Figure S21. Historical annual average observed DRP (blue) vs. historical annual average modeled DRP (red) for the UT model

40

Figure S22. Historical annual average observed DRP (blue) vs. historical annual average modeled DRP (red) for the LT model

41

Figure S23. Historical annual average observed DRP (blue) vs. historical annual average modeled DRP (red) for the HU model

42

Figure S24. Historical annual average observed TP (blue) vs. historical annual average modeled TP (red) for the UM model

43

Figure S25. Historical annual average observed TP (blue) vs. historical annual average modeled TP (red) for the OSU model

44

Figure S26. Historical annual average observed TP (blue) vs. historical annual average modeled TP (red) for the UT model

45

Figure S27. Historical annual average observed TP (blue) vs. historical annual average modeled TP (red) for the LT model

46

Figure S28. Historical annual average observed TP (blue) vs. historical annual average modeled TP (red) for the HU model

47

Figure S29. Historical annual average observed TN (blue) vs. historical annual average modeled TN (red) for the UM model

48

Figure S30. Historical annual average observed TN (blue) vs. historical annual average modeled TN (red) for the OSU model

49

Figure S31. Historical annual average observed TN (blue) vs. historical annual average modeled TN (red) for the UT model

50

Figure S32. Historical annual average observed TN (blue) vs. historical annual average modeled TN (red) for the LT model

51

Figure S33. Historical annual average observed TN (blue) vs. historical annual average modeled TN (red) for the HU model

52

Table S10. Crop yields as a percent change from the mid-century to historicalModel GCM Corn Soy Wheat'UT_SWAT' 'CSIRO_r10' -17 -10 -10'UT_SWAT' 'CSIRO_r4' -8 1 -16'UT_SWAT' 'CSIRO_r6' -18 -8 -16'UT_SWAT' 'CanESM' -14 -9 -14'UT_SWAT' 'MPI-ESM' -10 -8 -13'UT_SWAT' 'NorESM' -14 -2 -15'UM_SWAT' 'CSIRO_r10' -16 -6 -14'UM_SWAT' 'CSIRO_r4' -10 -2 -12'UM_SWAT' 'CSIRO_r6' -19 -9 -19'UM_SWAT' 'CanESM' -12 -12 -19'UM_SWAT' 'MPI-ESM' -7 -5 -12'UM_SWAT' 'NorESM' -13 -2 -16'OSU_SWAT' 'CSIRO_r10' -15 -19 -11'OSU_SWAT' 'CSIRO_r4' -13 -18 -5'OSU_SWAT' 'CSIRO_r6' -16 -21 -11'OSU_SWAT' 'CanESM' -13 -17 -5'OSU_SWAT' 'MPI-ESM' -9 -17 3'OSU_SWAT' 'NorESM' -13 -17 -12'LT_SWAT' 'CSIRO_r10' -19 -16 5'LT_SWAT' 'CSIRO_r4' -17 -17 10'LT_SWAT' 'CSIRO_r6' -18 -18 3'LT_SWAT' 'CanESM' -11 -17 6'LT_SWAT' 'MPI-ESM' -14 -14 9'LT_SWAT' 'NorESM' -14 -21 10'HU_SWAT' 'CSIRO_r10' -23 -22 -15'HU_SWAT' 'CSIRO_r4' -21 -18 -1'HU_SWAT' 'CSIRO_r6' -24 -24 -21'HU_SWAT' 'CanESM' -18 -23 -10'HU_SWAT' 'MPI-ESM' -16 -19 -4'HU_SWAT' 'NorESM' -21 -20 -17

MIN -24 -24 -21MAX -7 1 10

53

Table S11. Previous climate change studies done on the Maumee River WatershedCite Years Concentration

pathwayNumber of years assessed

Discharge TP DRP TN

Culbertson at al. (2016)

baseline = 1980-2009; mid-century=2040-2069

RCP 4.5 & 8.5 30 2% -6.60% -3.50% NA

Bosch et al. (2014) baseline: 1998-2005; high degree change: 2010-2039; low degree change: 2040-2069

A1F1, B1 8-year historical; 30-year future

+18 mm/yr -0.5 kg P/km2 +0.3 kg P/km2

+64 kg N/km2

Verma et al. (2015) baseline (ran w/ observed climate): 1995-2005; mid-century t1: 2045-2055; late century t2: 2089-2099

A1B 11 summer flows decrease, winter flows increase; t1: -8.5%, t2: 9.7%

t1: -8.5%; t2: +3.5%

NA NA

Johnson et al. (2015) baseline: 1971-2000; mid-century: 2041-2070

A2 30 0.50% 1.30% NA -0.40%

Cousino et al. (2015) baseline: 1985-2004; mid-century: 2046-2065; late-century; 2080-2099

RCP 4.5 & 8.5 20 -10.30% NA NA NA

Kalcic et al. (2019)baseline: 1980-1999; mid-century: 2046-2065

RCP 8.5 20 NA -10% -5% NA

54

References

Addor, N., Rössler, O., Köplin, N., Huss, M., Weingartner, R., & Seiber, J. (2014). Robust changes and sources of uncertainty in the projected hydrological regimes of Swiss catchments. Water Resources Research, 50, 7541–7562. https://doi.org/10.1002/2014WR015549.Received

Arnell, N. W. (1999). The Effect of Climate Change on Hydrological Regimes in Europe. Global Environmental Change, 9, 5–23.

Bosch, N. S., Evans, M. A., Scavia, D., & Allan, J. D. (2014). Interacting effects of climate change and agricultural BMPs on nutrient runoff entering Lake Erie. Journal of Great Lakes Research, 40(3), 581–589. https://doi.org/10.1016/j.jglr.2014.04.011

Bosshard, T., Carambia, M., Goergen, K., Kotlarski, S., Krahe, P., Zappa, M., & Schär, C. (2013). Quantifying uncertainty sources in an ensemble of hydrological climate-impact projections. Water Resources Research, 49(3), 1523–1536. https://doi.org/10.1029/2011WR011533

Bosshard, Thomas, Kotlarski, S., Zappa, M., & Schär, C. (2014). Hydrological Climate-Impact Projections for the Rhine River: GCM–RCM Uncertainty and Separate Temperature and Precipitation Effects*. Journal of Hydrometeorology, 15(2), 697–713. https://doi.org/10.1175/JHM-D-12-098.1

Cousino, L., Becker, R., & Zmijewski, K. (2015). Modeling the Effects of Climate Change on Water, Sediment, and Nutrient Yields from the Maumee River Watershed. Journal of Hydrology: Regional Studies, 4, 762–775. https://doi.org/10.1016/j.ejrh.2015.12.039

Culbertson, A. M., Martin, J. F., Aloysius, N., & Ludsin, S. A. (2016). Anticipated impacts of climate change on 21st century Maumee River discharge and nutrient loads. Journal of Great Lakes Research, 42, 1332–1342. https://doi.org/10.1016/j.jglr.2016.08.008

Dams, J., Nossent, J., Senbeta, T. B., Willems, P., & Batelaan, O. (2015). Multi-model approach to assess the impact of climate change on runoff. Journal of Hydrology, 529. https://doi.org/10.1016/j.jhydrol.2015.08.023

Giuntoli, I., Vidal, J. P., Prudhomme, C., & Hannah, D. M. (2015). Future hydrological extremes: The uncertainty from multiple global climate and global hydrological models. Earth System Dynamics, 6(1), 267–285. https://doi.org/10.5194/esd-6-267-2015

Johnson, T., Butcher, J., Deb, D., Faizullabhoy, M., Hummel, P., Kittle, J., … Witt, J. (2015). Modeling Streamflow and Water Quality Sensitivity to Climate Change and Urban Development in 20 U.S. Watersheds. Journal of the American Water Resources Association, 51(5), 1321–1341. https://doi.org/10.1111/1752-1688.12308

Karlsson, I. B., Sonnenborg, T. O., Refsgaard, J. C., Trolle, D., Børgesen, C. D., Olesen, J. E., … Jensen, K. H. (2016). Combined effects of climate models, hydrological model structures and land use scenarios on hydrological impacts of climate change. Journal of Hydrology, 535, 301–317. https://doi.org/10.1016/j.jhydrol.2016.01.069

Kay, A. L., Davies, H. N., Bell, V. A., & Jones, R. G. (2008). Comparison of uncertainty sources for climate change impacts: Flood frequency in England. Climatic Change, 92, 41–63. https://doi.org/10.1007/s10584-008-9471-4

Kalcic, M. M., Muenich, R. L., Basile, S., Steiner, A. L., Kirchhoff, C., & Scavia, D.

55

(2019). Climate change and nutrient loading: warming can counteract a wetter future. Environmental Science & Technology, in press.

Poulin, A., Brissette, F., Leconte, R., Arsenault, R., & Malo, J. S. (2011). Uncertainty of hydrological modelling in climate change impact studies in a Canadian, snow-dominated river basin. Journal of Hydrology, 409(3–4), 626–636. https://doi.org/10.1016/j.jhydrol.2011.08.057

Prudhomme, C., Giuntoli, I., Robinson, E. L., Clark, D. B., Arnell, N. W., Dankers, R., … Wisser, D. (2014). Hydrological droughts in the 21st century, hotspots and uncertainties from a global multimodel ensemble experiment. Proceedings of the National Academy of Sciences, 111(9), 3262–3267. https://doi.org/10.1073/pnas.1222473110

Sharpiro, S. S., & Wilk, M. B. (1965). An analysis of variance test for normality (complete samples). Biometrika, 52(3/4), 591–611. https://doi.org/10.2307/2333709

Thober, S., Kumar, R., Wanders, N., Marx, A., Pan, M., & Rakovec, O. (2018). Multi-model ensemble projections of European river floods and high flows at 1.5 , 2 , and 3 degrees global warming. Environmental Research Letters, 13.

Velazquez, J. A., Schmid, J., & Ricard, S. (2013). An ensemble approach to assess hydrological models ’ contribution to uncertainties in the analysis of climate change impact on water Solid Earth. 565–578. https://doi.org/10.5194/hess-17-565-2013

Verma, S., Bhattarai, R., Bosch, N. S., Cooke, R. C., & Kalita, P. K. (2015). Climate Change Impacts on Flow, Sediment and Nutrient Export in a Great Lakes Watershed Using SWAT. Clean Soil Air Water, 43(11), 1464–1474. https://doi.org/10.1002/clen.201400724

Vetter, T., Reinhardt, J., Flörke, M., van Griensven, A., Hattermann, F., Huang, S., … Krysanova, V. (2017). Evaluation of sources of uncertainty in projected hydrological changes under climate change in 12 large-scale river basins. Climatic Change, 141(3), 419–433. https://doi.org/10.1007/s10584-016-1794-y

Wilby, R. L., & Harris, I. (2006). A framework for assessing uncertainties in climate change impacts: Low-flow scenarios for the River Thames, UK. Water Resources Research, 42(2), 1–10. https://doi.org/10.1029/2005WR004065

Wilby, R. L., Whitehead, P. G., Wade, A. J., Butterfield, D., Davis, R. J., & Watts, G. (2006). Integrated modelling of climate change impacts on water resources and quality in a lowland catchment: River Kennet, UK. Journal of Hydrology, 330, 204–220. https://doi.org/10.1016/j.jhydrol.2006.04.033

56

Supporting Online Material for · Web view‡watqual3 routine is an adaption LimnoTech developed...

Documents

Transcript of Supporting Online Material for · Web view‡watqual3 routine is an adaption LimnoTech developed...