Minnesota Nitrogen Science Assessment and N Reduction Planning Tool D. J. Mulla, Department of Soil, Water, and Climate University of Minnesota, W. F. Lazarus Department of Applied Economics University of Minnesota, D. Wall Minnesota Pollution Control Agency GOALS • Assess nonpoint source nitrogen contributions to Minnesota rivers from a) the primary land use sources, and b) the primary hydrologic pathways under dry, average and wet climatic conditions • Determine the watersheds which contribute the most nitrogen to the Mississippi River, and combination of land uses and hydrologic factors having the greatest influences on the elevated nitrogen • Develop a nitrogen decision tool to estimate reductions in N loadings to surface waters at the watershed scale with various BMPs Reasons for Study Provide technical information to help establish Minnesota goals and strategies to address its contribution to: Nitrogen export to Gulf of Mexico Nitrate concentration impairments in surface waters which may arise due to new numerical nutrient criteria Minnesota technical assessments informing nutrient reduction strategy Technical Assessment Nitrogen Phosphorus Watershed outlet load monitoring (70+ watersheds) X X Major River Monitoring X X SPARROW Modeling of all HUC8 watersheds X X Stream Concentration monitoring 700 sites X X Water Quality Standards effects on loads X X Twin Cities effects on loads X X Temporal trends (50+ sites 1976-2010) X X Seasonal variability in loads and concentrations X In-stream losses X X Point source contributions X X Nutrient budgets to cropland X X X X Nonpoint sources to waters Nonpoint Pathways to Waters BMPs effectiveness – watershed N reductions - cost/benefit tool BMP adoption constraints (social) X Past progress with existing programs – quantifying reductions by sector X X X X Agroecoregions • Minnesota has 39 agroecoregions which represent broad regional variations in soil, landscape, climate, and crop or animal management systems • Agroecoregions are finer scale geographic units than aquatic ecoregions or Major Land Resource Areas • Each agroecoregion has unique limitations to production, for example drainage, irrigation, erosion, precipitation, growing degree days • Each agroecoregion also has unique features that influence non-point source pollution, for example drainage, erosion, leaching, karst, sandy soils, etc Agroecoregion Based N Inputs • Point data are available for crop acreage and livestock numbers • County statistics are available for crop harvest and fertilizer sales • N transformations in soil (mineralization, denitrification) and N losses (volatilization, leaching, drainage, etc) are based on soil and landscape factors (represented by agroecoregions) • Our approach is to estimate N inputs and outputs for agroecoregion units and then transform results back to watershed units Methods - N Sources & Pathways INPUTS METHODS /SOURCE Net Mineralization Burkart and James (1999 Inorganic Fertilizer MDA, NASS, Bierman (2011) Atmospheric Deposition EPA Legume Fixation Russelle and Birr, 2004; Meisinger and Randall, 1991 Planted Seeds Meisinger and Randall (1991) Purchased Animals MDA-USDA,NASS 2010 Animal Feed MDA, Stuewe (2006) OUTPUTS METHODS/SOURCE Crop Removal NASS Senescence Burkart and James (1999) Denitrification Burkart and James, 1999; Meisinger and Randall, 1991 Runoff Tile Drainage Leaching to GW N losses based on extensive literature search SWAT models, Water balance, River discharge data, research data Based on Precipitation and N rate Based on N rate for four groundwater pollution zones differing in risk for leaching Fertilizer Volatilization Meisinger and Randall, 1991 Manure Storage Losses Midwest Plan Service MWPS-18 2004, Univ. of MN Extension Service 2001 Animals Sold Milk, Eggs, Meat MDA-USDA,NASS, 2010 Nass County data (weighted avg) 2005-2009, NASS, 2010 Tile Drainage For each agroecoregion we used an extensive research database to estimate drainage losses based on: Precipitation during the growing season for dry, average, and wet years N rate (sum of fertilizer + manure application) Tile drainage loss categories NO3-N losses for corn, corn silage, wheat, barley, oats, sugarbeets, potatoes NO3-N losses for soybeans NO3-N losses for alfalfa Tile Drainage-Nitrate Losses: Multivariate analysis NO3-N Leaching For selected agroecoregions with high leaching loss: Estimated the rate of NO3-N leaching for dry, and wet years, using research data based on N rate (sum of fertilizer + manure application) “Average year data” is mean value of dry and wet years 180 Dry years (431 mm) 100 y = 0.0602*Nrate + 22.245 R² = 0.0871 80 60 40 20 Cumulative NO3-N leaching (lbs ac-1) Cumulative NO3-N leaching (lbs ac-1) 120 Wet years (687 mm) 160 140 120 100 80 60 40 y = 0.2945*Nrate + 37.6 R² = 0.459 20 0 0 0 100 200 N rate (lbs ac-1) 300 400 0 50 100 150 200 -1 N rate (lbs ac ) 250 300 NO3-N Leaching Zones For other agroecoregions: We scaled the rate of NO3-N leaching according to the potential risk of NO3-N contamination of groundwater in each agroecoregion based on a water quality monitoring database of 40,000 drinking water wells Proportion of Wells Exceeding 3 mg/L Nitrate Nitrogen per Agroecoregion Percent 0 - 2.0 2.1 - 4.0 4.1 - 6.0 > 6.0 N 40 0 40 80 Kilometers Groundwater NO3-N Denitrification Factors (used to estimate groundwater nitrate discharge) ndwater denitrification factor assigned to different agroecoregions. Agroecoregion Denitrification factor Blufflands, Rochester Plateau 0.25 Anoka Sand Plains, Alluvium and Outwash, Inter-Beach Sand Bars, Steep Valley Walls, Steeper Alluvium. 0.40 Forested Lake Sediments, Mahnomen Lake Sediments, Poorly Drained BE Till, Poorly Drained Lake Sediments, Red Lake Loams, Somewhat Poorly Drained Lake, Swelling Clay Lake Sediments, Very Poorly Drained Lake Sediments 0.60 Other agroecoregions 0.50 Drained soils 0.60 Surface runoff • An extensive database of river monitoring was used to provide river discharges in dry, avg and wet years • SWAT modeling for the following areas was available to estimate the percent of discharge attributable to runoff 7 Mile Creek (Wetter Clays and Silts) Root River (Undulating Plains) Karst (Blufflands, Rochester Plateau) Red River (Swelling Clay lake sediments, Very poorly drained lake sediments) Sunrise Creek (Central till, Anoka Sand Plains, Alluvium and Outwash) • For the remaining agroecoregions, runoff percentages were estimated from the closest SWAT results based on a runoff classification of agroecoregions N Losses = Discharge * Runoff (%) * N Concentration in Runoff Nitrogen concentration in cropland runoff for each Agroecoregion. Agroecoregion N concentration (mg L-1) 1 Drift & Bedrock Complex, Forested Lake Sediments, Mahnomen Lake Sediments, Northern Till, Northshore Moraine, Peatlands, Poorly Drained Lake Sediments, Red Lake Loams, Somewhat Poorly Drained Lake, Steep Poorly Drained Moraine, Swelling Clay Lake Sediments, Very Poorly Drained Lake Sediments,Wetter BE Till, Wetter Clays & Silts 3.51 2 Central Till, Coteau, Drumlins, Dryer BE Till, Dryer Clays &Silts, Dryer Till, Forested Moraine, Inner Coteau, Mesabi Range, Poorly Drained BE =Till, Rolling Moraine, Steep Dryer Moraine, Steep Stream Banks, Steeper Till, Stream Banks 1.82 3 Bufflands, Inter-Beach Sand Bars, Level Plains, Steep Valley Walls, Steep Wetter Moraine, Steeper Alluvium, Undulating Plains 0.73 4 Alluvium & Outwash, Anoka Sand Plains, Rochester Plateau 0.244 Region Forest N Export • 2006 NLCD-Deciduous, Evergreen, & Mixed Forest • ~11 million acres statewide • N export coefficients: • 2 lbs ac-1 in average year Urban/Suburban Runoff • 2006 NLCD- Developed >20% impervious • ~1 million acres statewide • Avg N export coefficients: 2.9 lbs ac-1 for surface runoff 1.1 lbs ac-1 for movement to GW Septic Systems •Septic N based on county data from MPCA Septic N to Groundwater = [(# Septics per county) *(Persons per household by county) *({9.1 lbs N per person}*{85% for denitrification losses})] *(% NOT IPHT) Septic N to Surface Water = [(# Septics per county) *(Persons per household by county) *(9.1 lbs N per person)] * (% IPHT) •Weighted to 2008 ZIP code populations to improve spatial accuracy of county data (MSP excluded from analysis) Methods – Watershed N Reduction Decision Tool • The Decision Tool is an Excel spreadsheet linked to a database of Minnesota soils, landscapes, cropping systems, management practices and crop enterprise budgets • Estimates of N reductions are based on research meta-data and BMP specific reduction coefficients • Estimates are tied to site specific characteristics such as soil, slope, climate, and baseline farm management practices and cropping systems N Reduction Decision Tool BMPs • Rate and timing of N fertilizer • Controlled drainage • Bioreactors • Planting cover crops • Planting perennial grass • Installing riparian buffer strips • Installing wetlands • Effects of individual BMPs as well as combinations of BMPs can be evaluated N Fertilizer BMPs • Existing N rates can be reduced to target rates which average 117 lb/ac for fall application in a corn-soy rotation • Reductions in N loading are estimated based on empirical relationships derived from extensive research databases for tile drainage, leaching and runoff • Spring or sidedress N rates are 30 lb/ac lower than fall applications and reduce N losses by 8% compared to fall applications • Spring application costs an extra $7/ac, while sidedress costs an extra $50/ac • Costs of N fertilizer average $0.55/lb • Price of corn is assumed $6.00/bu Controlled Drainage BMP • Controlled drainage reduces N losses from treated area in tile drainage by 40% • Installation costs are estimated at $162/ac on 1% slopes • Annual repair and maintenance costs are $2.82/ac Bioreactor BMP • Each bioreactor treats 40 ac, and has an area of 471 ft2 • N reductions are 13% based on the assumption that each bioreactor treats 30% of the drainage system • Total annualized net present value to install, maintain and replace bioreactors is $440 Cover Crop BMP • Cover crops can be successfully grown one in five years • Rye seed costs $0.22/lb, aerial seeding costs $25/ac, killing cover crop costs $22/ac • Overall reduction in N loadings in drainage and leaching average 10% over a five year period Perennial Grass and Riparian Buffer BMPs • Rye seed costs $11/lb or $8/ac • Other costs are $36/ac, including $10/ac for fertilizer (e.g. 60 lb N/ac) • Reduction in N loadings arise partially from replacing annual crops that require higher rates of N fertilizer • N loadings from perennial grass plantings and riparian buffers are assumed to be negligible Wetland BMP • Wetlands are assumed to cover 2% of the upland contributing area treated • Costs to install wetland are $1,565/ac • Annual capital and maintenance costs are $103/ac • Reductions in N loadings from wetlands are assumed to be 50% Suitable acres for BMPs • Fertilizer rate reductions are only possible in areas where existing application rates exceed University recommendations • Controlled drainage and bioreactors can be installed on tile drained land with slopes of 0.5%, 1% or 2% • Perennial grass can be planted on ag land with crop productivity ratings of 60% or less (marginal land) • Riparian buffers can be installed on ag land within 30 m of waterways • Wetlands can be restored on tile drained land with hydric soils and high Compound Topographic Index values Controlled Drainage Suitable Acres 0- 5,700 5,800 - 18,000 19,000 - 33,000 34,000 - 62,000 63,000 - 100,000 0 25 50 Miles 100 Restorable Wetlands Suitable Acres 0- 4,300 4,400 - 14,000 15,000 - 33,000 34,000 - 58,000 59,000 - 110,000 0 25 50 Miles 100 Perennial Cropland Suitable Acres 0 - 5,200 5,300 - 14,000 15,000 - 33,000 34,000 - 92,000 93,000 - 230,000 0 25 50 Miles 100 Riparian Buffers Suitable Acres 0- 11,000 12,000 - 28,000 29,000 - 49,000 50,000 - 83,000 84,000 - 220,000 0 25 50 Miles 100 User Inputs and Model Outputs • Select watershed and type of climate of interest • Select types of BMPs to install • Select percent of suitable acres in watershed for installation of BMPs • Model estimates effectiveness of each BMP at reducing N loadings • Model estimates cost (per lb of N removed or per ac) of installing each BMP • Model estimates overall watershed scale effectiveness and cost of installing multiple BMPs Results • Nonpoint Source N Loadings to Surface Waters • Watershed N Reduction Decision Tool Agricultural N Inputs Agricultural N Outputs Minnesota N Balance (lb ac-1) Manure and Fertilizer Volatilization 14.1 Milk, Eggs 2.5 Senescence 37.3 Crop Removal 111.6 Animal Feed 38.6 Denitrification 26.8 Manure 20.0 Fertilizer 70.3 Runoff 0.8 Drainage 6.0 Leaching 8.6 Animals Sold 5.7 Deposition 11.3 Net Mineralization 89.4 Fixation + Seeds 31.6 2.0 Purchased Animals 1.6 N Loadings to Surface Water by Source Comparison between Predicted and Measured Average N Loads Nonpoint Source N Loadings by Source Effect of Climate on N Loadings N Reduction Decision Tool Average cost/ac (see line) Average cost/lb of N Reduced N reduction from Current Key Reduce N rate 20% Reduce N rate 20%, spring preplant N 5.2% Reduce N rate 20%, spring preplant N 5.2%, restore wetlands 2.7% Reduce N rate 20%, spring preplant N 5.2%, restore wetlands 2.7%, cover crops 15% Reduce N rate 20%, spring N 5.2%, buffers 2.9%, wetlands 2.7%, cont. drain. 2.3% N rate 20%, spring N 5.2%, buffers 2.9%, wetlands 2.7%, cont. drain. 2.3%, cover crops 15% Conclusions • Total nonpoint source N loadings to Minnesota surface waters were estimated at 254 million lb during an average climatic year. This is about 6% of the total inputs of N on all Minnesota cropland • Statewide, losses of N to surface water from agricultural sources represent 88% of total nonpoint source losses Agricultural N loadings to surface waters from groundwater and drainage are about equal and each far exceed runoff losses • Statewide loadings of N to surface waters from forest, urban and septics represent 12% of total nonpoint source losses • The Minnesota River Basin accounts for 34% of N loadings from nonpoint sources, the Lower Mississippi accounts for 21%, the Upper Mississippi accounts for 18%, and the Red River of the North accounts for 9% Conclusions – Nonpoint Source N Loadings to Surface Waters • A comparison between the modeled nonpoint source N • • • loadings to Minnesota surface waters (in an average climatic year) and monitored N loadings (average of two typical years) was conducted for 33 MPCA monitored major watersheds across Minnesota Monitored N loadings were not used to calibrate the modeled nonpoint source N loadings, as the modeled N loadings were estimated independently, without calibration Linear regression between modeled and MPCA monitored N loads was very good, with an R² value of 0.69 Modeled N loadings across all monitored watersheds were 10% higher than monitored N loads, which is not surprising given that additional losses in predicted N loadings may occur as nitrate travels downstream to the mouth of the watershed Conclusions – Nonpoint Source N Loadings to Surface Waters • Climate has a significant effect on nonpoint source N loadings • • • • • to Minnesota surface waters Total statewide nonpoint source N loadings to surface waters for dry, average and wet years were predicted to be 106, 254 and 409 million lb, respectively During a dry year, the majority (46%) of nonpoint source N losses to surface waters arises from groundwater discharge During an average year, the nonpoint source losses from agricultural drainage (45%) increase relative to the losses from agricultural groundwater discharge (37%) in comparison with the losses during a dry year During a wet year, the majority of nonpoint source N losses statewide arise from agricultural drainage (49%) Discharge of groundwater from agricultural regions contributes another 34% Conclusions – N BMP Decision Tool • A watershed based N BMP Decision Tool was developed to assist planners evaluate strategies for reducing N loadings to Minnesota surface waters • The Tool allows users to select a target watershed, climate, and extent of adoption of various N reduction BMPs • BMPs are limited by an analysis of acres suitable for implementation Conclusions – N BMP Decision Tool • The Tool estimates N loading reductions for individual practices • The Tool estimates cumulative N loading reductions for combinations of BMPs at the watershed scale • The Tool estimates costs associated with implementing BMPs • Cost/lb of individual practices • Cost/ac of individual practices • Net annual costs for implementing all BMPs in a selected watershed Conclusions – N BMP Decision Tool • BMPs that are suitable for implementation over larger areas generally give larger watershed scale N loading reductions than BMPs that are limited to implementation in smaller areas, even though the latter may have high N reduction efficiencies per acre • Approaches to achieving N load reductions greater than 25% are challenging Thank you • Support for this research was provided by MPCA