ModellingModelling DiffuseDiffuse PollutionPollution onon WatershedsWatersheds UsingUsing aa GisGis--linkedlinked BasinBasin ScaleScale HydrologicHydrologic/water/water QualityQuality ModelModel A. Azzellino1, M. Acutis2, L. Bonomo1, E. Calderara1, R. Salvetti1 and R. Vismara1 1. Politecnico di Milano (Technical University), D.I.I.A.R. – Envir. Division, Piazza Leonardo da Vinci, 32 – 20133, Milano, Italy 2. Milano University, Crop Science Dept. (Di ProVe), Italy BackgroundBackground • Diffuse pollution enters the environment by multiple pathways. • Non point sources are generally estimated by using area- and pollutant- specific emission factors, function of land use (e.g. Autorità di Bacino del Po) • Despite the different temporal scale, instream direct measurements, when available, have often suggested that non-point indirect estimates may lead to large overestimations or to misleading results . Politecnico di Milano – DIIAR – Env. Div. AimsAims ofof thethe StudyStudy • To combine a water quality model, (USEPA- QUAL2E), with a GIS-linked physical-based model (SWAT) inin orderorder toto assessassess thethe sourcesource apportionmentapportionment ofof pointpoint andand nonnon--pointpoint emissionsemissions to the global quality of the rivers. • to simulate the factors driving diffuse pollution in large and complex watersheds. • To validate the models by comparing their predictions over a regional scale with instream direct measurements available from ARPA monitoring stations. Politecnico di Milano – DIIAR – Env. Div. AimAim ofof thethe StudyStudy:: ValidationValidation ofof thethe sourcesource apportionmentapportionment • Validation studies on diffuse pollution are not easy to design – no readily available tool enables to simulate the river water quality during rainfall events at large scales (i.e. basin-scale) and over the instantaneous temporal scale which is typical of water quality direct measurements • Average annual basis – Dry-weather scenario (i.e. point emissions over the mean annual flow regime) – Wet-weather scenario (i.e. point and non point emissions during rainfall events) Politecnico di Milano – DIIAR – Env. Div. StudyStudy AreaArea LombardyLombardy Cherio river Watershed 60 km Mincio river Watershed Politecnico di Milano – DIIAR – Env. Div. CherioCherio RiverRiver WatershedWatershed • Drainage area: 150 km2 • Watershed covers both the mountainous and the plain zone (approximately half and half of the percentage area) • Dominant coltures: • Maize (59%), Wheat (30%) Other (11%) Politecnico di Milano – DIIAR – Env. Div. MincioMincio RiverRiver WatershedWatershed • Drainage area: 310 km2 Garda Lake • Plain zone within the Po Mincio River and Mincio Rivers, Mantua dominated by a very Lakes intensive agricultural land use • Dominant coltures: Po River • Maize (70%), Wheat (30%) Politecnico di Milano – DIIAR – Env. Div. QUAl2EQUAl2E ModelingModeling Reach3Tratto 3 Reach1Tratto 1 Cherio a Casazza Reach2Tratto2 T. Drione Reach4Tratto 4 T. Grone Reach8Tratto 8 Reach5Tratto 5 T. Tadone TrattoReach6 6 T. Malmera ReachesTratto 9 9-10–10 Cherio Reach11Tratto 11 T. Tirna Reach11Tratto 12 Cherio a Palosco ASSUMPTIONS • Constant streamflow (i.e. Median of 3 years monthly measurements) • Constant emissions (point loads calculated on the basis of annual emission factors) Politecnico di Milano – DIIAR – Env. Div. InputInput forfor PointPoint EmissionsEmissions Point Sources Emission Factors and Average Treatment Removals used to estimate the domestic wastewater loads. A.S + N A.S. + N, P Pollutant Emission Imhoff Removal A.S.* Removal Removal Removal %% %% -1 -1 BOD5 60 g BOD PE d 25 90 92 92 COD 129 g COD PE-1 d-1 25 85 85 85 Total-N 12.3 g N PE-1 d-1 15 35 65 65 Total-P 1.8 g PE-1 d-1 20 35 35 85 *A.S.: Activated Sludge PE: Population Equivalent Politecnico di Milano – DIIAR – Env. Div. QUAL2EQUAL2E ValidationValidation Mean Annual Dry-weather Scenario Cherio River Direct Measuremets SWAT simulation Qual2E Simulation • Point load effect simulation 4.0 was validated for every 3.0 parameter considered by 2.0 mg/l comparing simulated values 1.0 with the median of 3 years 0.0 monthly monitoring N-NH4 N-NO3 P-PO4 P-tot campaign (ARPA) Direct measurements SWAT Simulation QUAL2E simulation 1.5 • Then the same emission 1.0 mg/l factors for point loads, 0.5 validated with QUAL2E, 0.0 were used for the SWAT N-NH4 N-NO3 P-PO4 modeling Mincio River Point loads control the river quality even in watershed characterized by extensive agriculturalPolitecnico di Milano land – DIIAR use – Env. Div. SWATSWAT SIMULATION:SIMULATION: INPUTS •• DEMDEM gridgrid at a 40 m • Weather database spacing – generated on the basis of site specific average monthly temperature and • ERSAL maps on landland useuse precipitation parameters and soilsoil characteristicscharacteristics. (i.e.min,max, std deviation) – ERSAL land use codes were translated into the • Three lakes: SWAT classes of land • Three lakes: cover – Endine and Garda Lakes headwaters of Cherio and Mincio rivers – Mantoa lakes within Mincio Watershed. Politecnico di Milano – DIIAR – Env. Div. SWATSWAT SIMULATION:SIMULATION: Irrigation network Irrigation network Politecnico di Milano – DIIAR – Env. Div. SWATSWAT SIMULATION:SIMULATION: METHODS Irrigation • The plain zone of the study area, network and particularly the Mincio Basin area, is characterised by a very complex system of irrigation channels that overlays the natural hydrographic network. Pseudo-Hydrography •A pseudo-hydrography was Generated by the generated by means of the Watershed SWAT WD tool Delineation tool (“Burn In” option). • Such a simplification reflects reasonably the average pathways of the diffuse pollutants to the main stream Politecnico di Milano – DIIAR – Env. Div. ResultsResults Cherio River SWAT simulations NO3-N NH4+-N Ptot 2500 1000 • Diffuse load is much lower in 2000 800 Cherio than in Mincio Basin 1500 600 1000 400 (kg/month) Diffuse NO3-N load 500 200 Diffuse load NH4+-N • Diffuse loads increase 0 0 (kg/month) Ptot and 0246810121416 downstream SWAT subbasin • The presence of Mantoa lakes NO3-N NH4_N P-PO4 within Mincio basin smoothes 400000 20000 the downstream trend 300000 15000 200000 10000 (kg/month) 100000 5000 Diffuse NO3-N load Diffuse load NH4+-N NH4+-N load Diffuse and Ptot (kg/month) Ptot and 0 0 7 21244256637074818791 SWAT subbasin Mincio River Mantoa Lakes Politecnico di Milano – DIIAR – Env. Div. ResultsResults Comparison with the indirect emission estimates approach • QUAL2E has been used to simulate the water quality corresponding to streamflows higherMonthly than trend the of 75° non percentile.point loads 250 20 •200 Non point loads were either estimated by means of SWAT modeling or by 15 ) using AdBPo indirect emission factors 150 10 Pollutant100 Arable Grasslands Forest Other load Nitrogen (mm rain percolation and runoff kg N (ha-1 y-1) 20 – 30 5 – 7 3 –10 5 -1 -1 5 Phosphorous50 erosion and runoff kg P (ha y ) 0.5 - 1.3 0.3 – 0.5 0.2 – 0.3 0.2 –0.5 % of the total annual 0 0 • Point loadsApril were May kept constant June although July diluted August of September25% October • AdBPo diffuse loads, being evaluated on an annual basis, were distributed following the monthly precipitation trend over the monthly average of rainy days. Similarly the SWAT predictions were distributed over the same interval of rainy days. Politecnico di Milano – DIIAR – Env. Div. ResultsResults Comparison with the indirect emission estimates approach Cherio River Direct Measurement SWAT simulation AdBPo Direct Measurement SWAT simulation AdBPo 0.14 6 0.12 5 0.10 4 0.08 3 0.06 2 P-tot (mg/l) 0.04 N-NO3 (mg/l) 1 0.02 0.00 0 Direct Measurement SWAT simulation AdBPo Direct Measurement SWAT simulation AdBPo 0.40 14 0.35 12 0.30 10 0.25 8 0.20 6 0.15 4 P-tot (mg/l) 0.10 N-NO3 (mg/l) 2 0.05 0.00 0 Mincio River Indirect emission factors may lead to remarkable overestimations Politecnico di Milano – DIIAR – Env. Div. ConclusionsConclusions • The integration of QUAL2E and SWAT enabled to quantify overestimations of the diffuse loads contribution • It offers the opportunity to adjust the source apportionment of point and non point contributions to the instream total load within the SWAT model. • SWAT predictions resulted to be much more accurate than the indirect emission estimates for both the studied watersheds. Politecnico di Milano – DIIAR – Env. Div..