ASSESSMENT OF HEAVY METAL FLUXES IN A FOREST WATERSHED IN BASQUE COUNTRY (NORTHERN SPAIN) USING SWAT MODEL PERAZA-CASTRO1,2*, M., RUIZ- ROMERA1, E., SAUVAGE3,4, S., SÁNCHEZ-PÉREZ3,4, J.M 1 Department of Chemical and Environmental Engineering, Faculty of Engineering of Bilbao, University of the Basque Country (UPV-EHU), Alameda de Urquijo s/n, 48013 Bilbao, Spain 2 School of Health Technologies, Faculty of Medicine, University of Costa Rica, Rodrigo Facio Campus, San Pedro de Montes de Oca, San José, Costa Rica 3 University of Toulouse, INPT, UPS, Laboratoire Ecologie Fonctionnelle et Environnement (Ecolab), Avenue de l’Agrobiopole, 31326 Castanet Tolosan Cedex, France 4 CNRS, Ecolab, 31326 Castanet Tolosan Cedex, France

e-mail: [email protected]

Scientific context: Nowadays, the main threat to the water body’s damage comes from non-point sources of pollution, as result of intensive agriculture and urban development (Boskidis et al., 2012). All of these sources, sediment represents the highest volume for weight of material transported to the sea. Metals can be transported in association with the sediment (adsorbed) or in solution (soluble contaminants) (FAO, 1993, Boithias et al., 2012). The fine sediment is important vector for pollutants transport such as heavy metals (Ankers et al., 2003), since it is a material highly enriched. During flood events, a significant proportion of heavy metal content from bottom sediments may be resuspended and transported to coastal zones, affecting negatively on the environment. In this sense, modeling is useful in assessing the impact of agricultural management and land use changes on water, sediment yield and pollutants as metals without altering the physical environment in the catchments.

Objective: To quantify annual particulate metal fluxes associated to suspended particulate matter (SPM) from a small forestry catchment to the Bay of Biscay (Northern Spain) during eleven hydrological years (2001-2012). Study area and data Methodology ♦ Annual fluxes of particulate metals were 3 1 The Oka River is located in the Basque Bay of Biscay ♦ Gauging station Flow (m s ) 10 min estimated using Waling and Webb method Country-northern Spain, within Urdaibai Turbidity (NTU) 10 min (Walling and Webb, 1985): Biosphere Reserve (declared in 1984 by SPM(mg l-1) 10 min UNESCO). Drainage basin: 31 km2 Elevation: 13 m to 605 m ♦ Water samples collected manually during Average precipitation: 1205 mm eight flood events occurred in 2009-2012 V (m3): annual water discharge, Average discharge: 0.64 m3 s-1 were carried to Chemical and Environmental Ci (µg l-1): particulate metals concentration, Vegetation: pasture, pinus, eucalyptus Engineering Laboratory (UPV/EHU) to (m3 s1): instantaneous river water flow and Main bedrock: Calcareous Flysch determinate particulate metal concentration n: number of measures. (Cu, , Pb, Cr, Zn, Al, Fe, Mn).

Modelling approach ♦ SWAT model was applied to asses the temporal variability of discharge, SPM and Ni during 2001–2012 at daily scale in Oka catchment. The details of calibration and validation can be consulted in our earlier paper Peraza-Castro et al., 2014. ♦ To evaluate the model performance between simulated and observed annual metal fluxes the following efficiency criteria were used: coefficient of determination (R2), volumetric efficiency (VE) and index of agreement (d). VE represents the fraction delivered at the proper time; its compliment represents the fractional volumetric mismatch (Criss et al., 2008). d represents the ratio of the mean square error and the potential error (Willmot, 1982). Both ,VE and d, were proposed to overcome the insensitivity of R2 and Nash-Sutcliffe Efficiency. Results and discussion Particulate metal concentration Regression equations R² ♦ With SPM and particulate metal concentrations obtained in eight flood events at the outlet, linear (pMe) regression model were employed to express its relationships. pCu (µg l-1) 0.0268*SPM + 1.1826 0.84 pNi (µg l-1) 0.1843*SPM0.6935 0.83 ♦ It was found that SPM correlated well with metals (R2>0.62). pPb (µg l-1) 0.0317*SPM + 3.3149 0.62 pCr (µg l-1) 0.0909*SPM + 2.5897 0.92 ♦ Based on these relationships, the long term observed and simulated metal concentrations could pZn (µg l-1) 0.135*SPM + 11.753 0.70 be computed from SPM. pAl (mg l-1) 0.0988*SPM0.9557 0.97 pFe (mg l-1) 0.038*SPM + 0.4693 0.92 pMn (µg l-1) 0.7004*SPM + 10.168 0.90 Simulated daily SPM and Ni ♦ Ni and other metal concentrations follows the SPM trend. In this example, simulated Ni ranged Simulated SPM Simulated Ni between 0.2-11.1 µg l-1, with a mean of 0.8 µg l-1 comparable with the observed mean of 600 0

-1 500 5

0.9 µg l . 1

- 400 10

1 300 Validation Calibration 15 -

200 20 SPM mg l mg SPM 100 25 µg l Ni 0 30 Particulate metal annual fluxes

Date (day)

Simulated and Observed annual Cu yield Simulated and Observed annual Cr yield Simulated and Observed annual Ni yield Simulated and Observed annual Pb yield Observed Cu yield Simulated Cu yield Observed Cr yield Simulated Cr yield Observed Ni yield Simulated Ni yield Observed Pb yield Simulated Pb yield Calibration 100 Validation Calibration 250 Validation 100 160 Validation Calibration R2=0.52 Validation Calibration R2=0.50 R2=0.63 R2=0.45 R2=0.52 140 R2=0.69 VE=0.53 R2=0.61 R2=0.50 VE=0.81

80 VE=0.84 VE=0.77 200 VE=0.73 80 VE=0.87

120

VE=0.81 d=0.69 VE=0.76 d=0.45 d=0.88 d=0.50 d=0.69 ) d=0.90

d=0.87 d=0.50 100

(kg) (kg) 60 (kg) 150 60

80

yield (kg yield yield yield 40 yield

100 40 60

Cr Cr Ni Pb

Cu yield yield Cu 20 40 50 20 20 0 0 0 0

Date (Year) Date (Year) Date (Year) Date (Year)

Simulated and Observed annual Zn yield Simulated and Observed annual Fe yield Simulated and Observed annual Al yield Simulated and Observed annual Mn yield

Observed Zn yield Simulated Zn yield Observed Fe yield Simulated Fe yield Observed Al yield Simulated Al yield Observed Mn yield Simulated Mn yield Calibration 600 Validation 2 100000 Calibration 180000 Validation Calibration 1800 Validation Calibration R =0.53 Validation R2=0.68 90000 R2=0.53 160000 R2=0.50 R2=0.52 1600 R2=0.55 R2=0.48 VE=0.80 R2=0.54 500 VE=0.86 VE=0.80 VE=0.71 VE=0.70 VE=0.77 VE=0.72 d=0.63 80000 VE=0.76 140000 1400

d=0.90 d=0.55

d=0.63 d=0.77 d=0.57 d=0.81 70000 d=0.80 )

400 120000 1200

(kg) (kg) (kg) 60000 100000 1000

300 50000 yield yield

yield(kg 80000 800

yield yield

yield yield 40000

Al Al

Fe Fe Zn 200 30000 60000 Mn 600 40000 400 100 20000 10000 20000 200 0 0 0 0

Date (Year) Date (Year) Date (Year) Date (Year)

Mean specific (kg km-2 y-1) metal yield during total period ♦ Exportation order of particulate metal yield: Al>Fe>Mn>Zn>Cr>Pb>Cu>Ni. Cu Cr Ni Pb Zn Fe Mn Al ♦ The years 2008/09-2009/10 produced the highest metal load while 2001/02 generated the lowest. Observed 1.6 4.7 1.6 3.2 12.8 1562 29.8 2682 Simulated 1.5 4.5 1.5 2.9 11.2 1544 29.3 2702 ♦ The mean specific annual of particulate metal export was compared with other European and American Specific particulate metal fluxes (kg km-2 y-1) of other catchments rivers with similar land use and the results are very similar each other; except for Pb and Zn that are River Author Country Land use Cu Cr Ni Pb Zn Fe Mn Al slightly higher, which is attributed to the slight industrial and agricultural activity in the area. Mero Palleiro et al., 2014 Spain Agroforestry 0.3 1.6 361 15.4 319 Corbeira Soto-Varela et al., 2014 Spain ♦ Al, Fe and Mn are more similar to the basins with agricultural landuse. These metals are major Agroforestry 0.3 1.6 260 8.3 205 Chester Branch Miller et al., 2003 EEUU Agricultural 0.4 1.8 1 0.7 3.3 1090 79 1550 compounds soil. Montoussé Roussiez et al., 2013 France Agricultural 1.6 4.1 2.6 1.1 7 1692 40 2416

ConclusionsCONCLUSIONS References ♦ With a good relationships between metal and SPM measured in field and a satisfactory SPM ♦ Criss, R.E., Winston, W.E., 2008. Do Nash values have value? Discussion and alternate proposals. Hydrol. Process. 22, 2723-2725. ♦ Jiao, W., , W., Hao, F., , H., , Y., , X., 2014. Combine the soil water assessment tool (SWAT) with sediment calibration by SWAT, is possible to compute the long term metal concentrations at daily time step. geochemistry to evaluate diffuse heavy metal loadings at watershed scale. J. Hazard. Mater. 280, 252-259. ♦ The statistical criteria indicate a fair and satisfactory annual metal fluxes simulation. ♦ Peraza-Castro, M., Ruiz-Romera, E., Montoya-Armenta, L.H., Sánchez-Pérez, J.M., Sauvage, S., 2015. Evaluation of hydrology, suspended ♦ Swat model is useful to estimate annual fluxes of pollutants which have punctual measures. sediment and Nickel loads in a small watershed in Basque Country (Northern Spain) using eco-hydrological SWAT model. Ann Limnol - ♦ Oka catchment showed a high annual metal fluxes variability during period study, which is related Int J Lim 51, 59-70. with sediment load. ♦ Annual simulated metal fluxes (kg) ranged from: 32.5-69 for Cu, 93.2-198 for Cr, 30.7-68.4 for Ni, ♦ 57-130 for Pb, 220-494 for Zn, 31525-68464 for Fe, 600-1297 for Mn and 53994-121724 for Al.