Dissolved organic matter dynamics in the boreal landscape mosaic: insights from Canada and Fennoscandia M.N. Futter, SLU Uppsala H.J. Laudon, SLU Umeå K.H. Bishop, SLU Uppsala P.J. Dillon, Trent University K. Rankinen, SYKE D. Rayner, Göteborg University D.N. Kothawala, Uppsala University P.G. Whitehead, University of Oxford A.J. Wade, University of Reading Talk Outline Surface water DOC from a mosaic of forest and mire landscape elements INCA-C: A dynamic model of organic carbon in a landscape mosaic Empirical testing of the landscape mosaic conceptual model Organic matter quality, colour and the landscape mosaic Future Climates Harp 4A Stream Dissolved Organic Matter in a Landscape Mosaic From Dillon and Molot (1997) www.the-colosseum.net/mages/MosaicNilo.jpg (c) Dolly Kothawala Another view of the landscape There is lots of data showing that catchments with a larger percent wetland export more DOM. Points on a regression of TOC export versus %wetland can be interpreted as a mixing model of TOC from forest and wetland landscape elements. From Dillon and Molot 1997 Export from a forest landscape element is equal to the regression intercept. Export from a wetland landscape element is equal to the regression intercept plus slope * 100% wetland. Another view of the mosaic The boreal landscape is comprised of forest, mire and surface water elements. Using the data from Dillon and Molot (1997), DOCExport = 2.39 + 0.261 * % Wetland Thus, DOCForest = 2.39 (2.39 + 0.261 *0) and DOCWetland = 28.5 (2.39 + 0.261*100) g/m2/yr Stream Mire Forest Mire Forest Mire Forest Stream Stream INCA Landscape and biogeochemical model The INCA modelling framework simulates a terrestrial biogeochemical processes in a landscape mosaic and subsequent surface water processing. Soil From Wade et al. 2002 Direct runoff Soil water Terrestrial process rates in INCA-C are positively dependent on soil temperature and moisture. Organic matter solubility is controlled by sulfate. Ground water Stream Controls on mass of soil solution DOC in INCA-C SMDMax − min (SMD, SMDMax ) dDOC × = Q (T − 20 ) Soil dt SMDMax (k SOC − (b [SO ] + k )DOC ) 2 − b1 D 0 4 q − DOC 86400 ⋅ v r + vd • • • • • • M Soil Temperature: Q(T-20) Soil Moisture: (SMDMax-min(SMD,SMDMax)/SMDMax Desorption of solid organic carbon: kDSOC: Sulfate mediated sorption: -b0[SO42-]b1DOC Mineralization: kMDOC Hydrologic Flux: DOC(86400 q/(vr + vd)) Modelling the mechanisms that control in-stream dissolved organic carbon dynamics in upland and forested catchments M. Futter, D. Butterfield, B.J. Cosby, P.J. Dillon, A.J. Wade and P.G. Whitehead 40 43 Modeled Measured Upper 95 Lower 95 38 33 Aug-2000 Jul-1999 Jan-2000 Jun-1998 Dec-1998 Oct-1996 Apr-1997 Nov-1997 Mar-1996 Sep-1995 Jan-1994 Feb-1995 Jun-1993 Aug-1994 Dec-1992 May-1992 Oct-1990 Apr-1991 Nov-1991 Mar-1990 Jul-1988 Feb-1989 -7 Aug-1989 -2 -10 Jan-1988 3 -5 Jun-1987 8 0 Oct-1985 13 5 Dec-1986 18 10 Apr-1985 23 15 May-1986 28 20 Sep-1984 25 Mar-1984 [DOC] mg/l 30 Prediction Range (mg/l) 35 Date INCA-C, the Integrated Catchments model for Carbon simulates soil and surface water dissolved organic carbon (DOC) concentrations as a function of climate, hydrology and acid deposition. The model, which operates on a daily time step, was developed for application to natural and seminatural catchments. It has been applied to catchments in Canada, Finland, Sweden, Norway and the UK. The impacts of future climate change and sulphur emission reductions on acidification recovery at Plastic Lake, Ontario J. Aherne, M. Futter and P.J. Dillon Changes in DOC will affect the rate at which ecosystems recover from acidification. We developed a model chain linking downscaled GCM climate projections to a rainfall-runoff model (HBV) which drove INCA-C projections. Modelled DOC output from INCA-C was used to drive long term MAGIC (Model of Acidification of Groundwater in Catchments) simulations. Increasing DOC concentrations and droughtinduced mobilisation of reduced sulfur are projected to delay recovery. 1600 calibration 60.0 acid neutralising capacity 50.0 precipitation (mm) 1400 40.0 1200 30.0 1000 20.0 10.0 800 0.0 600 –10.0 400 –20.0 200 catchment runoff (mm) 0 1960 25.0 1980 2000 2020 2040 2060 2080 –30.0 –40.0 1950 2100 1975 2000 calibration 5.0 dissolved organic carbon (mg L–1) 2025 2050 2075 2100 pH (pondus Hydrogenii) 4.9 20.0 4.8 4.7 15.0 4.6 4.5 10.0 4.4 4.3 5.0 4.2 temperature (°C) 0.0 1960 1980 2000 2020 2040 2060 2080 2100 Base Redox 4.1 4.0 1950 1975 2000 2025 2050 2075 2100 Modelling deposition and climate effects on DOC at Valkea Kotinen Peat #2 Forest Peat #1 Forest Lake Outflow Catchment map (upper left), lake (upper right), catchment outflow (lower left) and INCA-C catchment representation used in modelling Present-Day [DOC] 20 18 16 [DOC] (mg/l) 14 12 10 8 6 4 2 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 0 120 Date 100 [DOC] (mg/l) 80 60 40 20 Date Modelled (blue) and observed (grey) DOC in the lake (above) and catchment outflow stream (below) were simulated using present day (1990-2007) deposition and climate. Deposition and Climate Drivers of DOC 100 CLE D23 MFR 90 Simulated annual sulfate deposition (meq/m2/yr ) in southern Finland under currently legislated emissions (CLE, black), D23 (grey) and maximum feasible reductions (MFR, black) scenarios. 2 70 60 50 40 30 20 10 0 1860 1880 1900 1920 1940 1960 1980 2000 2020 2040 2060 2080 2100 12 Year A2 B2 10 Downscaled climate data were obtained from the PRUDENCE project 8 0 Average T ( C) 4 2 Year 2100 2090 2080 2070 2060 2050 2040 2030 2020 2010 2000 0 1990 Precipitation (not shown) is projected to be variable and with a small increasing trend. 6 1980 Annual average temperature at Valkea Kotinen under A2 (blue) and B2 (grey) scenarios. 1970 (0C) 1960 Sulphate Deposition (meq/m ) 80 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 0 Projected Daily [DOC] Daily modelled DOC in the lake (above) and stream (below) from 19612099 using parameter set from current-day calibration, MFR deposition scenario and SRES-A2 climate data Drivers of Change: Deposition and Climate 100 35 90 30 80 25 Frequency 60 50 20 15 10 40 5 20 0 -3 -2.8 -2.6 -2.4 -2.2 -2 -1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 30 10 Delta T 0 -180 -175 -170 -165 -160 -155 -150 -145 -140 -135 -130 -125 -120 -115 -110 -105 -100 -95 -90 -85 -80 -75 -70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 40 Deposition Change (meq/m2) 35 Frequency 30 Frequencies of change in deposition (above), temperature (top right) and precipitation (bottom right) resulting in a 1 mg/l increase in annual modelled [DOC]. 25 20 15 10 5 Delta Precip (mm) 400 380 340 360 320 300 280 260 240 220 200 180 160 80 140 60 120 100 0 40 20 -20 -40 -60 -100 -80 -120 -140 -160 0 -180 Less sulfate always leads to lower [DOC]. Modal values for climate suggest that warmer, wetter conditions will increase [DOC]. -200 Frequency 70 Annual DOC Flux from Lake (blue) and catchment outlet (grey) 9 8 6 2 Flux g DOC/m /yr 7 5 4 3 2 1 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065 2070 2075 2080 2085 2090 2095 2100 0 Date Modelled DOC flux from the lake and catchment outflow using the SRES B2 scenario and maximum feasible reductions of deposition (MFR). Areal exports from the lake are lower than catchment outlet because of in-lake losses. (Symbols represent the annual areal export and the lines are a 9-year running mean) Modelling seasonal and long-term patterns in stream dissolved organic carbon concentration in mire and forest dominated landscape elements at Svartberget, Sweden using INCA-C M. Futter, S.J. Köhler and K.H. Bishop An INCA-C model application was able to capture some TOC dynamics at the mire and catchment outflow but the overall quality of the simulation was a concern. The reasons for lack of fit are being addressed through the development of more appropriate models of stream flow generation and soil temperature and empirical testing of the “landscape mosaic” conceptual model. 50 70 Modelled Modelled Observed 45 Observed 60 40 [DOC] (mg/) 35 40 30 30 25 20 15 20 10 10 5 Date Date 01/2006 07/2005 01/2005 07/2004 01/2004 07/2003 01/2003 07/2002 01/2002 07/2001 01/2001 07/2000 01/2000 07/1999 01/1999 07/1998 01/1998 07/1997 01/1997 07/1996 01/1996 07/1995 01/1995 07/1994 01/1994 07/1993 0 01/1993 01/2006 07/2005 01/2005 07/2004 01/2004 07/2003 01/2003 07/2002 01/2002 07/2001 01/2001 07/2000 01/2000 07/1999 01/1999 07/1998 01/1998 07/1997 01/1997 07/1996 01/1996 07/1995 01/1995 07/1994 01/1994 07/1993 0 01/1993 [DOC] (mg/l) 50 Is there empirical support for the INCA ”landscape as a mosaic” conceptual model ? SVE Site M Kallkällsmyren Site 4 (Kryckaln) 19 ha 60% Forest/ 40% Mire SVV SiteV Västrabäcken Site 2 (Kryckaln) 13 ha 100% Forest SVW Site S Site 7 (Kryckaln) 50 ha 85% Forest/15% Mire 25 Jan 19,1994 TOC (mg/l) 20 15 There is a good steady state relationship between TOC export and % wetland. Is there a relationship between TOC and % wetland on individual dates within one catchment? y = 0.355x + 7.429 R² = 0.816 10 5 0 0 10 20 30 40 50 % Wetland 140 8 120 TOC in a landscape mosaic at Svartberget 7 Forest Mire NS 6 TOC concentration at Svartberget can be conceptualized as time varying contributions from forest and wetland landscape elements having the same unit runoff. Nov-03 Oct-02 May-03 Oct-01 May-02 Oct-00 Apr-01 Oct-99 Apr-00 Oct-98 Apr-99 Oct-97 0 Apr-98 -20 Oct-96 1 Apr-97 0 Oct-95 2 Apr-96 20 Oct-94 3 Apr-95 40 Apr-94 4 Oct-93 60 Apr-93 5 Oct-92 80 Nash Sutcliffe Statistic End Member TOC, mg/l 100 25 Jan 19,1994 Model fit is generally quite good (NS > 0.8) and there is a clear separation between forest and wetland TOC production. TOC (mg/l) 20 15 y = 0.355x + 7.429 R² = 0.816 10 5 0 0 10 20 30 % Wetland 40 50 Svartberget Landscape Mosaic model summary 70 The landscape mosaic approach is able to reproduce the observed data from SVV and SVE. It over-predicts TOC at SVW. (top), suggesting some in-stream losses. 60 SVV_Pred Modelled TOC 50 SVW_Pred SVE_Pred 40 30 20 A clear pattern emerges for monthly average TOC concentrations from forest and mire landscape elements (bottom). 10 0 0 10 20 30 40 50 60 70 Observed TOC This approach shows some value for understanding the behaviour of other elements, e.g. mercury 90 AvgOfForest 80 AvgOfMire TOC (mg/l) 70 60 50 40 30 20 10 0 1 2 3 4 5 6 7 8 9 10 11 12 Harp Streams DOC data Long term DOC data have been collected by Dillon and others from a series of headwater catchments in Central Ontario. Catchments have similar physiography but differing amounts of wetlands. Data from these catchments can be used to test the landscape mosiac approach. 5 4 6a 3a 6 3 Outflow 1 000 m 8 DOC concentration, 1983-1994 30 HP3 (Obs) 12 7 25 HP3A (Obs) HP4 (Obs) 10 6 20 8 5 4 15 6 3 10 4 2 5 0 0 18-Feb-82 2 1 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 35 0 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 25 HP5 (Obs) 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 35 HP6 (Obs) 30 18-Feb-82 30 HP6A (Obs) 20 25 25 15 20 15 20 15 10 10 10 5 5 5 0 18-Feb-82 0 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 18-Feb-82 0 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 Forest and Wetland End-Member DOC for Harp Streams 5 There is a clear separation between predicted DOC concentrations exported from forest and wetland end-members (below) 4 6a 3a 200 6 Wetland 3 180 Wood NS 160 Outflow 1 000 m DOC (mg / L) 140 8 DOC 6 120 100 80 60 40 4 2 20 y = 24.72x + 2.894 R² = 0.613 0 0 0 0.05 20-Mar-85 0.1 0.15 Wetland -20 0 15-Dec-82 28-Apr-84 10-Sep-85 23-Jan-87 06-Jun-88 19-Oct-89 03-Mar-91 15-Jul-92 27-Nov-93 21 Modelled (red) and observed (blue) DOC at Harp streams 5 4 6a 3a 6 3 Long-term DOC dynamics can be described as production from forest and wetland landscape elements. Including losses in surface waters (HP4) would improve model fit. Outflow 1 000 m 30 12 HP3 (Obs) 25 HP3A (Obs) HP3 25 10 HP4 (Obs) HP4 20 HP3A 8 20 15 15 6 10 4 5 2 0 0 DOC (mg /L) 10 18-Feb-82 14-Nov-84 35 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 HP5 (Obs) 5 0 18-Feb-82 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 25 18-Feb-82 35 HP6 (Obs) HP5 30 14-Nov-84 20 30 HP6 25 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 HP6A (Obs) HP6A 25 15 20 15 20 15 10 10 10 5 5 5 0 18-Feb-82 0 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 18-Feb-82 0 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 CDOM is increasing in northern Europe; is it the colour or the DOM (or both) ? Increased colour of drinking water supplies is a major concern in northern Europe (eg, colour has doubled in Oslo drinking water reservoirs). Increased colour may be a result of increased DOC input, or of the DOM becoming more coloured over time. Models able to predict changes in both DOM quantity and quality (colour) are needed. From Haaland et al. 2010 Colour:DOC ratios are not constant in Dorset lakes and streams While there are good long-term relationships between colour and DOM for lakes and streams (below), there can be large seasonal and between site variations (left). Assuming that colour:DOM ratios will remain constant in the future may not be justified. From Dillon and Molot 1997 Observed Colour of Harp Streams 5 Long-term colour records have been collected from the Harp streams. Colour (in Hazen units) is a measure of the difference in absorbance between 405-450 and 660-740 nm (from Dillon and Molot 1997). 4 6a 3a 6 3 Outflow 1 000 m 70 400 HP3 350 140 60 300 250 120 HP3A HP4 50 100 40 80 30 60 20 40 Colour, 1983-1994 200 150 100 10 50 20 0 0 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 500 0 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 18-Feb-82 350 450 300 400 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 500 HP6A 450 HP6 400 HP5 250 350 300 350 300 200 250 250 200 150 150 100 200 150 100 100 50 50 0 50 0 18-Feb-82 03-Jul-83 14-Nov-84 29-Mar-86 11-Aug-87 23-Dec-88 07-May-90 19-Sep-91 31-Jan-93 15-Jun-94 28-Oct-95 18-Feb-82 0 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 Colour: DOC Ratios from Harp forest and wetland landscape elements 18 Colour:DOC ratios were predicted using observed colour and endmember predicted DOC on each observation date. 14 12 10 8 6 4 2 Wetland Forest 0 1982 1984 1986 1988 1990 1992 30 1994 WetlandColourRatio WoodColourRatio 25 NS 20 Nash Sutcliffe Statistic Wetland DOC is more coloured than DOC from forests. There are seasonal and interannual patterns (some of which are statstically sgnificant, monthly Mann Kendall, p<0.05) in the colour of DOC exported from Harp forest and wetlands. Modelled Colour:DOC Ratio Colour:DOC ratio 16 15 10 5 0 -5 1983 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 Modelled (red) and observed (blue) colour in Harp Streams 5 4 Seasonal and inter-annual colour patterns can be simulated as a function of forest or wetland DOC and colour:DOC ratios. 6a 3a 6 3 Outflow 1 000 m 400 Colour, 1983-1994 160 HP3 350 300 120 250 100 200 80 150 60 100 40 50 20 HP4 250 HP3A_Modelled 14-Nov-84 11-Aug-87 07-May-90 500 HP5 450 HP5_Modelled 31-Jan-93 28-Oct-95 150 100 50 0 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 350 450 HP6_Modelled 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 HP6A HP6A_Modelled 400 250 300 18-Feb-82 500 HP6 300 400 350 HP4_Modelled 200 0 0 18-Feb-82 300 HP3A 140 HP3_Modelled 350 300 200 250 250 200 150 150 100 200 150 100 100 50 50 0 50 0 18-Feb-82 03-Jul-83 14-Nov-84 29-Mar-86 11-Aug-87 23-Dec-88 07-May-90 19-Sep-91 31-Jan-93 15-Jun-94 28-Oct-95 18-Feb-82 0 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 18-Feb-82 14-Nov-84 11-Aug-87 07-May-90 31-Jan-93 28-Oct-95 How might a changing climate affect future DOC dynamics? Haei et al. 2010 (in press) have shown that spring and summer soil solution [DOC] is a function of soil frost in the previous winter. How will soil frost dynamics change in the future ? Will soils be colder as a result of less snow or will there be less soil frost due to warmer temperatures ? 1300 Downscaled climate projections for Svartberget 1200 1100 1000 Precipitation (mm) 900 800 Rayner (in prep) has downscaled possible future climate at Svartberget. The precipitation downscaling routine is considerably more sophisticated than those used previously. 700 600 500 400 300 200 100 25 0 1960 1980 2000 2020 2040 2060 2080 2100 20 15 Precipitation (above) is projected to increase. Temperature (right) is also projected to increase. Modelled summer temperatures show a small increase, the greatest warming will occur in the winter months. Average Monthly T (C) 10 5 0 -5 -10 -15 -20 -25 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Implications for Winter Precipitation 700 Winter precipitation (October – April) is projected to increase (MK; p<1e-4) and a greater fraction will fall as rain (T < 0 during precipitation event). This will have profound consequences for streamflow. 600 Snow 500 400 300 200 100 2098 2094 2090 2086 2082 2078 2074 2070 2066 2062 2058 2054 2050 2046 2042 2038 2034 2030 2026 2022 2018 2014 2010 2006 2002 1998 1994 1990 1986 1982 1978 1974 1970 1966 0 1962 Oct – April Precipitation (mm) Rain Soil Temperature Modelling 2 0 We developed a new model based on Rankinen et al. (2004) predicting soil temperature at discrete depths from air temperature and precipitation . The model simulated snow pack aging, heat exchange to the surface and deep in the profile and soil freezing effects. It was calibrated to winter (Tsoil < 2 0C, NS=0.51) conditions at Svartberget. -1 -2 -3 Modelled -4 Observed 01/2002 07/2001 01/2001 07/2000 01/2000 07/1999 01/1999 07/1998 01/1998 07/1997 01/1997 07/1996 01/1996 -5 Projected Soil Temperature at Svartberget 1 300 Minimum T 0 250 -1 200 -2 150 -3 100 -4 50 0 -5 1960 1980 2000 2020 2040 2060 2080 2100 Days Soil T < 0 Days < 0 Minimum Soil T (C) 07/1995 Soil Temperature at 11 cm 1 Projected Snow Dynamics at Svartberget Days with Snow / Snow depth (SWE mm) 250 Days with Snow on Ground Average Snow Depth (SWE, mm) 200 150 100 50 0 1960 1980 2000 2020 2040 2060 2080 2100 Projected trends in Rain on Snow Events at Svartberget 60 Rain on Snow Events 50 40 30 20 10 0 1960 1980 2000 2020 2040 2060 2080 2100 Summary We’re developing better tools to downscale GCM projections and to model the effects of climate change on biogeochemically relevant catchment factors (eg. soil temperature, snow cover). This will be helpful in predicting not only changes in DOM concentration but also potential changes in quality (i.e. Colour). Harp 5 Stream We have demonstrated the value of using the landscape mosaic (eg. forest/mire) as a conceptual model of DOM dynamics in the boreal. These insights are especially useful for assessing future threats to drinking water quality in northern Europe.