6. Modelling the Impacts of Construction and Operation of Wind Farms on Peat
The recently completed ECOSSE project developed a dynamic and functional model that can be used at a range of scales to simulate C and nitrogen turnover in both highly organic and mineral soils (Smith et al. 2007). The impacts of the construction and operation of wind farms on organic soils has been modelled using a combination of ECOSSE and a C balance approach. This provides a full life cycle analysis, focussing on the potential impacts of peat removal, disturbance, drainage and dissection over both the lifetime of the wind farm and the longer term impacts on the peat land.
However, to run simulations of C and nitrogen turnover at an individual wind farm would require detailed site maps of soils and hydrology. This is not feasible in many instances due to the shortage of experimental data at the site. A set of transparent equations are required that will allow the losses of greenhouse gases (in CO 2 equivalents) to be estimated using amalgamated whole site measurements. These site-specific equations were derived from experimental data, and it was planned that they would be refined as needed using information from ECOSSE simulations. In practice, the association and coincidence between the derived equations and experimental measurements for the subset of losses from acid bogs and fens was highly significant. Therefore, there was no need to refine the equations using ECOSSE.
6.1. Practices that change carbon emissions from peat lands
6.1.1. Borrow pits
Construction of site roads, hard-standing, foundations and other structures require large quantities of stone ( i.e. approximately 2208-3500 m 3 stone per turbine and 5850-7708 m 3 stone per km of road, according to Entec UK Limited,2002, and Lewis Wind Power,2004). Unless an alternative source is available, it is normal practice to open borrow pits ( i.e. a pit or small quarry, dug to provide stone) to obtain materials for track construction on wind farm sites. Such pits require drainage, and so cut off ditches (surface ditches designed to divert water from the surface of the construction site) are installed to intercept surface water. On completion of the work, restoration of the borrow pits is usually done by the wind farm developer ( AMEC, 2005). This may be done by reinstating peat from the borrow pit overburden as well as from other parts of the site, such as peat excavated to construct roads and foundation areas. The impact of digging the borrow pit, and its restoration, should be included in the calculations of C loss from the site. For example, if it is restored by capping the surface with peat, subsequent emissions of C from the disturbed peat should be estimated on the basis of the restored hydrological and habitat conditions. Similarly, the impact of the temporary cut-off ditches on the peat land hydrology and C emissions should be estimated.
6.1.2. Access tracks
Depending on the conditions of the site, three different construction techniques are commonly used to build roads ( AMEC, 2005). The three road types are:
- Rock filled
Excavated roads are constructed on shallow peats (peat depth < 1m) with a steep cross slope (<3 0) (Lewis Wind Power, 2004). Excavation for this type of road is usually taken down to competent weight bearing strata or to the bedrock. Excavated roads are lower than the topsoil, and so have associated drainage, such as drainage ditches and check dams (constructed across the drainage ditch to lower the speed of concentrated flows) (Lewis Wind Power, 2004). The drainage around the excavated road represents a permanent lowering of the water table. The extent of the reduction in the water table around the drain and its likely impact on C emissions should be estimated.
The size of C emissions coming directly from the excavated peat depends on the treatment of the peat following excavation. If peat is left on the surface without restoration of the water table and habitat, it can be assumed that, over time, the losses will include all of the C content of the peat. If, however, excavated peat is preserved as part of the habitat restoration programme, only a proportion of the total C content may be lost before the peat C is stabilised.
On deep peats (peat depth > 1.5m), floating roads can reduce or avoid damage to the peat land soils, hydrology and ecosystems, minimising the need for removal of peat (Entec UK Limited, 2002, Lewis Wind Power, 2004). Floating roads are built on top of the peat and the road surface protrudes above the surrounding peat land. Normally drainage is not required, but sometimes it may become necessary to ensure that surface water flow is not impeded (Lewis Wind Power, 2004). If this is the case, the changes in hydrology associated with drainage of the floating road and its impact on C emissions should be estimated as for excavated roads.
Rock filled roads
Rock filled roads are constructed on shallow peat (depth < 1m) and deep areas of wet, unstable peat (depth > 1.5m) with a steep slope (< 20 0) perpendicular to the direction of travel (Lewis Wind Power, 2004).The amount of rock fill required is dependent on the depth of the peat, decreasing with decreasing depth. The bulk of the material displaced is water, which drains away without the need for further drainage channels ( AMEC, 2004). The impact of compaction of the peat on C emissions should be estimated. This can be done using experimental measurements of C losses from compacted peats, collated from the literature. As for floating roads, rock filled roads protrude above the surrounding peat land and so drainage is not normally required (Lewis Wind Power, 2004), but may become necessary in some cases to ensure surface water flow is not impeded.
6.1.3. Foundations for turbines and crane pads
Foundations for crane pads
Each wind turbine requires an area of hard-standing adjacent to the foundations (Entec UK Limited,2002)). This provides a stable base for turbine components during assembly and erection, and an area to site the cranes necessary to lift the tower sections, the nacelle and rotors into place. Hard-standing construction is carried out in conjunction with that of the foundations. Excavation for the crane pads is taken down to competent weight bearing strata or to the bedrock. The crane pads are usually left in place following construction to allow for replacement of components during the operation of the wind farm, and for decommissioning at the end of the wind farm lifetime.
Foundations for turbines
Each wind turbine requires a reinforced concrete foundation comprising a square base, typically (22mx22m) or (15mx15m) by 1-2m deep (Entec UK Limited,2002; Lewis Wind Power, 2004). The construction methodology for the wind turbine foundations depends on the stability and depth of the peat, and is specific to each location. Foundations need to be taken down to competent weight bearing strata, which normally requires excavation through the peat to either bedrock or glacial till (Lewis Wind Power, 2004). If the peat is shallow and stable, standard excavation techniques can be adopted (Lewis Wind Power, 2004). However, if peat is deep or unstable, a rock cofferdam (an enclosure beneath the water level, constructed to allow water to be displaced by air) is required to retain the peat and prevent it from flowing back into the excavation (Lewis Wind Power, 2004). The C emissions coming directly from the peat excavated from turbine foundations can be estimated similarly to emissions associated with peat excavated from roads and crane pad foundations.
6.1.4. Construction of the substation and temporary compounds
The area of the building for the substation is prepared by removing peat down to competent weight-bearing strata. This requires excavation through the peat to either bedrock or glacial till (Lewis Wind Power, 2004).Temporary compounds are constructed within the site and used as secure storage for construction materials. Temporary compounds are also constructed to provide temporary accommodation units. For construction of the temporary compound and substation, peat is excavated. However, this is usually small (for example at the planned Lewis site, 76,000 m 3) compared to that excavated for road construction (1454,000 m 3) and turbine foundation (1000,000 m 3) (Lewis Wind Power, 2004), and so will be omitted from the calculations.
6.1.5. Cable trenches
The typical method for laying cable on wind farms requires a trench (typically depth 0.5 m to 1m, width 1 m) (Entec UK Limited,2002; Lewis Wind Power, 2004). The cable trench usually runs beside access tracks. The cables are laid on a bed of sand. This method disturbs a large area of peat, comparable to the area disturbed by construction of access tracks. Furthermore, it could enhance drainage of the peat, by creating preferential flow pathways (Entec UK Limited, 2002). It is also possible that vegetation changes will occur along the trench. Colonisation of soft rush along cable trenches has been observed (see for example, the Environmental Statement for Black Law, North Lanarkshire, Entec UK Limited, 2002). Cable trenches should be included in the calculations especially where they do not run alongside access tracks.
6.1.6. Forestry operations
The potential effects of the wind farm development on forestry are:
- Removal of commercial forest area;
- Disturbance of ground by machinery;
- Tree debris and mulch remaining on site;
- Increase in the availability of water to the catchment.
Forestry plantations may need to be felled to improve the performance of the wind farm, and this will result in changes in the runoff regime at the site. The removal of trees reduces rainfall interception and evapotranspiration rates leading to a possible increase in runoff and stream flow. The precise effects of tree removal on runoff in upland areas are site specific, and would require detailed hydrological modelling for an accurate assessment to be made. Therefore, these effects will not be included in the current study. However, changes to the drainage regime, resulting in changes in C emissions from the soil will be included. The Forests and Water Guidelines, adherence to which is one of the requirements of sustainable forest management in the UK, provide advice on how to minimise runoff and should always be referred to in these circumstances.
6.1.7. Restoration of site after decommissioning
The environmental statement of a wind farm includes proposals for restoration of the site at the end of the operational time (Entec UK Limited, 2002, AMEC, 2005). The decommissioning process involves
- Dismantling wind turbines and transformers;
- Burying or clearing of the wind turbine foundations;
- Demolition of substations and compounds; and
- Restoration of all land affected, in accordance with best practice.
Wind turbines are removed using a crane. Turbine foundations and a hard standing area are broken up and cleared to a level from 0 to 0.5 m below ground level. Ground cover is reinstated and revegetated (Entec UK Limited, 2002; AMEC, 2005). Complete removal of structures would cause further disturbance of the peat surface. Therefore leaving them in place may be preferable to complete removal. In time, concrete will deteriorate and vegetation such as mosses and lichens will colonise cracks, and tracks will sink into the peat and become vegetated to some extent (Entec UK Limited, 2002). As already described in the above sections, peat land can become a source of CO 2 during the construction and operation of the wind farm, due to drainage resulting in a lowered water table, increased respiration and the loss of CO 2 fixing vegetation. Full restoration of the site should return the water table to its original level, potentially halting the increased C losses. If this is achieved, CO 2 emissions need only be accounted for up to the time of full restoration. To encourage recovery of the bog habitat, any drains should be blocked in order to assist rewetting of soils. The restoration plan for a site requires expert input and site specific strategies.
6.1.8. Improvement of altered peat lands as part of the habitat management plan
To ameliorate the potential impacts of wind farm development on the peat land habitat, proposals to restore areas of damaged peat land to its native state are often included in a habitat management plan. If successful, habitat improvement of these altered peat lands may reduce C emissions from the site. For a degraded peat land, the primary goal is to repair the ecosystem, if the damage is not too severe, or to rebuild it if there has been considerable loss of peat. This will require the normal peat land hydrology to be restored, allowing biogeochemical cycling, energy capture and eventually autogenic plant succession and the renewal of peat accumulation (Glaser and Janssens, 1986; Foster and Wright, 1990). Habitat improvement processes involve the blocking of drainage ditches to raise the water table, creating terraces by building bunds to retain water so distributing it more evenly, and altering microtopography to provide a variety of habitats for colonization. It may also be necessary to provide shade to lower the temperature and increase relative humidity near the peat surface. These processes impact CO 2 and CH 4 emissions. Waddington and day (2007) compared emissions from both an extracted (cutover) peat land and a restored peat land. Before habitat improvement, the seasonal fluxes of CH 4 from the two sites were not significantly different from each other. However, three years post-restoration, the seasonal CH 4 emissions at the restored site were 4.2 g m -2 CH 4 season -1, 4.6 times greater than at the cutover site. Tuittila et al. (2004) showed that under optimal conditions, reintroduction of Sphagnum fragments turned the cutaway site from being a C source into a sink of 23 g C m -2 per season (mid-May to the end of September). Rewetting lowered total respiration and increased gross photosynthesis, which resulted in a higher incorporation of CO 2 into the system (Tuittila et al., 1999).
6.2. Fate and treatment of excavated peat
Excavated peat is treated in a range of different ways. The losses of C from this source depend upon the treatment of the peat following excavation. Peat may be replaced in the excavated site, laid on both sides of a construction area ( e.g. a road) on top of the existing peat. It may be used in the restoration of peat land at the site (for example within felled areas of former conifer plantation), used in the landscaping of roadsides and borrow pits, dried and used as fuel, dried and used for horticultural or agricultural purposes, stored with lime or stored in a bund. Peat that is used for restoration of a site under anaerobic conditions could potentially retain a large proportion of its C content, but peat that is laid on the surface will loose a large proportion of the stored C. Peat that is dried and burnt as a fuel can be assumed to emit all of its stored C. The fate of the excavated peat should be accounted for when estimating the percentage C loss following excavation.
6.3. Drainage features associated with wind farm construction
6.3.1. Drainage of wind turbine foundations
The excavated foundations and hard-standing are kept free of water during the construction period to allow a reinforced concrete base to be put down. The usual sources of water ingress are ground, surface and rain water. Where local topographic conditions allow, the excavation for foundations is usually designed to be gravity draining ( AMEC, 2004). If this is not possible, the excavation is drained using a pump. Pumping water away from the foundations can have a major impact on the hydrology of the site. Expelled water can form additional drainage channels, which may result in wider drainage of the site than anticipated. These effects can only be described using sophisticated hydrological modelling and detailed survey information of the site. Therefore, as noted in the introduction to this section, such data cannot be assumed to be available, and thus these effects will not be included in the present study. The extent of drainage around the foundations, and the impact of the lowered water table on C losses should be estimated as for excavated roads.
6.3.2. Road drainage
Drainage of excavated roads
Excavated roads can include drainage ditches, cross drains and check dams.
Drainage ditches - Drainage ditches are open channels used to catch, direct and disperse water flow. For excavated roads on level ground, drainage ditches are required on both sides of the road. If there is a slope perpendicular to the road, drainage ditches are usually constructed on the uphill side of the road ( AMEC, 2005; Lewis Wind Power, 2004).
Cross drains - Cross drains are open top drainage channels, typically lined with stone, used to direct water from one side of the road to the other. Regular cross drains are constructed under the excavated road to maintain the natural flow of water (Lewis Wind Power, 2004).
Check dams - Check dams are usually required at regular interval in the drainage ditches constructed along an excavated road. These act as a silt/pollution trap, slowing the flow of water and allowing the sediment to settle out. They also help to direct the water into cross drains. The spacing of check dams is dependent on the gradient of the road, the spacing of cross drains and the depth of road excavation (Lewis Wind Power, 2004).
Drainage of floating and rock filled roads
The road surface of both floating and rock-filled roads protrudes above the surrounding peat land and may disrupt the surface water flow. Therefore, cross drains may be required to reduce disruption of surface water flow (Lewis Wind Power, 2004). The impact of cross drains is to reduce disruption of the site, so does not need to be included in the estimates of C losses. If topical drainage is needed to maintain the integrity of the floating or rock-filled road, the impacts on C emissions can be determined as for drains around excavated roads.
6.4. Summary of factors to be simulated
Table 6.4.1. Type of experimental measurements collated from literature
Provides evidence for changes in C emissions associated with…
- C losses from restored borrow pits giving depth of water table, plant inputs, location (climate) and soil type
- Restoration of borrow pits
- C losses from drained peat lands giving depth of water table before and after drainage, plant inputs, location (climate) and soil type
- Construction of temporary cut-off ditches during operation of borrow pits
- Drainage of excavated roads
- Any drainage associated with cable trenches
- C losses from restored habitats giving depth of water table, plant inputs, location (climate) and soil type
- Habitat restoration using excavated peats
- C losses from compacted peats
- Construction of rock-filled roads
Table 6.4.2. Simulations required
Provides evidence for changes in C emissions associated with…
- C losses at different depths of water table, climates and soil types
- Restoration and drainage of borrow pits
- Drainage of excavated roads
- Topical drainage of floating roads
- Topical drainage of rock filled roads
- Habitat restoration using excavated peats from roads and foundations.
- Any drainage associated with cable trenches
6.5. Effect of drainage on peats
Peat lands accumulate C because mean annual primary production exceeds annual organic matter decomposition (Clymo, 1983). Carbon accumulation is mainly due to the slow decomposition rate that is sustained by high water levels and anoxic conditions (Clymo 1983).
As described above, construction and operation of a wind farm usually requires drainage of water. Drainage can result in a reduction in the level of the water table, loss of habitat structure and subsidence of the peat surface (Lukkala 1949). The increase in soil respiration following a reduction in the water level following artificial drainage has often been regarded as an indicator of C loss from the accumulated peat ( e.g. Silvola 1986; Nykiinen et al. 1997). Hogg et al. (1992) suggested that C losses from peat following a reduction in the level of the water table depend on the quality of the peat. Peats that have previously been exposed to long periods of aerobic decomposition may be resistant to further decomposition (Bridgham and Richardson, 1992). A peat land with a water table 20 cm or more below the peat surface for most of the summer might not exhibit a significant increase in C release when areas of the peat are drained during wind farm construction. Many forested bogs and fens would have such a water regime. By contrast, in a peat land where organic strata near the surface have been continuously inundated, the peat containing highly labile organic matter is likely to decompose at increased rates when the surface is drained.
6.5.1. Impact on peat hydrology
The efficacy of drainage is related to the depth of ditching, distance between ditches, and the hydraulic conductivity of the peat (Boelter, 1972; Armstrong, 2000). A drainage ditch cut into peat provides an outlet for both surface and subsurface flow, but Boelter (1972) showed that the effect of open drains on ground water levels in peat was limited in horizontal axis. The reduction in the water level is greatest close to the ditch, and diminishes rapidly with distance, depending on hydraulic conductivity. For example, Prevost et al. (1997) found that drainage was most effective within 15 m of ditches in a drained tree-covered bog near Riviere-du-Loup, Québec, whereas Boelter (1972) found ditches were effective at up to 50 m in fibric peat in Minnesota, but ineffective beyond 5 m in more decomposed peat where the hydraulic conductivity was lower. A review of the available literature (Table 6.5.1) shows the extent of drainage effects are reported as being anything from 2 m (Burke,1961) to 50 m (Boelter, 1972) around the site of disturbance. Burke (1961) investigated the effects of drains on blanket peat in Glenamoy, western Ireland and concluded that drains installed in blanket peat have a localised effect: the horizontally lowered water level was found to a distance of only 2m. Stewart and Lance (1991) observed an even smaller extent of drainage at Moorhouse, with a lowering of the water table laterally observed 2m downslope (2.3m on steeper slope) and only 1m upslope (0.3m steeper ground). The flow of ground water through the saturated zone is governed by the hydraulic gradient and hydraulic conductivity of the soil (Gillman, 1994). The peat at Glenamoy was more Sphagnum-rich than Moorhouse and a limited number of measurements showed that hydraulic conductivities were generally an order of magnitude higher at the less decomposed Irish site. In contrast, Gilman (1994) showed that a strip of about 30m dried out more rapidly (due to transpiration) adjacent to ditch in early summer. Godwin and Bharucha (1932) studied the effect of drainage on Wicken fen and concluded that the effects of ground water flow to dykes at Wicken fen were minimal beyond 50 m. Coulson et al (1990) observed that drainage was efficient only at the lower altitude sites ( i.e. at Waskerley, laterally up to 15m). At high altitude, where the rainfall is greater, the ditches had little influence on the position of the water-table ( i.e. at Moorhouse, laterally up to 1.5 m).
Table 6.5.1. Reported extents of drainage effects for peats with different hydraulic conductivities.
Extent of drainage around site of disturbance (m)
Saturated hydraulic conductivity
(mm d -1)
Coulson et al (1990)
Stewart and Lance (1991)
Stewart and Lance (1991)
Prevost et al (1997)
Godwin and Bharucha (1932)
The extent of drainage around the site of construction strongly influences the total volume of peat impacted by the construction of the wind farm. Clearly, if only 1m of peat is drained, a much smaller volume will be affected than if 200m of peat is drained. We recommend that where sufficient measurements are available to describe the hydrological features of the area of the wind farm, these should be used together with a detailed hydrological model to simulate the likely changes in peat hydrology. However, in the absence of such detailed measurements, on a level site with uniform soil distribution, an estimate of the extent of drainage could be obtained by fitting a regression equation to the above data (Figure 6.5.1), giving the following equation for extent of drainage around the site of disturbance:
where E is the extent of drainage around the site of disturbance (m), and H is the hydraulic conductivity (mm d -1).
The results of a paired two sample t-test shows that there is no significant difference between the means of the measured values and the values estimated using the equation (P=0.9959).
Figure 6.5.1. Extent of drainage around the site of disturbance with respect to the hydraulic conductivity.
In practice, sites are rarely sufficiently level or uniform to use this equation. An estimate of the extent of drainage should be obtained by measurements at each site. Further research into potential methods for estimating the extent of hydrological conductivity from only simple measurements and generic equations is urgently needed.
6.5.2. Impact on carbon stocks
With the lowering of the water table due to drainage, a number of different effects can be observed. Increased respiration and release of CO 2 has been observed on draining peats in both laboratory and field experiment (Moore and Dalva 1993; Funk et al., 1994; Freeman et al., 1993; Martikainen et al.,1995). However, an increase in methane (CH 4) oxidation results in a reduction in CH 4 emission from drained peat (Martikainen et al., 1995; Glenn et al., 1993; Funk et al., 1994). Increased CO 2 emissions, a concomitant decrease in CH 4 emissions and increases in N 2O emissions have been reported (Aerts and Ludwig, 1997; Moore and Dalva 1993; Funk et al., 1994; Freeman et al., 1993; Martikainen et al., 1995). Changes in the mean CO 2, CH 4 and N 2O emissions as a result of a reduction in the level of the water table in different laboratory and field experiment are summarised below (Table 6.5.2.a, Table 6.5.2.b). In these studies, water table levels were reduced either in the laboratory (15-40 cm lower than the control (Moore and Dalva, 1993; Freeman et al., 1993; Funk et al., 1994) or emissions from drained and undrained peat soils were compared in the field ( e.g. Martikainen et al., 1995; Glenn et al., 1993; Roulet et al., 1993).
In a laboratory study Moore and Dalva (1993) showed that lowering the water table to a depth of 40 cm increased CO 2 fluxes by an average of 4.3 times and decreased CH 4 emission by 5.0 times as compared to emissions from saturated columns. Funk et al (1995) reported a three times increase in CO 2 emission as compared to control cores. Freeman et al. (1993) observed a two times increase in CO 2 emission and five times decrease in CH 4 emission in a laboratory study. In studying the effects of lowering the water table in a eutrophic fen, Aerts and Ludwig (1997) did not find any effect of lowering the water table on CO 2 emission, but the lowered water table significantly increased N 2O emissions. Note that in most natural peat lands, the availability of nitrogen will limit emissions of N 2O.
In field studies, the increase in CO 2 emissions is generally only observed with a lowering of the water table to a given depth (between 10 and 30-40 cm depending on the study), with no further increase in emissions observed with a further lowering of the water table (Silvola et al., 1996a; Chimner and Cooper, 2003). Chimner and Cooper (2003) suggested that this is due to a lack of easily oxidized labile C in the deeper soil layers. This is supported by the laboratory experiments of Hogg et al. (1992), in which in vitro drained samples from the 0-10 cm peat layer were observed to release around 10 times more CO 2 than samples from the 30-40 cm layer. This they attributed to the relatively large pool of non-structural carbohydrates in surface samples, deriving from recently dead plant biomass. In a field experiment in Finland, Martikainen et al. (1995) showed that drainage increased annual CO 2 and N 2O emissions and decreased CH 4 emissions. Silvola et al., 1996 reported that a lowering of the water table by 1 cm increased CO 2 fluxes by an average of 7.1 mg CO 2 m -2 h-1 at 12 0 C and 9.5 g CO 2-C m -2 year -1.
Table 6.5.2.a. Evidence from Laboratory Experiments
Soil organic Carbon (%)
(m d -1)
Water table (cm)
(g CO 2 m -2 d -1)
(mg CH 4 m -2 d -1)
N 2O emission
(µg N 2O m -2 d -1)
1 - 4
Funk et al., 1994
6 - 11
Moore and Dalva 1993
Freeman et al., 1993
Kechaverzi et al., 2007
0.3 - 2.1
0.01 - 2.2
0.07 - 1.5
0.1 - 1.7
0.2 - 2.5
52.8 - 190
Aerts and Ludwig 1997
Table 6.5.2.b. Evidence from Field Experiments
Soil organic Carbon (%)
(m d -1)
Water table (cm)
(g CO 2 m -2 d -1)
(mg CH 4 m -2 d -1)
N 2O emission
(µg N 2O m -2 d -1)
-3 - -5
Martikainen et al., 1995
- 23 - -33
-9 - -22
-22 - -30
+10 - +6
Chimner and Cooper 2003
+5 - +1
0 - -5
-6 - -10
-11 - -40
Bog (Low sedge)
Bog (Low sedge drained)
Silvola et al., 1996
Pine bog drained
Cotton grass Pine bog
Cotton grass pine bog drained
Cotton grass Pine bog
Cotton grass pine bog drained
Tall sedge fen
Tall sedge fen drained
Herb-rich flark fen
Herb-rich flark fen drained
6.6. Calculation of the effect of drainage and flooding on peat lands using IPCC default values
The effects of drainage and flooding of peat lands on CH 4 emissions are estimated in the Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories ( IPCC, 1997) using a very general approach, which includes only limited site specific information. However, these estimates are the standard used in national inventories, and so are included in the technical guidance for completeness.
6.6.1. Methane Emissions
Flooding of peat lands results in emissions of CH 4 due to anaerobic decomposition of vegetation and soil C. Methane emissions are highly variable and are dependent on the ecosystem type, the status of the ecosystem that is flooded ( i.e. above- and below-ground C, plant types, whether any pre-flooding clearing occurred, etc.) and on the depth and length of flooding. Rates of CH 4 emissions from freshwater wetlands are also strongly dependent on temperature, and therefore vary seasonally, as well as daily. The simple CH 4 flux calculation included in the IPCC Guidelines for National Greenhouse Gas Inventories ( IPCC, 1997, Vol3, pp.5-55 to 5-56) is based on the area of land flooded, an average daily CH 4 emission rate, and the number of days in the year that the land is flooded. Land is divided into groups based on characteristics such as length of flooding, vegetation type, and latitude ( IPCC, 1997, Vol3, Table 5-13). The CH 4 emission rate provided for acid bogs is 11 (1-38) mg CH 4-C m -2 day -1, based on the experiments of Aselmann and Crutzen (1989). The period of flooding is taken to be 178 days yr -1 based on the monthly mean temperature and the lengths of inundation. For lowland fens, the CH 4 emission rate is 60 (21-162) mg CH 4-C m -2 day -1 with a period of flooding of 169 days yr -1. The following equation is used to estimate CH 4 emissions in flooded land in acid bogs using IPCC default values:
where ECH4 is the total annual emissions of CH 4 (t CO 2 equiv. ha -1 yr -1), RCH4 is the annual rate of CH 4 emission ( RCH4 = 11 mg CH 4-C m -2 day -1 / 10 9 mg t -1 x 10 4 m ha -1 x 365 day yr -1 = 4.015 x 10 -2 t CH 4-C ha -1 yr -1), DF is the number of days in the year that the land is flooded each year ( DF = 178 days yr -1), and CCH4-C? CO2 converts CH 4-C to CO 2 equivalents ( CCH4-C? CO2 = (23 x 16/12) = 30.67 CO 2 equiv. (CH 4-C) -1). For fens, RCH4 = 0.219 t CH 4-C ha -1 yr -1 and DF = 169 days yr -1.
To calculate the reduction of CH 4 emissions due to wetland drainage, the area drained is multiplied by the difference in the average daily CH 4 emission rate before and after draining, and is multiplied by the number of days in a year that the wetland was emitting CH 4 prior to drainage. The number of days of CH 4 emissions prior to drainage can be approximated by using the number of days in the year that the wetland was flooded as shown above.
6.6.2. Carbon Dioxide Emissions
The rate of carbon dioxide (CO 2) emissions from drained organic soils depends on climate, the composition of the organic matter and the degree of drainage. Unlike mineral soils, where C levels approach a new equilibrium level following changes in land management, C losses from organic soils can be sustained over long periods of time, in principle, until the organic soil layer has been completely lost. As decomposition proceeds, the more recalcitrant fractions of organic matter, having slower decomposition rates, will accumulate. This would tend to reduce CO 2 emissions over time. The IPCC Revised 1996 Guidelines for National Greenhouse Gas Inventories ( IPCC, 1997) provides values for CO 2 emissions on drainage of organic soils for upland crops ( e.g., grain, vegetables) of 2.2 t C ha -1 yr -1 in boreal regions (Finland, Russia), 7.9 to 11.3 t C ha -1 yr -1 in temperate climates ( USA and Western, Central and Eastern Europe, China, Japan) and 21.9 t C ha -1 yr -1 in Florida and coastal California (Armentano and Menges, 1986). Following the figures used in the IPCC guidelines, the CO 2 emissions are assumed to range from 7.9-11.3 t C ha -1 yr -1, and so are taken to be the mid-point of the range x 3.66 (t0 convert to CO 2) = 35.2 t CO 2 ha -1 yr -1. Therefore, CO 2 emissions from soils can be estimated from the IPCC default values using the following equation:
where ECO2 is the total annual emissions of CO 2 (t CO 2 ha -1 yr -1), RCO2 is the annual rate of CO 2 emission ( RCO2 = 35.2 t CO 2 ha -1 yr -1), DF is the number of days in the year that the land is flooded each year. Note that the same values for RCO2 are given in the IPCC guidelines for acid bogs and fens. When the soil is undrained, the period of flooding is taken to be 178 days yr -1 for acid bogs and 169 days yr -1 for fens based on the monthly mean temperature and the lengths of inundation. When the soil is drained, the period of flooding is assumed to be zero ( DF = 0 days yr -1).
6.6.3. Nitrous Oxide Emissions
Nitrous oxide emissions are dependent on the amount of available nitrogen in the soil, and on applications of nitrogen as fertiliser and organic manures ( IPCC, 1996). Since the levels of available nitrogen are very small, and fertilisers and manures are not usually applied, nitrous oxide emissions are assumed to be negligible.
6.7. Calculation of the effect of drainage and flooding on peat lands using site specific equations
6.7.1. Derivation of site specific equations
The experimental data used to derive site specific equations were collated during the development of the ECOSSE model (Smith et al, 2007). Experimental data were collated in order to verify the performance of the model at these sites. Having verified the accuracy of the model, multiple simulations could be run to generate values that would be used to obtain a simple summary equation of the results. In this way, the noise in the experimental data could be removed, allowing greater confidence in the site specific equations. However, in practice, there was no need to use the ECOSSE model to remove noise in the experimental data, because for the subset of peat lands on acid bogs and fens, a good fit could be obtained directly from the experimental data. If the equations were to be extended (for example to mineral soils), it is likely that the noise in the data would obscure the development of a good regression equation, and the original approach (using ECOSSE) would again be needed.
Linear multiple regression was done using Minitab 15 to derive equations for both CO 2 and CH 4 emissions. The regressions used peat depth, air temperature, pH, water table, rainfall, soil C content and bulk density as predictors. Only the parameters that contributed significantly to the variance were included in the final equation.
The parameters controlling CO 2 emissions are air temperature, water table depth and peat depth. The site specific equation for CO 2 emissions is
where RCO2 is the annual rate of CO 2 emissions (t CO 2 (ha) -1 yr -1), T is the average annual air temperature (° C), dpeat is the peat depth (m) and dwater is the water table depth below the surface (m). This equation was derived from 41 experiments (see Table 6.5.2), and has an associated R 2 value of 53.8%, P <0.0001. By statistical convention, if P<0.01 this relationship can be considered to be highly significant.
The parameters controlling the CH 4 emission are air temperature, pH of the top 20 cm of peat and water table depth. The equation for CH 4 emission is:
where RCH4 is the annual rate of CH 4 emissions (t CH 4-C (ha) -1 yr -1), T is the average air temperature (° C), pH is the soil pH and dwater is the water table depth (m). This equation was derived from 40 experiments (see Table 6.5.2), and has an associated R 2 value of 52.7%, P <0.0001. By statistical convention, if P<0.01 this relationship can be considered to be highly significant.
6.7.2. Global sensitivity analysis
The sensitivity of the derived equation to changes in the input variables was analysed using MATLAB Script. A global sensitivity analysis adjusts more than one input variable at a time. Multiple sensitivity analysis takes the form of a factorial analysis, which runs the model for all combinations of inputs. The probability density functions for the input variables were drawn from the pre-defined parameter space and used to generate sample combinations of input variables. The range of input variables used in the sensitivity analysis is given below:
CO 2 analysis
Temperature (T) = 0-15 ° C
Peat depth (d peat) = 0-10 m
Water table depth (d water) = 0-10 m
CH 4 analysis
Temperature (T) = 0-15 ° C
Water table depth (d water) = 0-10 m
pH = 3-10
The contribution index was used to express the sensitivity of the model to the inputs (Gottschalk et al., 2007). This index represents the importance of each variable, also taking interactions between the variables into account. The contribution index is calculated by running a Monte Carlo simulation, consisting of multiple runs of the model, with all input factors sampled from the probability density functions defined for each input. The Monte Carlo simulation is then repeated for each of inputs included, holding selected input constant at its default value, whilst allowing all others to vary within the pre-defined ranges. The distribution of the difference in the model outputs (CO 2 and CH 4 emissions) from the first Monte Carlo simulation represents the global sensitivity. This gives a quantitative estimate of the contribution of each input variable to the global sensitivity. The contribution is expressed as the normalised percentage change with respect to the global sensitivity.
Carbon dioxide emissions ranged from 4 to 193 t CO 2 ha -1yr -1, with water table depth providing the highest contribution to the global sensitivity as shown below.
This range of emissions is within the normal range observed for peat lands, and the strong positive response to water table depth is as expected, reflecting the increased CO 2 emissions that have been observed in drained soils.
Figure 6.7.1. Contribution of selected variables to global sensitivity of CO 2 emissions
Methane emissions ranged from -14.6 to 0.9 t CH 4 ha -1 yr -1, again with water table depth providing the highest contribution to the global sensitivity. The analysis for CH 4 are shown below. This range of emissions is within the normal range observed for peat lands, and the strong negative response to water table depth is as expected, reflecting the increased CH 4 emissions that have been observed in flooded soils.
Figure 6.7.2. Contribution of selected variables to global sensitivity of CH 4 emissions