Millennia peatland model
Peatlands in the UK have mostly been 'growing' for the last 6-8k years, since after the ice-age. Basically, plants take up carbon from the atmosphere through photosynthesis and put this carbon into plant biomass (leaves and roots). However, plants or parts of their biomass die annually, decomposing according to their litter quality and the environmental conditions (temperature and moisture). Therefore annual litter cohorts are laid down, mostly within the layer above the mean water table (acrotelm), and as plant matter decays only very slowly under wet and cold conditions in the uplands, the undecomposed organic matter is 'locked away' over time as the carbon enters the saturated zone (catotelm) - as long as conditions do not change (i.e. high water tables and low temperatures). Consequently, over millennia organic matter accumulates, leading to peat formation. The MILLENNIA model (Heinemeyer et al., 2010) captures this long-term peat accumulation in relation to variable (i.e. Holocene) climate data, estimates of mean annual temperature (MAT) and mean annual precipitation (MAP), and resulting changes water table and corresponding vegetation dynamics over time. The model can switch from long-term build up during the Holocene to recent predictions using available climate data from actual locations. The model has been validated and successfully compared to other UK peatland models in Clark et al. (2010) who pointed out several advantages of this model over others (mainly the dynamic (and total) peat depth, dynamic water table and vegetation feedbacks). We have since further developed the monthly model in collaboration with M. Carroll on his UKPopNet cranefly work (i.e. Carroll et al., 2015). This model performs well in comparison to site hydrological data (see previous figure), a prerequisite for modelling accurate carbon dynamics. However, the winter months still show some discrepancies to site data as there is no modelled snow representation. However, annual mean temperatures also seem crucial for key decomposers in peatlands, enchytraeids (see Briones et al., 2007), which will be subjected to climate change alongside other biota crucial for bird populations, such as crane flies (Carroll et al., 2011). We will therefore run the MILLENNIA model with climate change scenarios to determine likely impacts on such key species.
Importantly, the field flux measurements and the empirical models of carbon uptake (photosynthesis) and release (respiration) in connection with root exclusion will allow comparing field based estimates of net primary productivity (NPP) to predictions by the MILLENNIA model based on the real site climate data. Currently, the MILLENNIA model derives net carbon inputs (NPP) based on a simple empirical relationship to actual evapotranspiration (AET). This will offer an important validation for the MILLENNIA model and enable comparison of the simple versus a more process level driven calculation of NPP (considering light conditions and vegetation specific NPP estimates). A further opportunity will be the model inclusion of available vegetation and treatment specific water flux data. This will enable an estimation of the treatment impact on the water balance and discharge (flow patterns) in connection with runoff relationships to water table dynamics and vegetation. We will, over time, attempt to achieve this goal in a combination of GIS and the MILLENNIA model, also as part of the up-scaling attempts to larger regions (making use of the variance estimates from the ‘between site’ comparison to provide an uncertainty measure on model input parameters and processes).
Above: the MILLENNIA model, considering past climate (during the Holocene - in the UK in general about 6,000-8,000 years before present), dynamic vegetation (litter quality) and water table as affected by climate and rooting depth (evapotranspiration), topography and drainage .
Below: preliminary modeled water table depth (WTD) data compared to site measurements from the three sites (note the good overall agreement, but lower WTD at the Nidderdale and Whitendale sites - likely due to site specific drainage - STDEV ~4 cm) v.1 (left) unadjusted model runs; v.2 (right) adjusted model runs, increasing bedrock drainage (modeled via hydraulic conductivity and specific yield in the bedrock) at Nidderdale and Whitendale.
Below Figures: We did a comparison of modeled versus paleao-ecological water table depth (WTD) reconstructions (Heinemeyer & Swindles, 2018) for Moor House (NNR), a well-studied upland blanket bog site in the UK, which revealed a very good agreement between the two methods (Figure 4 & 5); not only did the WTD predictions agree over an about 100 year time period, they also agreed with the available actual water table measurements over the past ~20 years (ECN monitoring; data citation code: ‘ECN:AH2/12 & ECN: AH2/14’), and the model allowed comparing water table depth for an unmanaged (wet) and a grouse moor managed scenario (dry) including maintenance of drainage ditches and burn rotations (Figure 6). Most interesting was the period between 1840s and 1940s, an unexpectedly drier period based on TA predicted hydrological conditions compared to the wetter conditions predicted by the MILLENNIA model based on climate only (Figure 6). Moor House was a formal shooting estate throughout exactly this period (i.e. 1842-1951), based on grouse bags and predator control information (ECN data provided by Rob Rose from the Centre for Ecology and Hydrology (CEH) Lancaster, personal communication). Therefore, burning in connection with drainage seems to be the most likely factor explaining the water-table lowering:
Most importantly, the model allowed us to explore the grouse moor management impacts of drainage and burning (at Moor House since about the 1850s). We included drainage impacts with a cycle of natural infilling and maintenance and burning reducing net primary productivity (NPP) as part of a burn management cycle (i.e. natural regrowth over time). Please see the actual paper for more information (Heinemeyer & Swindles, 2018). Importantly, the model enables making predictions of changes in carbon accumulation, C fluxes and GHG emissions (CO2 & CH4 emissions) for both scenarios (i.e. unmanaged 'natural' vs. managed 'grouse moor'):
Expand section to read all:
The MILLENNIA model scenario period of 1851-1950 (Figure 8; right) showed a positive (C gain) annual soil C budget (b) of 12.7 ± 22.7 g C m-2 for the unmanaged scenario (with a mean water table depth (WTD) of 4.7 ± 1.5 cm; a) compared to a negative (C loss) budget of -38.6 ± 110.1 g C m-2 for the grouse managed scenario with drainage (mean WTD of 7.7 ± 2.1 cm) assuming a 10-year NPP recovery time after burning (resulting in 51.3 g C m-2 yr-1 less soil carbon gain under grouse management). The corresponding peat depth increments predicted by the MILLENNIA model during 1851-1950 showed a mean peat accumulation of 0.03 ± 0.59 cm for the unmanaged (no shoot) versus a loss of -0.06 ± 0.61 cm for the grouse managed scenario (i.e. 0.09 cm yr-1 less peat depth increment). Moreover, the 5-year NPP re-growth scenario (data not shown) resulted in about 45% lower mean annual soil C budget loss (-21.6 ± 93.7 g C m-2) and half the peat increment loss (-0.03 ± 0.63 cm) compared to the 10-year NPP recovery scenario. In comparison to the default burn and drain scenario, the burn but no drain 10-year NPP re-growth scenario (data not shown) resulted in about 30% lower C and peat accumulation losses with a mean annual soil C budget of -28.4 ± 108.7 g C m-2 and -0.04 ± 0.63 cm mean annual peat depth increment (with a mean WTD of 4.8 ± 1.9 cm). The burn but no drain 5-year NPP re-growth scenario reduced this further (i.e. 80% lower soil C and peat increment losses with a mean annual soil C budget of -10.9 ± 92.6 g C m-2 and a mean annual peat depth increment of -0.01 ± 0.64 cm.
During 1831-1850, the period of drainage only (i.e. no burning), the managed scenario reduced the mean annual soil C budget by 27.8 g C m-2 to -7.2 ± 15.8 g C m-2, which reflected an average reduction in mean annual WTD by 3.9 cm to 8.3 cm (a). These changes in soil C budget under drainage only corresponded to an annual peat depth increment reduction by 0.06 cm to -0.02 ± 0.63 cm compared to the unmanaged scenario (b). The grouse moor management (i.e. 1851-1950) not only impacted C dynamics via reduced water-tables from drainage (a), it also altered C inputs and thus C dynamics via reduced NPP following burning. Overall, drained and 10-year NPP recovery scenarios reduced both mean annual C losses from soil CO2 fluxes (308 ± 106 g C) and annual CH4 emissions (4.9 ± 8.3 g C) compared to the unmanaged scenario for which mean annual values were 417 ± 60 g C for CO2 and 13.2 ± 20.4 g C for CH4 net emissions (i.e. including methane oxidation, ebullition and plant-mediated transfer (PMT) processes via sedge leaves and stems). However, whereas the no drain 10-year NPP burn scenario decreased CO2 (292 ± 110 g C) and increased CH4 (9.8 ± 18.0 g C) emissions (a), the 5-year NPP scenario (data not shown) increased both CO2 fluxes (348 ± 92 g C) and net CH4 emissions (11.2 ± 18.8 g C) emissions, reflecting quicker vegetation re-growth and thus NPP and PMT recovery.
Up-scaled to an intense grouse moor management period of 100 years these C losses (i.e. 1850-1950) relate to around 5 kg C m-2 less soil C or 10 cm lower peat depth accumulation than predicted for the unmanaged scenario. Notably, burning and drainage contributed equally to the carbon loss, via reduced NPP (i.e. less litter input) and enhancing decomposition (i.e. lower WTD). However, the MILLENNIA model did not account for any potential charcoal impacts on peat properties, hydrology and decomposition. In particular, impacts on peat bulk density with possibly changes in water holding capacity and charcoal inputs representing an inert C pool with possible additional effects on decomposition rates might need to be considered in future model developments.