Monitoring soil carbon and nation wide carbon inventories Raisa Mäkipää Finnish Forest Research Institute (Metla) with Mikko Peltoniemi, Margareeta Häkkinen, Petteri Muukkonen and Aleksi Lehtonen at Metla, and colaborators Jari Liski and Kristiina Karhu at SYKE Taru Palosuo and Marcus Lindner at EFI
Outline Introduction Forest monitoring in Finland Soil C modeling & Nation wide soil C inventories Model evaluation and validation Repeated soil sampling Within site spatial variation of soil C & sampling desing at plot level Model based stratification & soil survey
Introduction - Forest monitoring in Finland National Forest Inventory (NFI) since 1922 First nation-wide vegetation survey 1951-53 as a part of NFI Forest monitoring permanent sample plots (n=3000) established by NFI in 1985-86, remeasured in 1995 (trees and understorey veg.) Soil sampling of the permanent plots in 1985-1989 (upland soils, subsample) and in 2001-2002 (peatlands, all 978 plots) a sub-sample of plots re-measured in 2005-2006
Results of vegetation survey 1951-1995 Abundance of blueberry (Vaccinium myrtillus)
Significant changes in bryophyte abundances Hylocomium splendens in 1950s, 1980s and 1990s Significant changes: dark/grey shading Source: Mäkipää & Heikkinen 2003 JVS 14: 467-508 year 1995 1985 1953
Forest monitoring in Finland National Forest Inventory (NFI) since 1922 First nation-wide vegetation survey 1951-53 as a part of NFI Forest monitoring permanent sample plots ( n=3000) established by NFI in 1985-86, remeasured in 1995 (trees and understorey veg.) Soil sampling of the permanent plots in 1985-1989 (upland soils, subsample) and in 2001-2002 (peatlands, all 978 plots) ICP Level 1/Forest Focus monitoring of forest crown condition
Soil C modeling & Nation wide soil C inventories No repeated soil survey available for C inventories Soil carbon modelling can be used in national GHG inventory (IPCC 2003, 2006) We integrated a dynamic soil C model (Yasso) to NFI data on forest resources forest inventory data (aggregated) growing stock, area (forest land, no peat), growth indices, harvests, natural mortality biomass models / BEF(t) biomass turnover i.e. modelled litter input to soil soil decomposition model YASSO Output: Forest carbon balance in Finland 1922-2004
Carbon balance of forests in Finland 1922-2004 C sink or source (Tg / year) 15 10 5 0-5 -10 Biomass Dead wood Soil organic matter + litter 1922 1932 1942 1952 1962 1972 1982 1992 2002 Year Liski et al. 2006. Annals of For. Sci. NOTE peatland soils excluded
Validity of the applied modelling approach tested: simulated vs. chronosequency 12 11 k s = 0.0058 ± 0.0010 kga -1 m -2 10 9 Carbon (kgm -2 ) 8 7 6 5 4 3 2 1 0 Mesic site, Norway spruce Mesic site, Scots pine Sub-xeric site, Scots pine Measured organic layer k m = 0.0047 ± 0.0014 kga -1 m -2 (p = 0.0011) 0 10 20 30 40 50 60 70 80 90 100 110 120 Simulated rate of change in soil C agreed with measured change in humus layer of a chronosequency Stand age (a) Source: Peltoniemi et al. 2004. GCB 10:2078-2091
Model evaluation- Uncertainty of model based soil carbon sink estimates in Finland Upland soils Carbon sink (Tg) -5 0 5 10 1990 1992 1994 1996 1998 2000 2002 Year 2.5% C.l. 25% 50% 75% 97.5% Source: Monni et al. 2006. Climatic Change, in press.
Model evaluation - differences between potential soil models (Yasso, ROMUL) Site: mesic Scots pine stand in middle boreal zone Source: Palosuo et al. manuscript in preparation
Repeated soil sampling Question: Does repeated soil C measurements confirm predicted increase in soil C stock Soil C models predict increase in soil C stock for forest stands over 20-years Rate of change is known to be largest in organic layer
Repeated soil sampling - Material 38 stands (24 pine, 14 spruce), now 40-80 years measured 1985-89 (one composite sample per plot) New meassurements with good spatial information (n=40 per plot) in 2005 -> kriging based estimates of mean and variance of soil C Soil samples from organic layer CHN-analysis with LECO Fig. Example of sample plot; location of old and new sample points
Measured change in C stock of organic layer Increase of C stock in 36 of 38 plots (significant in 7 plots) (Häkkinen et al.)
Results Significant increase in C stock measured Average of C stock of all the old measurements: 1444 g C m -2 ) C stock, new measurements: 1849 g C m -2 Change of 404 g C m -2 & Annual change 23 g C m -2 was significant With this rate of change and variation the change could be detected by sampling >27 plots (in this forest class of intermediate age) In younger stands between site variation is most likely higher and in older stands rate of change is lower -> nro of sampled plots should be higher
Sampling desing at plot-level Questions What should be spatial location of sample points (to avoid correlated samples)? How many samples per plot/stand are needed to obtain reliable plot level estimates of soil C stock?
Material Spatial auto-correlation of carbon stock in humus layer (within site variation) studied 5 young Scots pine stands (measured 2004) 4 young Norway spruce stands (measured 2005) 1 old pine stand (Liski 1995) 1 old spruce stands appr. 100 soil samples / plot Fig. Example of sampling design within one plot Y Coord -1000-500 0 500 1000-1000 -500 0 500 1000 X Coord
Distance between sampling points To avoid correlated samples distance between sampling points should be > 7 m γ 0.5 0.4 0.3 Y Coord -1000-500 0 500 1000 2.5 kg m 2 2 1.5 1 0.2-1000 -500 0 500 1000 X Coord 0.1 0.0 0 250 500 750 1000 r (cm) Range 637.06 cm Nugget 0.183 Sill 0.319 Fig. Spatial autocorrelation in one sample plot Source: Muukkonen, Häkkinen Mäkipää, manuscript
Effect of sample size (n) on variance 100 75 Ratio 50 25 0 0 20 40 60 80 100 120 n Ratio = (variance / n) / mean
Model based stratification & soil survey Question Models can provide best correlates with expected changes of soil carbon: 1.Can they be used to stratify sampling? (to select an optimal subsample of plots for measuring changes in soil C stock) 2.How much sampling efficiency would be improved?
Model based stratification - Method Permanent sample plots (1721 plots on mineral soil): dom. tree species, site fertility, age of the stand, temperature, precipitation, typical stand development scenario with MOTTI simulator Stratification Optimal sampling (Neyman) Proportional (to size of strata) Simple random
Simulated changes, Motti stand simulator & Yasso Example: recently cut spruce site on poor soil in central Finland Source: Peltoniemi et al.
Strata Probability Density Function Cumulative Distribution Function 4 f( C 10a ) 0.0 0.1 0.2 0.3 F( C 10a ) 0 1 2 3 4 3 2 1 2 0 1 2 3 2 0 1 2 3 C 10a (kgm 1 ) C 10a (kgm 1 ) Optimal 4 strata according to modelled rate of change Source: Peltoniemi et al.
Gain from stratification Source: Peltoniemi et al.
Discussion on model based stratification 1. Can they be used to stratify sampling? Yes, as any method that can provide a correlate to soil C change Approach should be applicable for any other target parameter as well 2. How much sampling efficiency would be improved? It depends on how precise are the soil C measurements -> Select paired repeated samples; or take enough samples and use spatial analysis It depends on how precise are the soil C simulations Predictions of future are difficult (harvests, thinnings) Without successful prediction of stand future model based stratification is not effective
Thank you for your attention Further information www.metla.fi/hanke/843002/ Project on Monitoring changes in the carbon stocks of forest soils and earlier projects www.metla.fi/hanke/3306/ www.efi.int/projects/integrated www.efi.int/projects/uncertainty and Raisa s email raisa.makipaa@metla.fi
References Liski, J., Palosuo, T., Peltoniemi, M. & Sievänen, R. 2005. Carbon and decomposition model Yasso for forest soils. Ecological Modelling 189: 168-182. Liski, J., Lehtonen, A., Palosuo, T., Peltoniemi, M., Eggers, T., Muukkonen, P. & Mäkipää, R. 2006. Carbon accumulation in Finland's forests 1922-2002 - an estimate obtained by combination of forest inventory data with modelling of biomass, litter and soil. Annals of Forest Science. Monni, S., Peltoniemi, M., Palosuo, T., Lehtonen, A., Mäkipää, R. & Savolainen, I. 2006. Uncertainty of forest carbon stock changes - implications to the total uncertainty of GHG inventory of Finland. Climatic Change, accepted. Peltoniemi, M., Mäkipää, R., Liski, J. & Tamminen, P. 2004. Changes in soil carbon with stand age an evaluation of a modeling method with empirical data. Global Change Biology 10: 2078-2091. Peltoniemi, M., Palosuo, T., Monni, S. & Mäkipää, R. The factors affecting the uncertainty of sinks and stocks of carbon in Finnish forest soil and vegetation. For Ecol Managem 232: 75-85. Manuscripts in preparation, see www.metla.fi/hanke/843002