Accounting Greenhouse Gas Emissions for Irish Agriculture: Known/Unknown - M.I. Khalli
Climate Change Research Programme EPA Greenhouse Gas Modelling Workshop held on 24 November 2010 in the Gresham Hotel, Dublin
Reporting of total anthropogenic emissions &removals of GHG to EU & UNFCCC Six sectors: 1. Energy 2. Industrial processes 3. Solvents and other products 4. Agriculture 5. Land Use, Land Use Change and Forestry 6. Waste
Methodological Development Tier 1: Simple approach, relies on default emission factor (EF) drawn from previous studies and even somewhat on Activity Data (AD) Tier 2: Complex approach, requires detailed country-specific data derived from enhanced characterisation- disaggregated. Tier 3: Models (ecosystem/process-based), taking into account the country-specific measured data as well as soil and environmental conditions. Moving from Tier 1 to Tier 2 and 3 depending on robust data available under Irish conditions
Carbon and Nitrogen Accounting • Tier 2 approach over Tier 1 would provide better estimates, depending on the variability of soil organic carbon (SOC) and N dynamics. • Tier 3 approach reflects robust emission accounting and identify mitigation options but needs to include more variables regulating GHG emissions. • Tier 3 also provide a flexible and a robust way to assess how different scenarios and measures for land use management (LUM) and change (LUC) can affect soil C and N dynamics. • Even the best models require measurement-based validation at field scale and therefore benchmark sites are required. • Combining modelling and geostatistical techniques may be a better option to assess and project soil C and N stocks/emissions.
Preliminary Concepts: SOIL CARBONMONITORING, ACCOUNITNG & REPORTING Step 6: Develop SOC Map (Arc- GIS) and update/improve data Step 1: Data acquisition (National Database & Others Step 5: Total C stock by • Identify locations, relevant) integration, meta-analysis missing land parcels & land transition factors & new soil C data. • Predict coefficients of change for major land use categories. • Update LULUCF. • Prioritise research gaps. Step 2: Data Compilation (Depth Distribution: LU, Soil Step 4: Develop 3D SOC type, Climate, etc.) model by substitution of empirical models Step 3: Synthesize/develop empirical models using pedotransfer functions
Step 1: Data Acquisition CORINE Land Cover (LC), National Soil Database (NSDB), Kiely et al. (2009), Land Parcel Information System (LPIS), Soil Maps and Others• 1 km Buffer on Irish National Grid: SOC under a LC contains a Great Soil Group (GSG) >50% area
Number of Sites/Land Cover and Great SoilGroup (GSG) represented Grassland Rough Arable Others Gleys 83 10 7 Podzols 15 3 NA Brown Podzolics 50 1 12 Soil depth: 0-10 cm 111 SOC (confidence 75%), no Bulk Grey Br Podzolics for 9 16 density Brown Earth 66 NA 5 Some anomalies in representing5major soil group Lithosols 3 NA Specific LU absent 4 Rendzinas 2 NA Peats 18 21 6 (?) Regosol/Sand 0 0 0 Total 350 51 46 581 NA = Not available
Number of sites and GSG represented Kiely et al. (2009) database Grassland Arable Rough Forest Peat 29 (7) 12 (4) 10 (4) 9 (5) 11 (3) Soil depth: 0-50 cm, no matching SOC with bulk density (BD) Representation of all GSGs under a LC is not available Specific LU information, as of NSDB, are absent SOC contents are highly variable with NSDB.
Step 2: Data Compilation(Depth Distribution: LU/LC, Soil type, Climate, etc.) • In addition to 50 cm depth, SOC for arable and grassland measured at 100 cm depth are also included. • Non-linear relationship between soil depth, SOC and bulk density (BD) are adopted. • Empirical equations are developed to estimate SOC and BD (to calculate soil mass) down to 100 cm except Rendzinas to 50 cm.
Step 3: Synthesize/develop empirical models using pedotransfer functions Data for SOC in the NSDB are up to10 cm depth and that original data are taken to calculate its stocks as: SOC (Z 10cm) = SOCz10 SOC for depths (Z) >10 cm are calculated using empirical models developed from the measured/interpolated SOC ratio functions with depth as: SOC (Z >10cm) = a e(b*z)*SOCz10 Due to lack of BD information in the NSDB, empirical models are also developed from measured/interpolated data to calculate it, as: BD (Z=10-100 cm) = a e(b*SOCz)
SOC distribution ratio with soil depth: GrasslandGreat Soil LC Specific LCS (All)Group Soil Type Specific (STS, Mean) (LCS, Mean)Gleys 1.3397*e(-0.034*z)*SOCz10; (R2 = 0.998) 1.3620 1.3071 * e(-0.035*z) *e(-0.034*z)Podzols 1.4432*e(-0.040*z)*SOCz10; (R2 = 0.953) *SOCz10 *SOCz10Brown Podzolics 1.4275*e(-0.035*z)*SOCz10; (R2 = 0.999) (R2 = 0.999) (R2 = 0.894)Grey B. Podzols 1.2800*e(-0.034*z)*SOCz10; (R2 = 0.995)Brown Earth 1.4356*e(-0.034*z)*SOCz10; (R2 = 0.999)Lithosols a 1.0611*e(-0.057*z)*SOCz10; (R2 = 0.974)Rendzinas b 1.9042*e(-0.040*z)*SOCz10; (R2 = 0.968)Peats c 0.9206*e(-0.037*z)*SOCz10; (R2 = 0.918)Sand d 0.8167*e(-0.019*z)*SOCz10; (R2 = 0.890)a= df rough; b= df IFS 12, 22 &31, rep BE & peat mineral; c= df from both grass * peat; d= Original
BD from pedotransfer function (SOC): GrasslandGreat Soil STS (Mean) LCS (Mean) LCS (All)GroupGleys 1.4725*e(-0.085*SOCz); (R2 = 0.998) 1.3582 1.3949 *e(-0.074*SOCz); *e(-0.084*SOCz);Podzols 1.7859*e(-0.104*SOCz); (R2 = 0.918) (R2 = 0.990) (R2 = 0.643)Brown Podzolics 1.1509*e(-0.044*SOCz); (R2 = 0.964)Grey Br. Podzols 1.4306*e(-0.089*SOCz); (R2 = 0.998)Brown Earth 1.2400*e(-0.047*SOCz); (R2 = 0.988)Lithosols a 0.8593*e(-0.033*SOCz); (R2 = 0.908)Rendzinas b 1.1730*e(-0.050*SOCz); (R2 = 0.936)Peats c 1.1078*e(-0.003*SOCz); (R2 = 0.830)Sand d 1.1858*e(-0.0025*SOCz); (R2 = 0.956)a= df rough; b= df IFS 12, 22 &31, rep BE & peat mineral; c= df from both grass * peat; d= Original
Rough and Arable SOC distribution ratio with soil depth BD from pedotransfer function (SOC)
Step 4/5: Depth distribution of SOC stocks for each GSG STS equations better represent SOC stocks with depth for a particular soil.LCS would provide similar estimate ofSOC stocks in a LC but either over- orunder-estimate for a soil type
Depth distribution of SOC stocks for major LC± peat Large variability in SOC stocks under a LC can be reduced by separating peats from other soil types STS could best estimate of SOC density. For 0-30 cm: Grass = 1 Rough = 1.57 (+67 t) Arable = 0.74 (-30 t) Representative samplings for peats could better estimate SOC under a LC.
LU areas covering IS and GSG derivedfrom overlaying LPIS, GSM and ISM ISM/GSM LPIS Map
OC stocks (STS) in Indicative soils (IS) & GSG SOC stocks are calculated using the equations developed but covering soils of ISM and GSM • Giving higher level of disaggregation for SOC across soil depth
OC stocks (STS) in IS & GSG Showing higher SOC stocks than grassland in all soil types and depths
SC stocks (STS) in IS & GSG Demonstrating lower SOC stocks than grassland and rough. Peats under arable are misplacement/anomalies
Disaggregated total SOC stocks (STS)under grassland (LPIS 2004) Calculation: LC LU GSG ISM Area (ha) Pasture = 4,328,569 Rough = 3,185 Hay = 81 Silage = 1,173 Total = 4,333,009 Disaggregation of grassland using LPIS is non-realistic due to identification problems of LU by farmers but CSO
Disaggregated total SOC stocks (STS)under arable crops (LPIS 2004) CSO reported area = 424,000 ha: This underestimation is related to areas misplaced /identification error in the LPIS but exist, requiring re-synthesis
National SOC stocks: Other stocks derivedfrom Eaton et al. (2008)
Conclusions and further studies The empirical approaches provide robust estimate of SOC stocks for the development of Tier 2 through 3 and thereby for LUC. It can further be improved through elimination of following anomalies: * Missing/misplaced LU area in the LPIS * Missing SOC data for soil types under various LU * Inclusion of LUM and Input categories in the LPIS, advantageous Update/improve data for LPIS and refine SOC & develop Maps (Step 5 & 6). Accounting N2O emission for Irish agriculture using same data sources. LULUCF: Land transition factors (LU, LUM & Input) through Meta-analysis, leading to Tier 2 development. Develop/validate models for GHG accounting through geo-regression using LU, soil & environmental variables . Identify research gaps
Acknowledgements Christoph Müller and Tom Bolger, UCD Phillip O’Brien and Frank McGovern, EPA Ger Kiely, UCC Gary Lanigan and Karl Richards, Teagasc Researchers from UCD, TCD, UL, UCC... ….