The Net Carbon Flux from Deforestation and Re-growth in the Brazilian Legal Amazon: Comparison of Bookkeeping and Process Based Approaches Adam I. Hirsch1, Richard A. Houghton1, William S. Little2 1The Woods Hole Research Center, P.O. Box 296, Woods Hole, MA. 02543, USA 2The Woods Hole Oceanographic Institute, Woods Hole, MA., USA Presented at the International LBA Scientific Conference 2nd Contact: Adam at [email protected] Bookkeeping vs. Process-Based Approaches • 1. 2. 3. Bookkeeping model (Houghton et al., 2000) Fixed forest productivity – proportional to mature forest biomass Doesn’t consider litter and soil carbon Mature forest biomass is taken from maps generated from measurements • 1. 2. 3. 3. Process-Based Model Productivity is a function of climate, soil qualities, and calculated Leaf Area Index Includes full water balance using Penman-Monteith evapotranspiration Carbon passes from the atmosphere, through detritus and soil carbon, and back into the atmosphere Mature forest biomass is generated internally by the model • Both use the same deforestation rates, but different land use dynamics The Net Flux Calculation F(t) = Net Carbon Flux in year t (1970-1998) fi = Fraction of deforested area in climate class i F(t)= fi *Ci a(t- ) *r( ) Ci = Total ecosystem carbon storage in climate class i i=1 τ =1 a(t-τ) = Annual Deforestation rate, ha/yr r(τ) = Flux response curve, normalized by steady state Ci Average biomass Adding contribution of deforested area from all years to present 100 ∑( 29 )∑ τ τ The CARLUC Carbon Cycle Model: a process-based carbon cycle model used to determine Ci and the net C flux due to disturbance and recovery, which are part of r(τ) Dynamic Land Use Change Model: Used to determine timing of land use transitions, which is part of r(τ) The CARLUC Carbon Cycle Model Atmosphere Wood Products Charcoal NPP Stems Leaves Coarse Woody Debris Roots Detritus Humus NPP = 0.012*0.45*α*f(SW)*f(T)*f(VPD)*PAR*(1-e-0.7*LAI) “Climate class” = index for scaling model results to the Basin level • NPP and water balance based on 3-PG (Landsberg & Waring, 1997) • α = optimal quantum efficiency • 0.45 = discounting factor for plant respiration • f(X) = multipliers representing water, temperature, and humidity limitations, viz. 3-PG • Allocation and live carbon turnover similar to CASA (Potter et al., 1997) • Climate inputs from East Anglia CRU 1961-1990 0.5o monthly averages • Radiation input from Dr. Rachel Pinker, 1990-1992 0.5o monthly averaged PAR • Soil Texture from Dr. Chris Potter (8 km) • Maximum PAW from Paul Lefebve at WHRC • Aggregating the Amazon into “climate classes” allows substantial savings over simulating every 64 km2 pixel. • During deforestation, 20% of biomass is combusted, 8% harvested for wood products, 2% is converted to charcoal, and 70% left to decompose, as in Houghton et al. (2000). m tu rn -0.1 -0.2 af t ov er u ro rn ot ov e CW tu rn r o D tu ver rn ov er al ph a CU E SL A m in T m a ra x T di at io hu n pr m ec us ip tu rn ov be e lo f hu r w fr a gr o u mu c s ab n ov d a l e to loc st em s le st e BETA CARLUC Sensitivity Testing 0.3 0.2 0.1 0 -0.3 -0.4 Parameter β = (total C+10% – total carbon-10%)/total Cdefault Ecosystem Carbon Partitioning – Rio Branco Mature Forest DPM 1% RPM 12% HUM 14% CWD 4% Live Wood 65% Fine Root 2% Live Leaf 2% DPM = Decomposable Plant Material HUM = Humus RPM = Resistant Plant Material CWD = Coarse Woody Debris IOM = Inert Organic Matter ~ 1.5*(DPM + RPM + HUM) Soil Carbon Model • Based on the Roth-C model (Jenkinson, 1990) • Calibrated to match Trumbore et al. (1995) C and bulk ∆14C measured to 8 m depth near Paragominas in 1992 • Used to calculate fraction of decomposition passing to humus vs. CO2 Ci = Steady State Ecosystem Total Carbon Storage Deforestation Pattern: fi Fraction of 64 km2 Pixels Classified as “Cleared” in TRFIC 1986 or 1992 Amazon Maps 100 ∑ f *C i ι =1 i = 145 tC/ha (slightly lower than Houghton et al., 2000) Aboveground Live Biomass Comparison 22 field sites from Houghton et al. (2001) : 148±43 tC/ha Model: 135±30 tC/ha Biomass Recovery Dynamics A G L B , tC /h a 100 CARLUC BASELINE 80 HOUGHTON et al. (2000) 60 Salimon & Brown (2000) 40 Uhl et al. (1988) light 20 Uhl et al. (1988) moderate 0 0 5 10 15 20 25 Years • Houghton et al. (2000) assumed 75% recovery of biomass in 25 yr • CARLUC predicts much slower growth, coincidentally falling equally close to observations from the dominant land use pattern in Eastern Amazonia (Uhl, Buschbacher & Serrão, 1988). • We have not yet included the initial high-productivity pioneer phase in the model Dynamic Land Use Change Model FAO Skole & Tucker, 1993 1 2 INPE 4 3 1 = Primary Forest Clearing: a(t-τ) 3 = Houghton et al. (2000) Abandonment 2 = Simulated Abandonment 4 = Simulated Re-clearing Primary Forest Agricultural Land Re-growing Forest Calibrated with 1986, 1992, 1996 TRFIC Land Cover Classification Maps Comparison with Satellite-Derived Information Cleared Forest Area A = Houghton et al. (2000) B = Land Use Model C = TRFIC maps Secondary Forest Area A = Houghton et al. (2000) B = Land Use Model C = TRFIC maps (too low) • Houghton et al. (2000): annual abandonment = 30% of annual deforestation, no re-clearing of secondary forest • Present model: approx. 7 years in agriculture followed by 7 years as re-growing before being re-cleared • TRFIC maps may underestimate secondary forest area (viz. Alves & Skole, 1996) r(τ) = Normalized Flux Response Curve – Combination of the 2 Models 3 Cycles of Cut and Burn, then 7 Years Fallow Flux Normalized by Steady State Carbon Storage Net Carbon Flux Results 1970-1998 Cumulative Net Release ~ 4.5 GtC Uncertainty (1 SD ~ 35%) estimated using Monte-Carlo technique for mature forest biomass and flux response Summary and Conclusions • Results are very similar to Houghton et al. (2000) • In present study, slower biomass recovery offset by greater area of re-growing forest predicted by land use model and slightly lower AGLB • Land use model would benefit from accurately coregistered multi-temporal Basin-wide land cover maps, allowing pixel by pixel change detection • Ecological model would benefit from more realistic treatment of secondary succession • We intend to use the model to study the carbon budget impact of Basin-wide logging and accidental fire
© Copyright 2026 Paperzz