Science Rationale Status of Deforestation Measurement Some Comments Dave Skole Tropical deforestation is related to: Carbon cycle and biotic emissions/sequestration Ecosystems and biodiversity Water and the hydrologic cycle Land use and cover change dynamics Natural resource management Measurement needs Special need for high resolution data Degradation, cf. logging (carbon) Spatial pattern of fragments, clearing, regen. (eco/biodiversity) Size and distribution of clearing for heat flux (water/climate) Spatial interaction of land use agents (LCLUC) Forest inventory and measurement (NRM) Main points for carbon Deforestation is dynamic losses and regeneration Deforestation is complex varies over time and place Deforestation occurs through a range of states, from very obvious clearing to subtle degradation Land use change leaves a mosaic of various cover types and cover states in the landscape These mosaics have memory, that is manifested in long term sources, and sinks in regrowth and soil OM storage Changes in stocks changes in area, changes in density -- and changes in fluxes, which vary with time
Geography and timing Some important issues include geography and timing Geography in the broader context to include spatial pattern Past deforestation may currently be regenerating; in regions where current deforestation is declining and there are larger regenerating areas (reflecting a history of large deforestation rates), such asynchronies may be important. Considerable evidence for large areas of regeneration, and for considerably variable rates of clearing Multiple changes in one landscape The current landscape is a mosaic, or record, of current and past land use and cover changes Variation exists at fine temporal and spatial scale Variation exists across classes of cover (from conversion) and within classes of cover (from modification or degradation) History has created a more complex landscape We know nothing about the processes which form this landscapes over time, nor do we have good measures (maps) of these landscapes themselves. Our prognostic ability is severely limited Observations: extent and density Carbon flux over space We can now focus on making direct observations of changes in forest extent (both increase and decrease) and density This can be done using frequent (annual) observations from high spatial resolution remote sensing in conjunction with a coupled land use-carbon models. This approach complements, but is more direct in determining the land use component, than use of other measures of changes in forest carbon from stand inventory data alone (Casperson et al. 2000)
Carbon flux over time 8 7 7.2 Net Flux 6 5 4 3 3.8 6.2 5.3 5.5 4.6 2 1 0-1 1.5 0.9 1 2 3 4 5-0.5 6 7 8 9 Year
Global Emissions from Land Use Change da = F + B -O-b 3.5 = 5.3 + 1.8-2.0 3.5 = 5.1 Difference is 1.6 -- The Missing Sink 90% due to deforestation [20% descrease Forest Area] Historically Total emissions of C [deforestation and fossil-fuel burning] 450 PgC [180-200 PgC from land use change] + 90 ppm CO 2 in the atmosphere 1 Pg C = 1,000,000,000,000,000 g C [40 ppm due to changes in land use] From 1850 to 1990 (a billion tones) 124 Pg emitted due to land use change 60% in tropical areas %40 in temperate areas Houghton et al. 1999, Houghton 1999, Defries et al. 1999, IPCC-TAR 2001 Global Carbon Budget - The fate of CO 2 Period 1990-1996 7.9 Pg C/yr (6.3 Pg Fossil Fuel) (1.6 Pg Land Use) After IPCC, TAR 2001 3.7 PgC/yr - Atmosphere 2.9 PgC/yr - Oceans 1.3 PgC/yr - Terrestrial Ecosystems 6,3 Fossil Fuel Annual Net Flux of Carbon (Pg) 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Net Annual Flux of Carbon from Changes in Land Use North America Tropical Asia Latin America Africa China 1840 1860 1880 1900 1920 1940 1960 1980 2000 Year Houghton 1999
Global CO 2 Budgets (Pg/yr) 1980 s 1990-95 Atmospheric Increase +3.3±0.1 +2.9 ±0.1 Emissions (fossil fuel, cement) +5.5 ±0.3 +6.3 ±0.4 Ocean-Atmosphere Flux -2.0 ±0.6-2.4 ±0.5 Land-Atmosphere Flux -0.2 ±0.7-1.0 ±0.6 - Atmospheric constraints of Global C sources and sinks - Location of Global C Sources and Sinks CO 2 Flask Network and Inverse Modeling Land-Use Change(80 s) +1.6 (0.5 to 2.4) Residual Terrestrial Sink - 1.8 (-3.7 to +0.4) IPCC, TAR 2001 NOAA-CMDL 1999 Inverse Model Estimates of CO 2 Uptake (7 Models) Biological C Sources and Sinks - 0.7 to - 2.4 Pg C/yr 0.0 Pg C + 1.6 Pg C/yr - 1.6 Pg C/yr - 0.0 Pg C/yr IPCC, TAR, 2001
Inverse Modeling Calculations of C Sources and Sinks North America: 1.6 PgC/yr Euroasia: 0.5 PgC/yr Pg C/yr Inverse Modeling Calculations of Terrestrial Carbon Sources and Sinks -0.5-0.3-1.3-0.1 Fan et al. 1998 TM2 1985-1995 GlobalView-CO 2 Ciais et al 2000 Current Terrestrial Sinks Potential Driving Mechanisms CO 2 fertilization Nitrogen fertilization Climate change Regrowth of previously harvested forests Reforestation / Afforestation Regrowth of previously disturbed forests Fire, wind, insects Fire suppression Decreased deforestation Improved agriculture Sediment burial Future: Terrestrial Carbon Management (e.g., Kyoto) Land Use/Cover Change
Slashing and Burning Forest Pasture Burned
Forest (selective logging) Logging access roads Burning Skidder trails Slashed Pasture Landsat ETM+, RGB 5/4/3 230/069-2000 (selective logging) Logging storage areas (patios) Current approaches: deforestation Tree fall gaps Stumps FAO: semi decadal survey Originally focused on country reports Somewhat unverifiable; aggregate Regional case studies: synoptic epochs Landsat Pathfinder: pan-amazon, C. Africa, SE Asia Spatial analysis with Landsat
Global Deforestation Based on Hot Spots Current approaches: Forest extent Global analyses of Land cover: forest extent for a date IGBP, UMd, USGS, FAO Percent Tree Cover: snapshot first time to introduce concept of density Change in Percent Tree Cover
Selective logging, degradation and cryptic deforestation
Selective logging detection Visual Interpretation Technique Automatic Analysis (textural algorithm band5) Obvious logging Subtle logging Buffer radius around log landings Note: The buffer radius for the State of Rondonia and Acre is 450 m. Other States buffer radius is 180 m.
Forest, Deforestation, and Regeneration - 1996 Mato Grosso,, Brazil, path/row 226/068 Forest, Deforestation, Regeneration and Logging - 1996 Forest, Deforestation, and Logging - 2002 Mato Grosso,, Brazil, path/row 226/068 Mato Grosso,, Brazil, path/row 226/068
Forest, Deforestation, Regeneration, Logging, and Fire - 2002 Selective logging detected in 1996 Mato Grosso,, Brazil, path/row 226/068 Selective logging detected in 1999
Validation: Mato Grosso case study Comparison between Landsat and Ikonos Methodology validation: Rondonia case study Validation: Rondonia case study Textural image (Landsat ETM+, Band 5) Buffer radius (450m) around log landings Forest inventory Patio, road, and tree locations
Validation: Rondonia case study (results) Concluding comments Small changes in large numbers Amazon has been the easy case Degradation will be important for carbon even if its less so for land use: logging is the easy case Accuracy for conventions may be more difficult than for carbon Data acquisition scenarios are bleak