Nationwide Biomass Modeling of Bio-energy Feedstocks Chris Daly, Mike Halbleib, Matt Doggett David Hannaway Sun Grant Western Region GIS Center Oregon State University Corvallis, Oregon, USA GROUP PRISM
Introduction GIS Program Objective: Gain an understanding of the spatial distribution of current and potential biofuel/bio-energy feedstock resources across the country Envisioned outcome: A series of national geo-referenced grids that describe the actual and potential productivity patterns of various feedstocks
Methods to Accomplish This Currently Collecting production information from field trials and the literature; some regions developing models to make spatial estimates Issues Data representativeness- Data are taken from relatively few locations under widely varying management practices and in different years, and span small portions of the environmental gradient – very messy, difficult to extrapolate from data alone Regional consistency - Difficult to coordinate regional results into a national, “wall-to-wall” assessment that is consistent across the country “Potential” not the same as “existing” - Unclear how potential biomass production of new crops will be estimated nationwide, esp. under future climates
An Environmental Suitability Modeling Framework Two main objectives: Develop gridded estimates of current and potential feedstock resources across the entire conterminous US, constrained by climate, soil, and land use patterns Provide a spatial framework for biomass data collection and field trials: What additional data do we need and where?
Biomass Yield Internet Map Server Percent of Maximum Yield SSURGO Soil Maps Environmental Model PRISM Climate Maps Internet Map Server Biomass Yield Observed Yield Terrain/Land Cover Constraints
Environmental Model “Limiting Factor” Approach Relative Yield (0,100%) = Lowest production resulting from the following functions: Water Balance Winter Low Temperature Soil Properties pH Salinity Drainage
Monthly Water Balance Model P ETa KS Kcmid AWC TAW Droot Dr
Monthly Water Balance Model Water stress coefficient KS = (TAW - Dr) TAW = = total avail. water cont. = AWC Droot AWC = avail water content (NRCS data) Droot = rooting depth* Root zone moisture depletion Dr = Drm-1 + (Eta(m-1) – Pm-1) ETa = actual evapotranspiration P = precipitation Evapotranspiration Eta(m-1) = ET0(m-1) KS(m-1) Kcmid ET0 = Reference evapotranspiration (based on PRISM climate data) Kcmid = Crop coefficient, mid-growth stage* * User input
Monthly Water Balance Model Temperature coefficient C2 = (max*-Tday)/max*-optimum*) * User input Winter Wheat Monthly Relative Yield (water balance) RY = KS (C2L e ((L/R)(1-C2R))) Water stress coefficient Temperature growth curve
Final Water Balance Relative Yield Calculating Final Water Balance Relative Yield Mo J F M A S O N D RY 5 50 90 80 30 70 60 10 GP 1 FP Growth Period* N=3* Floating N-month* max yield Final RY = N-month max average RY within the Growth Period * User input
Winter Temperature Constraint Function Low End - Winter survival High End - Chilling Requirements Winter Wheat
Soil Constraint Functions Soil pH Soil Salinity Winter Wheat Soil Drainage
Environmental Model Relative Yield Dryland Winter Wheat, Local Varieties
Biomass Yield Internet Map Server Percent of Maximum Yield SSURGO Soil Maps Environmental Model PRISM Climate Maps Internet Map Server Biomass Yield Observed Yield Terrain/Land Cover Constraints
“Usable” Land Cover and Terrain Masks Forest, Urban, Tundra omitted Ag, Grass, Shrub, Savanna allowed Terrain Slopes > 7% omitted Local high ridges and peaks omitted
Environmental Model Relative Yield Dryland Winter Wheat, Local Varieties, “Usable” Land
RMA Reported Yield, 2000-2009 Mean Dryland Winter Wheat, All Varieties/Management, County Average
RMA Reported Yield, 2000-2009 Mean Dryland Winter Wheat, All Varieties/Management, “Core” Counties Only “Core” Counties 30-30-30 ≥30 model cells/county ≥30% of county “usable” ≥30 RMA reports/county
National Relative Yield vs. RMA Reported Yield Dryland Winter Wheat, All Varieties/Management, “Core” Counties Only
Final Winter Wheat Straw Yield Dryland Winter Wheat, Natl. Regr., All Varieties/Mgmt, “Usable” Land 0.4 Harvest Index
Using PRISM to Transform Relative Yield to Actual Yield For each pixel, PRISM develops a regression between modeled relative yield and RMA county-average yield. The result is a yield map that has been locally transformed from relative yield to actual yield. Observed Yield (Bu/ac) Modeled Relative Yield (% of optimum)
Final Winter Wheat Straw Yield Dryland Winter Wheat, PRISM Regr., All Varieties/Mgmt, “Usable” Land 0.4 Harvest Index
National Relative Yield vs. RMA Reported Yield Dryland Winter Wheat, All Varieties/Management, “Core” Counties Only
Dryland Winter Wheat, National Regr., All Varieties/Management Outlier Counties Dryland Winter Wheat, National Regr., All Varieties/Management
Grant County, Washington RMA County Average Yield for Dryland Winter Wheat Higher than Modeled PRISM Precipitation Dryland Wheat Cropland Data Layer (30 m) Irrigated Crops RMA county average reflects higher precipitation area in northern corner of county Modeled county average reflects nearly all land in county
RMA County Average Yield for Dryland Winter Wheat Higher than Modeled Wood County, Ohio RMA County Average Yield for Dryland Winter Wheat Higher than Modeled Corn/Soybean (white) Dryland Wheat (brown) SSURGO Soil Drainage (4 km) Cropland Data Layer (30 m) Native soils very poorly drained, reduced yields in model Actual - field modification of soil drainage
RMA County Average Yield for Dryland Winter Wheat Higher than Modeled Wood County, Ohio RMA County Average Yield for Dryland Winter Wheat Higher than Modeled Corn/Soybean (white) Dryland Wheat (brown) Cropland Data Layer (30 m) SSURGO Soil pH (4 km) Native soils very acidic, reduced yields in model Actual - field modification of soil pH
Environmental Model Relative Yield Switchgrass, Lowland Varieties, VERY PRELIMINARY
Switchgrass Modeled Yield Maps Lowland Cultivars (Tulbure et al., In Prep) All Cultivars (Wullschleger et al., 2010)
Next Steps Update environmental input grids and further refine model Update PRISM and SSURGO datasets Model on a time series, not just long–term mean data Soil constraints – perhaps don’t include in model - think of as costs of management (P. Woodbury) Collaborate with ORNL, GIS regions, and species teams Share data Share expertise Collaborate on final products KDF connections GROUP PRISM
Next Steps Develop potential biomass maps for nationally important feedstocks Corn stover Switchgrass Other small gain residue Willow, Poplar Energycane Sorghum Grasses on CRP land Miscanthus GROUP PRISM
Sun Grant Yield Data are Essential Validate, transform, and improve modeled yields No RMA data for many crops Use modeled maps to identify locations where field validation is needed (e.g., Great Plains precipitation gradient), and where data might be suspect The more (good) data we have, the better our maps will be GROUP PRISM