Can a model transferability framework improve ecosystem service estimates? A case study of soil carbon sequestration in Tillamook Bay, OR, USA

Size: px

Start display at page:

Download "Can a model transferability framework improve ecosystem service estimates? A case study of soil carbon sequestration in Tillamook Bay, OR, USA"

Jean Manning
5 years ago
Views:

1 Can a model transferability framework improve ecosystem service estimates? A case study of soil carbon sequestration in Tillamook Bay, OR, USA Lauri Green* and Theodore DeWitt *ORISE, US Environmental Protection Agency ACES Conference Jacksonville, Florida 2016

2 Transferability Apply estimate/model from one site to another Previous site New site 2

3 Model transferability framework Goal: 3

4 Model transferability framework 1. Specify 2. Assess 3. Define 4. Obtain 5. Compare 6. Evaluate model/estimate needed logistical constraints acceptability criteria previous applications context variables model performance 4

5 Case study Forested watershed of Tillamook Tillamook 5

6 Carbon sequestration as an ecosystem service CO 2CO2 CO 2 CO 2 CO 2 CO 2 CO 2 Key Points Soils capture carbon & offset CO 2 emissions Industries buy carbon credits $$ CO 2 CO 2 Oregon regional leader CO 2 6

7 COMET Farm 1. Select model CarbOn and Management & Evaluation Tool DAYCENT model Carbon sequestration Tonnes CO 2 captured Converted to credit value $12.95 per tonne: calcarbondash.org Converted to kg C m -2 7

8 2. Assess logistical constraints Easy to use Cost Good documentation Spatially explicit Data available COMET-Farm X/ 8

9 3. Define acceptability criteria Coefficient of variation < 30% Tonnes CO 2 captured Carbon value ($) kg C m -2 9

10 4. Obtain previous applications Among sites Tillamook 10

11 4. Obtain previous applications Applied COMET Farm model to each site Data Entered 1. Shapefiles 2. Dominant vegetation 3. Age of forest (50 yrs) 4. Grow only scenario 11

12 Temperature Precipitation 5. Compare context variables Intrinsic vs extrinsic Not in the model (Extrinsic variables) Vegetation cover Vegetation species In the model (Intrinsic variables) Carbon sequestration in soil Bulk density Land management Ecotype Soil type 12

13 5. Compare context variables Site State Ecotype Soil type Vegetation PPT (mm) Temp C Canopy (%) Bulk (g cm -3 ) Lower Arkansas AR Wet Plains Unconsolidated Cropland Iroquois IL,IN Plains Carbonate Rock Cropland Gallatin MT Mountains Carbonate Rock Evergreen Non-acidic Clackamas OR Mountains Volcanics Evergreen South Fork Trinity CA Mountains Carbonate Rock Evergreen Non-acidic Upper Chehalis WA Mountains Volcanics Evergreen Non-acidic Tillamook OR Mountains Volcanics Evergreen Non-acidic Summer Lake OR Hills Volcanics Grassland, Scrub Smith MT Mountains Carbonate Rock Grassland, Scrub Upper Verde AZ Mountains Carbonate Rock Grassland, Scrub North Fork Humboldt NV Mountains Mixed Rock Grassland, Scrub Deep Fork OK Wet Plains Carbonate Rock Grassland, Scrub Clinton MI Artificial

14 5. Compare context variables Qualitative approach Site State Ecotype Soil type Vegetation PPT (mm) Temp C Canopy (%) Bulk (g cm -3 ) Tillamook OR Non-acidic Mountains Volcanics Evergreen Upper Non-acidic Chehalis WA Mountains Volcanics Evergreen Non-acidic Clackamas OR Mountains Volcanics Evergreen Summer Non-acidic Grassland, Lake OR Hills Volcanics Scrub

15 6. Evaluate model performance Site Qualitative approach State Million tonnes CO 2 captured Credit value $Millions COMET Farm kg C m -2 Tillamook OR Upper Chehalis WA Clackamas OR Summer Lake OR COV 98% 94% 22% 15

16 5. Re-compare context variables Soil variables Site State Ecotype Soil type Vegetation PPT (mm) Temp C Canopy (%) Bulk (g cm -3 ) Tillamook OR Non-acidic Mountains Volcanics Evergreen Upper Non-acidic Chehalis WA Mountains Volcanics Evergreen Non-acidic Clackamas OR Mountains Volcanics Evergreen Summer Non-acidic Grassland, Lake OR Hills Volcanics Scrub

17 5. Re-compare context variables Multivariate approach Site State Ecotype Soil type Vegetation PPT (mm) Temp C Canopy (%) Bulk (g cm -3 ) Tillamook OR Mountains Non-acidic Volcanics Evergreen Clackamas OR Mountains Non-acidic Volcanics Evergreen Upper Chehalis WA Mountains Non-acidic Volcanics Evergreen South Fork Trinity CA Mountains Carbonate Rock Evergreen Lower Arkansas AR Wet Plains Unconsolidated Cropland Deep Fork OK Wet Plains Carbonate Rock Grassland, Scrub Iroquois IL,IN Plains Carbonate Rock Cropland Clinton MI Artificial Gallatin MT Mountains Carbonate Rock Evergreen Upper Verde AZ Mountains Carbonate Rock Grassland, Scrub Smith MT Mountains Carbonate Rock Grassland, Scrub North Fork Humboldt NV Mountains Mixed Rock Grassland, Scrub Summer Lake OR Hills Non-acidic Volcanics Grassland, Scrub

18 5. Re-compare context variables Multivariate quantitative approach Similarity Hierarchical Cluster Analysis Site Extrinsic Precipitation Temperature Intrinsic Vegetation cover Bulk density 18

19 6. Evaluate model performance Multivariate approach Million tonnes CO 2 captured Credit COMET value Farm $Millionskg C m -2 Site State Tillamook OR South Fork Trinity CA Clackamas OR Upper Verde AZ COV 34% 33% 63% 19

20 5. Re-compare context variables Ecotype, Vegetation, Climate, Soil Site State Ecotype Soil type Vegetation PPT (mm) Temp C Canopy (%) Bulk (g cm -3 ) Tillamook OR Mountains Non-acidic Volcanics Evergreen South Fork Trinity CA Mountains Carbonate Rock Evergreen Clackamas OR Mountains Non-acidic Volcanics Evergreen Grassland, Upper Verde AZ Mountains Carbonate Rock Scrub

21 6. Evaluate model performance Site State COMET Farm kg C m -2 Tillamook OR South Fork Trinity CA Iroquois IL,IN Deep Fork OK Smith MT Gallatin MT Upper Verde AZ North Fork Humboldt NV Clackamas OR Upper Chehalis WA Summer Lake OR Lower Arkansas AR Clinton MI USDA kg C m -2 Soil Carbon Overview Framework Case Study Next Steps 21

22 6. Evaluate model performance Quantitative approach Site State % Difference USDA/COMET Tillamook OR Upper Chehalis WA 15 Clackamas OR 19 South Fork Trinity CA 25 Gallatin MT 59 Deep Fork OK 67 Upper Verde AZ 77 Smith MT 173 Lower Ark AR 195 Clinton MI 437 Summer Lake OR 440 Iroquois IL,IN 2094 North Fork NV 2117 Top three 22

23 6. Evaluate model performance Quantitative approach Million tonnes CO 2 captured Credit value $Millions COMET Farm kg C m -2 Site State Tillamook OR Upper Chehalis WA Clackamas OR South Fork Trinity CA COV 31% 31% 33% 23

24 5. Re-compare context variables Ecotype, Vegetation, Climate, Soil Site State Ecotype Soil type Vegetation PPT (mm) Temp C Canopy (%) Bulk (g cm -3 ) Tillamook OR Mountains Non-acidic Volcanics Evergreen Upper Chehalis WA Mountains Non-acidic Volcanics Evergreen Clackamas OR Mountains Non-acidic Volcanics Evergreen South Fork Trinity CA Mountains Carbonate Rock Evergreen

25 Conclusions No Meet acceptability criteria? Yes Done! Refine acceptability criteria Add application sites Select different model 25

26 Conclusions Goal: 26

27 Next steps Additional multivariate analyses Add continuous variables to HCA Include categorical variables DFA to identify groups and context variables PCA to identify context variables Apply additional models Yasso, ROTHC, InTEC 27

28 Acknowledgements Oak Ridge Institute for Science and Education Pat Clinton, Chanda Littles, Jari Liski, Mark Easter, Melissa Errend, JB Moon Contact me: Lauri Green 28

29 South Lake Trinity most similar to Tillamook Hierarchical Cluster Analysis Similarity Site 29

30 South Fork Trinity most similar Million tonnes CO 2 captured Credit value $Millions COMET Farm Site State kg C m -2 Tillamook OR South Fork Trinity CA Clackamas OR Upper Chehalis WA Upper Verde AZ

31 Look for additional similarities to reduce COV? PPT Temp Canopy Site State Ecotype Soil type Vegetation(mm) C (%) Non-acidic Tillamook OR Mountains Volcanics Evergreen South Fork Trinity CA Mountains Carbonate Rock Evergreen Bulk (g cm -3 ) 31

32 Estimate Transfer 3. Assess logistical constraints 4. Conduct estimate-transfer analysis 5. Is the result acceptable? Transferability Decision Tree Model Transfer 1. Specify the estimate wanted 1. Specify the model wanted 2. Define acceptability criteria Review & collect data from literature Define context variables Compare context niches Evaluate estimate variance at previous sites, similar context 2. Define acceptability criteria 3. Assess logistical constraints 4. Conduct model-transfer analysis Compile previous applications of the model Define context variables Evaluate model performance at previous sites, similar context 5. Evaluate performance Done! Yes No Revise acceptability criteria No Acceptable Done! Redefine question Use a model / a different model Methods if estimate transfer not advised Create new model using existing data Conduct primary research to obtain estimate or create model from site of interest Methods if model transfer not advised Overview Framework Case Study Next Steps 32

Geography and the Early Settlement of China

Geography and the Early Settlement of I N T E R A C T I V E S T U D E N T N O T E B O O K How did geography affect life in ancient? P R E V I E W Physical Features 1. Circle the physical features that