Reducing uncertainty in carbon cycle science of North America: a synthesis program across United States and Mexico

50 %
50 %
Information about Reducing uncertainty in carbon cycle science of North America: a...

Published on December 23, 2016



1. Reducing uncertainty in carbon cycle science of North America: a synthesis program across United States and Mexico Rodrigo Vargas Department of Plant and Soil Sciences University of Delaware CoPIs: Nathaniel Brunsell University of Kansas Daniel Hayes University of Maine Contact: Agroclimatology PD meeting December 16-18, 2016 San Francisco, CA

2. Interagency Carbon Cycle Science (FY 2014) (2014-67003-22070)

3. • Synthesize new existing datasets and models across the United States (U.S.) and Mexico in a consistent analysis framework. …directed towards improving our understanding of forest and soil carbon dynamics, and the validation of terrestrial ecosystem models. The specific objectives: a) Harmonize available datasets b) Develop the synthesis approaches for scaling information c) Develop a to identify knowledge gaps. Objectives

4. Biederman, et al (2016) Global Change Biology 22:1867–1879. Villarreal, et al (2016) Journal of Geophysical Research-Biogeosciences 121:494-508. Petrie, et al (2016) Journal of Geophysical Research- Biogeosciences 121:280-294. Reimer, et al (2016) Progress in Oceanography 143 (2016) 1–12 McKinney et al (2015) IEEE 11th International Conference on e-Science: 108-117. Programa de Investigación en Cambio Climático (PICC) (2015) Reporte Mexicano de Cambio Climático. (Mexican Report on Climate Change. Group I: Scientific Bases, Models and Modeling). FAO and ITPS (2015) Status of the World’s Soil Resources (SWSR) – Main Report. Vargas , et al (2015) EOS, 96. doi:10.1029/ 2015EO037893 Reimer, et al (2015) PLoS ONE. 10(4):e0125177 Cueva, et al (2015) Journal of Geophysical Research-Biogeosciences 120:737-751. King, et al (2015) Biogeosciences 12:399-414 Milne, et al (2015) “Soil Carbon: science, management and policy for multiple benefits”. CABI. 10-25. Banwart, et al (2014) Carbon Management 5:185-19 Hengl, et al. SoilGrids250m: global gridded soil information based on Machine Learning (in review) PlosONE Vargas R, et al. (in review) Enhancing interoperability to facilitate implementation of REDD+: case study of Mexico. Carbon Management Publications

5. Soil carbon across North America - For decades the USA and Mexico have collected soil organic carbon (SOC) information. - Can we describe the spatial variability of SOC across North America? - Can we relate observations with biophysical information to predict SOC?

6. • Digital soil mapping (predictive soil mapping) - Computer-assisted production of digital maps of soil properties. - Use of field and laboratory observational (data and methods) with spatial and non-spatial inference systems. Digital soil mapping + many others

7. United States Mexico International Soil Carbon Network Federal agencies NRCS N=94778 1938-2010 INEGI Legacy Series 1 N=21153 1969-2001 USGS N=5623 1928-2006 INEGI Legacy Series 2 N=2805 1999-2009 Oak Ridge National Lab N=588 1992-2006 INEGI – National land degradation project N=2472 2008-2012 Other institutions (e.g. Universities, Long Term Ecological Research sites) N=2330 1905-2009 CONAFOR – INFyS N=3061 2009-2011 TOTAL=103319 analyzed samples TOTAL=29491 analyzed samples NRCS = Natural Resource Conservation Service USGS = United States Geological Survey INEGI = National Institute for Statistics and Geography CONAFOR = National Forest Commission SOC databases

8. SOC databases

9. 0-30cm n=12,360 SOC database for USA (ISCN)

10. • Randomized sample from INEGI series 1 & 2 0-30cm n=12,997 SOC database for Mexico (INEGI and CONAFOR)

11. SOC = f(S,C,O,R,P,A,N)+e Soil- soil type maps Climate, climatic properties Organisms, land cover and natural vegetation Relief, terrain parameters from DEM`s Parent material, geological maps Age, the time factor N, space, relative position e, autocorrelated random spatial variation Dokuchayev 1883->Jenny 1941->McBratney et al., 2003,-> Grunwald et al., 2011 Conceptual model for SOC variability

12. Model evaluation (e.g. cross valitacion, AIC, BIC, Cp) Variable selection (e.g. linear model) Prediction to new data (e.g. random forest Cubist) Uncertainty Estimation (e.g. different models, Global/local) SOC = f (Soils, Climate, Organisms, Parent material, Age, Space) + error

13. Digital soil mapping Guevara and Vargas (in preparation)

14. Hypothesis driven Machine learning * Median Statistical performance (explained variance)

15. Unexplained variability

16. Model r raca RMSEraca r inegi RMSE inegi MX (Pg) US (Pg) linear 5km 0.45 0.94 0.47 0.23 16.6 ± 2.16e-05 23.16 ± 2.96e-05 rf 5km 0.46 0.95 0.33 0.24 17.4 ±2.68-e05 21.53 ± 3.36e-05 SOC stocks across North America (PRELIMINARY) 29.3Pg for 0-30 cm depth (SSURGO; Bliss et al 2014) for CONUS 14.2 Pg for 0-20 cm depth (+- 3.9 Pg; Murray-Tortarolo et al 2015) for Mexico RaCA = Rapid Assessment of US Soil Carbon (USDA) INEGI = Series 1& 2 RaCA INEGI Soil carbon density: CONUS = 2.8 Mg km-2 Mexico = 8.5 Mg km-2

17. Towards a continental map of SOC for North America SOC next steps

18. - - This approach represents a regional baseline estimate of SOC (0-30cm) including variability - - Useful in future soil sampling planning (i.e. for inventory, SOC monitoring networks) aiming to reduce areas dominated by high variability - -This approach is reproducible (and semi automated) and can be periodically updated with new data and new covariates (Land use time 1, land use time2 and so on) Conclusions

19. Vargas et al (in review) Stakeholder Scientists Interoperability

20. Vargas et al (in review) Interoperability Interoperability is a collective effort with the ultimate goal of sharing and using information to produce knowledge and apply knowledge gained, by removing conceptual, technological, organizational, and cultural barriers.

21. Vargas et al (in review) Interoperability

22. Interagency Carbon Cycle Science (FY 2014) (2014-67003-22070) THANK YOU

Add a comment