MAPPING CURRENT AND FUTURE POTENTIAL SITE PRODUCTIVITY: ARE PROCESS-BASED MODELS NEEDED? Aaron Weiskittel, University of Maine School of Forest Resources Nick Crookston, US Forest Service Rocky Mountain Research Station Phil Radtke, Virginia Tech INTRODUCTION • Measures of site productivity are needed: – Growth and yield projections – Management regimes – Landscape classification • Most common means of quantifying in forestry is site index – Applicable to forested places • Other quantitative measures have been used with varying degrees of success: – Maximum MAI – Maximum basal area – Yield INTRODUCTION • Site index has often found to have a limited relationship with soil, climate, and physiographic variables (Carmean 1975) • Monserud et al (1990) offered several reasons for this: – Number of samples often low – High within- and betweenstand variability in estimates – Failure to measure the true causes of site productivity INTRODUCTION • Process-based models and remote sensing tools offered as a more effective tool for assessing site productivity (Swenson et al. 2005; Waring et al. 2006, etc.) – Mechanistic – Integrate soil and climate – Can map productivity across large regions and all biotic communities – Predict future productivity INTRODUCTION Latta et al. (2010) Coops et al. (2010) Both empirical and process-based models predict a change in future site productivity, but differ in the level of change RESEARCH QUESTION • How to better represent potential changes in future site productivity in an empirical growth and yield model like FVS? – Use a process-based model – Re-fit equations to include climate – Relate site index to climate RESEARCH OBJECTIVES 1. What is the relationship between climate and several measures of site productivity in the Western US? 2. What are driving variables influencing each measure of site productivity? 3. Does it matter what climate model used to make the assessment? 4. What is the future forecast for changing productivity? METHODS • Individual tree height-age data obtained from USFS FIA – n=83,016 – 61 different species – Douglas-fir most common (38%) • Estimates of site index standardized using Monserud (1984) equation – Flexible model form – Avoids species and regional differences – Strong correlation between standardized and observed site index • Tree-level estimates averaged for each plot (n=21,554) METHODS • Climate data obtained for each location based on fuzzed lat/long – USFS Moscow lab (http://forest.moscowfsl.wsu.edu/climate/; Rehfeldt 2006) • 1961 to 1990 • Monthly resolution • Developed using ANUSPLINE • Point estimates – DAYMET (http://www.daymet.org/) • 1980 to 2003 • Daily resolution • Truncated Gaussian weighting filter(Thornton et al. 1997) • Grid estimates (1 km2) METHODS • Climate variables used to develop different temperature and moisture indices identified by Rehfeldt et al. (2006) METHODS • Estimates of gross primary production (GPP) obtained from 2 sources based on fuzzed lat/long – 3-PG (Nightingale et al. 2007) • Relatively simple process-based model • Relies on monthly climate and soils (held constant) data – MODIS (Running et al. 2004) • Driven by a NASA satellite sensor • Daily estimates of current APAR, LAI, and surface climate • ε*(APAR/PAR)*PAR METHODS • Variable importance assessed using RandomForests – Nonparametric technique – Iteratively ran so that the least influential was dropped until only 2 left (~35 initial variables) • Maps of current and future site index generated from RandomForests model • Multiple models developed – Climate only – Climate and GPP – Climate, GPP, and physiographic variables (lat, long, elev) RESULTS Limited relationship between site index and GPP, but MODIS GPP was a significant improvement over 3-PG RESULTS • Both site index and MODIS GPP highly related to climate variables • Generally, 7 variables was most effective • 2 variables explained between 70-78% of variation Results of RandomForests Fit RESULTS Climate Climate/Physiographic Variable %IncMSE Variable %IncMSE prdd5 49.0 elev 83.7 mmax 48.3 mmax 77.1 mmindd0 36.0 sdi 72.6 maptd 33.7 prdd5 71.7 sdi 33.0 gsptd 60.0 gsptd 32.1 lat 59.1 tdgsp 25.7 maptd 55.6 RMSE (m) 4.59 RMSE (m) 3.75 • Relatively little difference between DAYMET and Moscow • Neither MODIS or 3-PG GPP in top 10 of influential variables for site index • Results dependent on whether physiographic variables included RESULTS RESULTS mindd0 gsp RESULTS Distributions of site index for future climate change scenarios Climate Mean change -6.3 m Climate/Physiographic Mean change +4.5 m RESULTS Climate Climate/Physiographic RESULTS Climate Climate/Physiographic RESULTS Climate Climate/Physiographic DISCUSSION • Process-model output significantly related to derived climate variables – Is the complexity necessary? • What’s elevation a proxy for? – Imprecise climate estimates – Solar radiation – Soils*Climate interaction – Gas concentrations • Is there a better measure of potential productivity? LIMITATIONS • Assumes a static relationship between site index and climate – Genetics? – Influence of CO2? • Role of past and potential future disturbances not considered • Uncertainty in stand-level estimates of site index not addressed – Only 1-4 sample trees/plot • Climate an effective predictor of current site index across western US CONCLUSIONS • Site index influenced by both temperature and moisture limitations – Elevation is an important variable in addition to climate • PNW Coast Range most vulnerable to a changing climate • Site index an imperfect measure, but allows relatively easy modification of FVS predictions • • • REFERENCES • • • • • • • • • Carmean, W.H., 1975. Forest site quality evaluation in the United States. Advances in Agronomy 27, 209-269. Coops, N.C., Hember, R.A., Waring, R.H., 2010. Assessing the impact of current and projected climates on Douglas-fir productivity in British Columbia, Canada, using a process-based model (3PG). Canadian Journal of Forest Research 40, 511-524. Latta, G., Temesgen, H., Adams, D., Barrett, T., 2010. Analysis of potential impacts of climate change on forests of the United States Pacific Northwest. Forest Ecology and Management 259, 720-729. Monserud, R.A., 1984. Height growth and site index curves for inland Douglas-fir based on stem analysis data and forest habitat type. Forest Science 30, 943-965. Monserud, R.A., Moody, U., Breuer, D.W., 1990. A soil-site study for inland Douglas-fir. Canadian Journal of Forest Research 20, 686-695. Nightingale, J.M., Coops, N.C., Waring, R.H., Hargove, W.W., 2007. Comparison of MODIS gross primary production estimates for forests across the U.S.A. with those generated by a simple process model, 3-PGS Remote Sensing of Environment 109, 500-509. Rehfeldt, G.E., 2006. A spline model of climate for the Western United States. In, General Technical Report RMRS-GTR-165. USDA Forest Service, Rocky Mountain Research Station, Fort Collins, CO, p. 21. Rehfeldt, G.E., Crookston, N.L., Warwell, M.V., Evans, J.S., 2006. Empirical analysis of plant-climate relationships for the Western United States. International Journal of Plant Science 167, 1123-1150. Running, S.W., Nemani, R.R., Heinsch, F.A., Zhao, M.S., Reeves, M., Hashimoto, H., 2004. A continuous satellite-derived measure of global terrestrial primary production. Bioscience 54, 547560. Swenson, J.J., Waring, R.H., Fan, W., Coops, N.C., 2005. Predicting site index with a physiologically based growth model across Oregon, USA. Canadian Journal of Forest Research 35, 1697-1707. Thorton, P.E., Running, S.W., White, M.A., 1997. Generating surfaces of daily meteorology variables over large regions of complex terrain. Journal of Hydrology 190, 214-251. Waring, R.H., Milner, K.S., Jolly, W.M., Phillips, L., McWethy, D., 2006. Assessment of site index and forest growth capacity across the Pacific and Inland Northwest U.S.A. with a MODIS satellitederived vegetation index. Forest Ecology and Management 228, 285-291.