Estimating wood growth across the Hubbard Brook valley: an analysis linking imaging spectroscopy and ecosystem model PnET

Printer-friendly version

Spatial wood growth across the Hubbard Brook valley, New Hampshire, U.S.A. was estimated by an ecosystem model PnET-II. The valley was delineated into grid cells in a 30-m resolution. For each cell, PnET-II was run for either of deciduous stands or evergreen stands, or both for the mixed. The mountain microclimate model MTCLIM was applied to estimate the climate variables (maximum and minimum temperature, precipitation, and PAR) across the valley. The canopy N was estimated by hyperspectral remote sensing from SpecTIR. Remote sensed canopy N was validated against 18 field plot canopy N measurement. We used an empirical approach that allowed partitioning of image-derived % N estimates into the relative proportions of deciduous and evergreen forests and the foliar N concentration of each component. The model was also valiated using the 18 plot wood growth measurement from the valley wide forest surveys in 1995 and 2005. The surveyed wood growth shows the forest at Hubbard Brook is no longer aggrading. After validation, PnET-II was run to estimate the wood growth across the whole valley.  Predicted wood growth ranged from about 150 g m-2 year-1 to 460 g m-2 year-1 with a mean of 330 g m-2 year-1. Aspect trends in predicted wood growth indicate that wood growth in south facing area is greater than in north facing area, reflecting the remote sensed canopy trend and the fact that south facing areas are less shaded with higher solar radiations. There are not significant trends in wood growth with slope, temperature, and precipitation. The elevation trend is nonlinear, increasing from low to mid elevations and then decreasing at upper-elevation sites where evergreen stands become increasingly dominant. Spatial patterns also reveals that experimental watersheds 5,4, and 2 show in order the highest wood growth rates that corresponds to the their ages after clearcuts. The combined effect of forest management, soil, and species composition is reflected on the observed patterns of canopy nitrogen in these watersheds, which controlled the productivity. Overall, canopy N is the primary driver to regulate wood productivity. It proved the importance of remote sensing of canopy N in this study. Linking remote sensing with process-based ecosystem models can be an important approach to provide critical information about forest dynamics at landscape scales.