Assessment of Forest Productivity Using an Ecosystem Process Model, Remotely Sensed Lai Maps and Field Data

Abstract

The process-level ecosystem simulation model Biome-BGC was run aspatially for a highly managed forest, the Idarwald. The area (49°45’N, 7°10’E) is climatically dominated by oceanic influences and mainly consists of managed stands of Norway spruce (Picea abies). Biome-BGC was run for 90 years using model parameters adapted to the local environmental conditions. The parameters include meteorological input data (1988-2002) from nearby Morbach weather station, ecophysiological constants for Picea abies and initialisation variables from Running & Gower (1991). Annual whole-plant mortality (AMT) fraction was adapted to the situation of a managed forest, where biomass is removed during thinning and logging measures in developing stands. AMT was calculated to a value of 0.017 yr -1 . Fire mortality fraction was set to zero. Model outputs were compared to remotely sensed leaf area index (LAI) and field measured variables (stem volume and radial stem growth). Airborne imaging spectrometer (HyMap) data had been used to derive an LAI Map of Idarwald forest (cross-validated RMSE = 0.54 m 2 m -2). Age courses of modelled and remotely sensed LAI both had a maximum of 5 m 2 m -2 at about stand age 15 and decreased to 3 m 2 m -2 at age 80. Measurements of stem volume [m 3 ha -1 ] were converted to stem carbon (SC) [kgC m -2 ] and compared to modelled values. At age 35, Biome-BGC overestimated SC by factor 2. From age 55 to 75 modelled SC was in good agreement with measured SC. Radial stem increment (RSI) data [mm yr -1 ] was available for the period 1990-2002 for a test plot located nearby Idarwald forest. Model outputs of "C pool to total stem C" [kg m -2 d -1 ] were converted to RSI. Modelled and measured values of RSI agree for most of the years. The RMSE between modelled and measured RSI (n=9) was relatively low (0.10 mm yr -1). In a summary, model outputs of Biome-BGC did relatively well agree with measured data. Obviously, Biome-BGC reasonably simulated ecosystem processes in a managed forest.