Estimation of total, direct and diffuse PAR under clear skies in complex alpine terrain of the National Park Berchtesgaden, Germany

Information about total, direct and diffuse photosynthetically active radiation (PAR) is required in simulation models to estimate carbon gain and growth of vegetation. While there are existing worldwide networks of stations where direct, diffuse and reflected global radiation are measured using standardized methodology, no such network exists for PAR. In complex mountainous terrain, few published studies have examined even global radiation distribution as influenced by topography. We have developed a model to estimate the total, direct and diffuse photosynthetically active radiation in complex terrain. The model includes: (1) a parametric atmospheric model to extrapolate atmospheric conditions to any given location in complex terrain (which directly determines the potential PAR radiation that can be received on a horizontal surface) and (2) a topographic model, which accounts for the alteration of PAR radiation caused by terrain. Validation of the model was undertaken first for a baseline valley site at Schönau, an essentially flat surface with simple surrounding terrain. The hourly step atmospheric conditions were fit using direct PAR measurements at Schönau, and then applied in the parametric model to simulate the total and diffuse PAR. The results demonstrate that the parametric model provides good and fairly good simulations for total and diffuse PAR, respectively. In a second step, the model was tested for total PAR at five sites with distinctive topographic characteristics in the National Park Berchtesgaden, Germany, and for direct and diffuse radiation at two of the five sites with direct and diffuse PAR measurements. The model simulated total PAR well with high R2 (all >0.90). The NRMSE varies from 8% to 26% depending on sites. Although high R2 were found for direct PAR at the two sites Kederbichl and Bartholomä (both 0.92), lower R2 of 0.64 and 0.65 were obtained for the diffuse PAR simulation. While the NRMSE for diffuse PAR was also lower (0.23 for Kederbichl site and 0.26 for Bartholomä site), this difference was not as pronounced (0.31 for Kederbichl site and 0.28 for Bartholomä site for direct PAR). The R2 decreases to 0.14 and 0.33 for the two sites in the diffuse PAR simulation, if the diffuse PAR is treated isotropically. Thus, the results suggest that consideration of anisotropic distribution of diffuse radiation is required in PAR extrapolation models within complex alpine terrain.