Long-term variations in leaf area index and light extinction in a Fagus sylvatica stand as estimated from global radiation profiles

The paper presents a model that estimates leaf area index (LAI) from above- and below-canopy global radiation measurements. The approach is based on a detailed description of radiation regime above and below the canopy. One EuroFlux site, a beech forest ( Fagus sylvatica L.) at Hesse, France was selected for model testing. The time-courses of daily LAI from 1996 to 2001 were simulated by the model. Since the simulated LAI curves include some random daily variations due to the quality of global radiation measurements, these were further smoothed by Fast Fourier transform and convolution using a 5-day window. The direct model output and the smoothed data by Fast Fourier transform and convolution using a 5-day window all agreed well with the in situ measurements. The r(2) of daily modeled LAI, Fourier transformed and convolution smoothed LAI with respect to measured LAI were 0.85, 0.89 and 0.84, and the corresponding RMSE were 0.56, 0.62 and 0.42 respectively. Relative evident noise was noted during the "leaf constant'' period in each year. The analysis of normalized deviation during this period in response to environmental parameters revealed that no statistical relationship could be found for daily average temperature, VPD, wind velocity and fraction of diffuse radiation. However, in 65% of cases with positive normalized deviation during the "leaf constant'' period precipitation occurred, indicating an effect of precipitation on model output. The daily and seasonal variation of the estimated extinction coefficient k was examined, since treating k as a constant may lead to large changes in the estimates of LAI via direct application of the Lambert-Beer law. Finally, the paper has also compared MODIS LAI predictions for the tower site during the year 2001 with the convolve smoothed model output. While a similar pattern is found during midsummer, MODIS underestimated LAI during the "leaf constant'' period. Larger as yet unexplained deviations are found during winter and spring before the "leaf constant'' period. Since the model is not very sensitive to weather conditions, it can be used to provide daily LAI from the daily global radiation measurements and offers an opportunity to include phenological information into gas exchange models which require LAI as input.