Comparison of various methods for estimating the mean growing season percent photosynthetic photon flux density in forests

Five methods for estimating the mean growing season percent photosynthetic photon flux density (PPFD) were compared to continuous measurements of PPFD throughout the growing season within a young bigleaf maple stand on Vancouver Island (Canada). Measured PPFD was recorded continuously as 10-min averages over the growing season (May 18-October 14, 1996) using 52 gallium arsenide phosphide photodiodes in the understory and a LI-COR quantum sensor (LI-190SA) in the open. Photodiodes were randomly located on a systematic grid of points and represented a wide range of above canopy openings which were classified into three different types of light environments: closed canopy, gaps of various sizes, and open canopy. Objectives of this study were to compare different methods for estimating the, growing season %PPFD and to determine the efficiency of these methods in the three light environments. At each photodiode location, instantaneous light measurements using a Ceptometer on sunny days around noon and a LAT-2000 Plant Canopy Analyzer were made and hemispherical canopy photographs were taken. 10-min averages recorded by the photodiodes during completely overcast sky conditions were used as surrogate values for a method that uses instantaneous measurements on overcast days. Finally, a new light model (LITE) developed to estimate growing season %PPFD in a deciduous canopy was tested. All these five methods provided estimates of growing season %PPFD and are much less time consuming than continuous measurements of %PPFD using photodiodes. The three most accurate (r(2)>0.89) methods to estimate the growing season %PPFD were the 10 min averages on overcast days, the diffuse non-interceptance calculated using the LAI-2000, and the gap light index (GLI) calculated from the hemispherical canopy photographs. These three methods performed similarly in each type of light environment. Aid-rough the relationship between the LITE model and the growing season %PPFD was good (r(2)=0.79), the model systematically underestimated light transmission. The instantaneous sunny days around noon method was the least efficient method (r(2)=0.68) for estimating the growing season %PPFD, although replacing instantaneous measures with the mean of two 10-min averages improved r(2) to 0.84. Estimates on sunny days tended to be low in low light and high in high light. Practical considerations such as equipment availability, cost, sampling and processing time, sky conditions, and the number of microsites to be sampled should be taken into account in the selection of the suitable method for a particular study. (C) 1998 Published by Elsevier Science B.V. All rights reserved.