A LiDAR Driven 3D Model For Canopy Sun-induced Chlorophyll Fluorescence

Sun-induced chlorophyll fluorescence(SIF)has been used to be an indicator of photosynthesis of vegetation.The observation of SIF on vegetation at different spatial and temporal scales can reflect the actual photosynthetic and physiological state of vegetation.However,there are still many uncertainties in the process of observing,analyzing and utilizing SIF.Understanding the interaction between SIF and vegetation structure from the mechanism and analyzing the main factors affecting the activation of SIF will help to better understand the internal relationship between SIF and photosynthesis and biomass.Thus,radiative transfer models are necessary tools to interpret and utilize SIF signals.SIF signals are relatively weak and need to be quantitatively described under the influence of various factors such as environment,vegetation and physiology,which brings challenges to the construction of SIF radiative trasfer models.In recent years,many scholars have developed SIF radiative transfer models in different scales,providing a solid theoretical basis for the development of SIF remote sensing.However,there are many assumptions in the existing canopy models,and most of the simulated scenes are virtual,which cannot stand for the SIF distribution and canopy SIF signal in real scenes.In addition,ray tracing method is usually used to simulate the radiative transfer process in 3D models,which is not efficiently with large computation time.This research developed a model to simulate the 3D distribution of SIF based on Li DAR and radiative transfer processes.First,we collected model input dataset in Baima plot including Li DAR data,direct and diffused photosynthetically active radiation(PAR),canopy SIF,leaf biochemical parameters etc.To construct the model,we first incorporated a leaf model to simulate SIF at leaf level.Then we developed an upscaling method to derive canopy SIF.Finally we validated the model through comparing the simulated diurnal variation of SIF with measured canopy SIF with the same footprint.Based on the model we characterized the 3D distribution of SIF in Baima forest.The main research contents and conclusions are as follows:(1)Establishing the 3D distribution of SIF method based on the 3D distribution of canopy PAR.We incorporated the Fluspect model,the atmosphere radiative transfer module in SCOPE,multiple scattering coefficience of sunlit and shaded leaves in 4-scale model and finally simulated the 3D distribution of SIF successfully base on Li DAR data.This method also included the upscaling method from leaves to canopy to simulate the canopy SIF with considering the multiple scattering effect of SIF.All the codes were implemented in Matlab.(2)Model validation.In this study,we carried out field experiment to collect model input data and canopy SIF observation simultaneously for model validation including the experiments for single tree and forest canopy.In the experiment of single tree,there was little difference between simulated canopy SIF and measured SIF(R~2=0.9357).In Baima plot,Nanjing,the simulated SIF was consistent with observed canopy SIF before 11:00,but lower than the observation result during12:00-14:00,and declined faster after 15:00.The R~2 between simulated SIF and the observed SIF was 0.7127 and 0.7216 with 1-minute and 10-minute interval respectively.The trend of simulated SIF was very much in line with that of BF5logging PAR.And the R~2 between them was above 0.9 with both 1-minute and10-minute interval.Through the sensitivity analysis,we found neighbouring point distance(NPD)is a key factor to the model precision.The R2 and RMSE between the simulated and measured data declines and increases with the increase of NPD when NPD is above 0.2 meter.There is a significant increase in canopy SIF with the increase of chlorophyll content,especially when the chlorophyll content under50?g/cm~2.(3)Characterizing and quantifying the spatial-temporal distribution of SIF.In this study,we characterized the spatial-temporal distribution of SIF and PAR based on the model we developed and in-situ data in Baima plot.The simulated PAR was separated into photosynthetic and non-photosynthetic part and SIF was simulated base on the PAR in photosynthetic part.We found that this part of SIF distribution had the similar pattern with PAR.The difference in the value and spatial distribution of SIF excited by direct irradiance was small in the morning and afternoon.At noon,the value of SIF in the whole scene in larger,espetially the unshaded leaves at the top and outer edge of the canopy.The high-value area of SIF excited by direct irradiance moved with the change of solar azimuth and zenith angle.However,the distribution of SIF excited by diffuse irradiance showed low little difference in pattern during a day but high variation with height for the method of diffused PAR simulation.In a single tree,SIF generally increases with height.In this study,we developed a model to simulate the forest 3D distribution of SIF in a real scene by incorporating leaf and canopy radiative transfer models as well as the upscaling method.All the input data were collected by field experiment.The performance of the model is up to expectations but still remain some problems.In the future,many works can be done to improve the method,reduce model assumptions and carry out cross validation between models.