continuous observation of vegetation indices, leaf area index (LAI) and fraction of absorbed photosynthetically active radiation (fPAR) using an integrated low-cost near surface sensing system

Abstract

Monitoring vegetation indices, fraction of absorbed photosynthetically active radiation (fPAR) and leaf area index (LAI) has advanced our understanding of biosphere-atmosphere interactions. Although there are continuous observations for each variable, monitoring vegetation indices, fPAR and LAI simultaneously is still lacking. Recent advances of technology provide unprecedented opportunities to integrate various low-cost sensors as an intelligent near surface observation system for monitoring ecosystem structure and functions. In this study, we developed a Smart Surface Sensing System (4S), which can automatically collect, transfer, process and analyze data, and then publish time series results on public-available website. The system is composed of micro-computers, micro-controllers, multi-spectral spectrometers made from Light Emitting Diode (LED), micro cameras, and Internet module. We did intensive tests and calibrations in the lab. Then, we conducted in-situ observations of normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), fraction of absorbed photosynthetically active radiation (fPAR), and leaf area index (LAI) continuously at a rice paddy field during the growing season. NDVI and EVI obtained by 4S showed linear relationships with those from a reference hyperspectrometer (R2 = 0.98

NDVI, R2 = 0.96

EVI). 4S derived fPAR and LAI were comparable to LAI-2200 and destructive measurements in both magnitude and seasonal trajectory. We retrieved vegetation indices, fPAR and LAI independently and continuously and show that after the reproductive stage, fPAR remained constant, whereas LAI and NDVI decreased continuously after their peak because of non-photosynthetic materials such as grain and yellow leaf. In addition, using vegetation index to estimate fPAR has limitation because the spectral reflectance could not capture the diurnal pattern. On the other hand, fPAR changes abruptly depending on the sky conditions and the amount of light transmitted. We believe that 4S will be useful in the expansion of ecological sensing networks across multiple spatial and temporal scales.