Evaluation of seasonal variations of remotely sensed leaf area index over five evergreen coniferous forests

Seasonal variations of leaf area index (LAI) have crucial controls on the interactions between the land surface and the atmosphere. Over the past decades, a number of remote sensing (RS) LAI products have been developed at both global and regional scales for various applications. These products are so far only validated using ground LAI data acquired mostly in the middle of the growing season. The accuracy of the seasonal LAI variation in these products remains unknown and there are few ground data available for this purpose. We performed regular LAI measurements over a whole year at five coniferous sites using two methods: (1) an optical method with LAI-2000 and TRAC; (2) a direct method through needle elongation monitoring and litterfall collection. We compared seasonal trajectory of LAI from remote sensing (RS LAI) with that from a direct method (direct LAI). RS LAI agrees very well with direct LAI from the onset of needle growth to the seasonal peak (R2=0.94, RMSE=0.44), whereas RS LAI declines earlier and faster than direct LAI from the seasonal peak to the completion of needle fall. To investigate the possible reasons for the discrepancy, the MERIS Terrestrial Chlorophyll Index (MTCI) was compared with RS LAI. Meanwhile, phenological metrics, i.e. the start of growing season (SOS) and the end of growing season (EOS), were extracted from direct LAI, RS LAI and MTCI time series. SOS from RS LAI is later than that from direct LAI by 9.3±4.0days but earlier than that from MTCI by 2.6±1.9days. On the contrary, for EOS, RS LAI is later than MTCI by 3.3±8.4days and much earlier than direct LAI by 30.8±7.2days. Our results suggest that the seasonal trajectory of RS LAI well captures canopy structural information from the onset of needle growth to the seasonal peak, but is greatly influenced by the decrease in leaf chlorophyll content, as indicated by MTCI, from the seasonal peak to the completion of needle fall. These findings have significant implications for improving existing RS LAI products and terrestrial productivity modeling.