Stratifying forest overstory and understory using the Global Ecosystem Dynamic Investigation laser scanning data

Abstract

Most multi-layer natural forest structures, usually containing obvious vertical structures including overstory, understory, and grass, show significant differences in structure, phenology variations, and photosynthetic capacity. However, it is still challenging to extract the waveforms of overstory and understory in forests with varied canopy covers and topographic conditions using the Global Ecosystem Dynamic Investigation (GEDI) data. This work successfully used the GEDI L1B data to extract the waveforms of the overstory and understory, taking the multiple peaks in the waveform, canopy cover, and terrain slope into consideration and investigated the effects of the GEDI's geolocation errors, canopy cover, overstory height, and terrain slope on the forest stratification. Our results showed that: (1) The GEDI-based mean overstory height bias was −6.50 m with an RMSE of 7.07 m and improved by 1.40 m and 2.53 m using our method, respectively. The GEDI-based mean understory height bias was 2.85 m with an RMSE of 3.83 m. (2) The correlation (R2) of ALS- and GEDI-based canopy cover in forests without understory was 0.11 higher than that in those with understory. (3) The topography, canopy cover, overstory height, and geolocation showed apparent effects on forest stratification using the GEDI L1B data. Our study provides a solid foundation to better understand the multi-layer forest structure in terms of energy and matter exchange mechanisms and precision forest management using large-footprint spaceborne laser scanning data.