Upscaling drought disturbance monitoring in the Amazon forests by combining terrestrial laser scanning and field spectroscopy

With increasing frequency and intensity of extreme drought events in Amazon forests due to climate change, understanding their impact on the ecosystem is crucial. One critical challenge lies in how drought disturbance will alter the 3D forest structure and composition locally, and how we can monitor these forest changes globally. Addressing this challenge, we present new unique data for the Caxiuanã throughfall exclusion field experiment in eastern Amazonia - the longest running drought experiment in the tropics. Our data consists of two complementary sources: terrestrial laser scanning (TLS) and field spectroscopy. TLS is increasingly recognized as a pivotal technology for forest monitoring, offering highly detailed measurements of 3D vegetation structure. On the other hand, spectroscopy can provide the spectral signature of the main biophysical components of the forest scene. We show that the fusion of both enables the creation of realistic virtual forests, which can serve as input for (optical) radiative transfer models (RTM), allowing for the upscaling of forest monitoring. The study site consists of two 1ha research plots established in 2001 in the Caxiuanã National Forest, a tropical lowland rainforest in the state of Para, northeast of Brazil (1∘43′ S, 51∘27′ W). The first plot is a control plot without any manipulation. The second plot is the throughfall exclusion (TFE) experiment, where plastic panels at 1-2m above the ground cover the entirety of the area, thereby excluding 50% of canopy rain throughfall. For both plots, we collected TLS and field spectroscopy measurements in October 2023 (dry season). Plots were scanned using a RIEGL VZ400i in a 10x10 grid with both an upright and tilt scan at each position (600khz). Individual scans are subsequently co-registered in RISCAN-PRO to form a single complete point cloud. Reflectance spectra (350 nm – 2500 nm) of leaves, bark and soil were collected using an ASD FieldSpec 4 Hi-Res radiospectrometer. For 51 individual trees, nine top of canopy leaves were sampled and measured (2x per leave), and 4 bark reflectance measurements at ca. 2m above ground level were taken. For 29 individuals, leaf measurements were repeated pre-dawn and at midday to check for any diurnal spectral variation. Soil measurements were taken at patches lit by direct sunlight. Spectra were averaged to derive a single reflectance curve for each individual and component. Figure 1 shows the steps for conversion of a segmented TLS point cloud into a virtual forest for radiative transfer modelling. First, the different components (soil, stem and branches, leaves) in the 3D point cloud are semantically segmented and the individual trees processed into some volumetric representation. The different components can then be spectrally parameterized, allowing for the creation of a structurally and radiometrically realistic digital replica of the forest: a ‘digital twin forest’. These digital twins can be used as input for (optical) radiative transfer modelling. RTMs enhance our ability to monitor and understand the coupling between emitted, reflected or scattered electromagnetic (EM) waves and a 3D scene, and therefore are a powerful tool to establish the physically-based link between in situ structure and how it is monitored with satellite remote sensing. Explicit digital forest twins will be constructed according to the abovementioned processing pipeline for both the Caxiuana control and drought plot. This will then allow for running RTM models such as DART (https://dart.omp.eu/#/) and Eradiate (https://www.eradiate.eu/site/) to simulate and compare remote sensing products, and increase our understanding of the impact of drought on 3D forest structure as seen from space. A next step is to move from ‘static’ 3D digital twins to ‘dynamic’ 4D digital twins, to incorporate the temporal dimension. For Caxiuana, both plots have been scanned before in the years 2015 and 2018. Co-registering these data opens avenues for unexplored research into the impact of drought on Amazon forests. Furthermore, ongoing and future work within the context of the SPACETWIN project (https://spacetwin.ugent.be/) focuses on enhancing our understanding on forest disturbances and recovery from space. The project will focus on the impact of drought, but also fire logging and insect disturbances across a range of tropical and temperate forest ecosystems. SPACETWIN aims to integrate realistic TLS-derived digital twins of forest structure with spectroscopic measurements as RTM input to build an emulator for optical, lidar and radar satellite data that will allow for near real-time interpretation of disturbances.