Conference Agenda

Precise data on ecosystem restoration projects can enhance scientific research on monitoring the long-term effectiveness of restoration efforts and the consecution of restoration objectives using remote sensing technologies. Through the combination of two FAO tools, (i) the Framework for Ecosystem Restoration Monitoring (FERM) that is used for compiling and publishing ecosystem restoration data and (ii) the System for Earth Observation Data Access, Processing and Analysis for Land Monitoring (SEPAL) that allows to produce sophisticated and relevant geospatial analyses we can monitor restoration actions on the ground.

Indicators and their corresponding metrics are the way to track the progress of restoration efforts. FERM provides the user the possibility to monitor ecosystem restoration through specific indicators. A good practice for restoration projects is having indicators measured on the ground by the project monitoring team. Furthermore, the use of earth observation technologies provides the possibility to determine scientific baselines as well as to monitor change over time through indicators, including after project completion.

With SEPAL, we will monitor two forest ecosystem restoration projects and indicators integrated into the FERM platform.

For monitoring forest restoration & agroforestry activities we will use SEPAL to create a yearly mosaic over the project location and calculate some indices (Normalized Difference Vegetation Index (NDVI), EVI) followed by a time series example of the project location using a built in CCDC algorithm. With the combination of all these approaches we will identify the gradual regrowth of vegetation in areas targeted for restoration.

For mangrove ecosystems, SEPAL will map aboveground biomass (AGB) using remote sensing techniques, by calculating the NDVI and Soil-Adjusted Vegetation Index (SAVI). This approach leverages vegetation indices as proxies to estimate AGB, providing essential insights into mangrove distribution, carbon storage potential, and ecosystem health.

The increasing frequency and intensity of droughts and heat waves driven by climate change have led to a significant increase in tree mortality worldwide. However, the lack of accurate and consistent data on the location, timing, species and structure of dead trees across vast geographical areas limits our understanding of climate-induced tree mortality. Furthermore, standing dead, dying, and habitat trees are crucial indicators of forest health and biodiversity but are often overlooked in existing forest resource mapping systems.

To address this, we present novel advancements in mapping individual tree mortality events using high-resolution (≤ 0.5 m) multi-temporal Earth Observation data, including both satellite and aerial imagery, combined with deep learning techniques. Our approach represents the first steps towards building an open large-scale database of individual tree mortality events across time. We have trained several U-Net-based deep learning models for detecting individual dead and dying trees from a wide array of imagery, enabling the creation of wall-to-wall datasets on tree mortality at national scales. We show results from the first nationwide individual tree mortality mapping, demonstrating the accuracy of sub-meter resolution satellite imagery in providing annual tree mortality data. We also discuss the challenges and limitations associated with detecting and characterizing detected dead trees across entire countries. We also show the accuracy of sub-meter resolution satellite imagery in providing annual tree mortality data using deep learning for several study areas.

We welcome scientists across the globe to contribute to creating a database on individual tree mortality events to support a wide range of tree mortality data needs in different scientific disciplines.

**"Integrating Indigenous Knowledge and Earth Observation for Carbon Monitoring in Amazon Reforestation"**

The Ecuadorian Amazon is at the forefront of reforestation efforts, driven by Indigenous communities working to restore and protect this critical ecosystem. Mario Vargas Shakaim, an Indigenous leader from the region, will present on the reforestation initiatives led by his community and the innovative approach of Project Shakaim. This project combines traditional ecological knowledge with Earth Observation (EO) data to quantify carbon sequestration in newly reforested areas.

Project Shakaim leverages satellite data alongside Indigenous land management practices to accurately measure the carbon stored in reforestation sites, providing crucial data for understanding carbon dynamics in tropical forests. By integrating these knowledge systems, the project enhances the precision of carbon monitoring while ensuring that local ecological insights guide reforestation efforts. This approach offers a replicable model for combining community-led initiatives with advanced monitoring technologies to address climate change.

This talk will illustrate how Indigenous perspectives enrich scientific approaches to ecosystem restoration and climate mitigation. Attendees will gain a deeper understanding of the role of Indigenous knowledge in improving carbon measurement accuracy and how such integrative approaches can inform global reforestation strategies.

The EU Nature Restoration Law came into force on August 18, 2024. With research project RestorEO we contribute to biodiversity and ecosystem restoration and conservation activities and reporting duties in Austria by developing and testing a wall-to-wall EO-based monitoring system for selected biodiversity indicators. We focus on three habitat areas (1) forests (2) cultural grassland, and (3) wetlands.

Forests are biologically diverse ecosystems that provide habitat for a multiplicity of plants, animals and micro-organism. For the forest use case, we develop methods that assess forest condition and detect standing and lying deadwood from both Sentinel-2 and LiDAR data. Additional product developments deal with forest fragmentation and connectivity based on GuidosToolbox by the EC Joint Research Center, and an estimation of organic carbon. The common forest bird index (Article 12(2)), an important indicator of biodiversity restoration, is closely linked to these parameters.

For grasslands, our developments focus on a Sentinel-2 based monitoring of the mowing intensity and yellowness of grassland plots. These are two parameters that indirectly reflect plant species diversity and that are used in butterfly habitat modeling. Better information on plant species diversity can serve as an indicator to detect changes related to the grassland butterfly index for areas where field surveys and butterfly counts are missing.

For wetlands, our remote sensing approaches address the detection of drainage channels and shrub encroachment in ecologicaly important moors and heathlands to support Austrian biodiversity monitoring tasks.

With this contribution, we want to present some of the RestorEO mapping results for the different habitats and biodiveristy indicators and discuss the potential and limitations for operational integration of these remote sensing based parameters in national activities and reporting for ecosystem and biodiversity restoration.

To strengthen and operationalise the relationship between condition variables and Ecosystem Services (ES) as defined by the System of Environmental Economic Accounting Ecosystem Accounting framework (SEEA EA), is essential to integrate condition variables into models of ES potential, which is the capacity of the ecosystem to provide the service. This study focuses on the relationship between the ES flood control and key condition variables of the urban ecosystem measured with satellite remote sensing data, these are: imperviousness and tree cover density, which are an input in the ES model of flood control. The data sources are Copernicus High Resolution Layers for the EU. The model was adjusted to include Tree Cover Density making it responsive to an indicator of the Nature Restoration Regulation (NRR) for the urban ecosystem.
Additionally, the model uses CLC+ classes which enabled the analysis of the role of urban green spaces on flood control potential, urban green spaces is also an indicator of the NRR.
This study demonstrates that changes in condition variables, such as a decrease in tree cover density, can significantly impact the ES potential. The decrease in the ES potential is translated into a decrease of the ecosystem service flow. In conclusion, this study operationalizes SEEA EA condition variables and ecosystem service accounts, demonstrating the linkages between them and their potential to support policy objectives, particularly the Nature Restoration Regulation.

Climate adaptation in cities occurs at several levels including the conservation and restoration of green spaces. To support the implementing and monitoring large scale greening projects, we propose a multisource strategy that can be adapted to a variety of input images and data sources (VHR, Sentinel-2, airborne, Lidar, tri-stereo) with the aim to characterize ecological corridors in detail. We apply a first pass of a SegNet like model trained to segment canopy surfaces in RGB images (a version of this model has been adapted to grayscale images and can thus be used to go back in time). This canopy detection can be coupled with a detection of isolated trees, the latter making it possible to obtain the finest trees thanks to a RetinaNet type model specified for the detection of small, isolated trees. Each of these elements can be combined with digital height models, obtained by lidar or stereoscopic reconstruction, to refine the accuracy of the typology by assigning height strata classes. Finally, the use of Sentinel 2 time series, at the scale of the objects detected, makes it possible to refine the typology with phenological considerations. The methods developed can be adapted to images with resolutions ranging from 5cm to 50cm, with great robustness and invariance to acquisition conditions.

The Amazon rainforest is thought of as an important global carbon sink, but changes in local climate, extreme events, and human disturbance often result in it becoming a source of atmospheric carbon. Understanding shifting patterns in tree mortality is crucial to determining the carbon budget of the Amazon, but little is known about the extent, rate, and causes of mortality of large canopy trees, which contain most of the forest carbon and are hypothesised to be most at risk with climate change.

To address this data gap, we developed an algorithm that accurately detects canopy tree mortality across the Amazon Basin using a time series of Planet NICFI data from 2018 to 2024 at 4.77m pixel spacing. We detect mortality events by identifying changes in trend over time in the multispectral reflectances caused by either a decrease of photosynthetic activity or shadowing from adjacent trees.The same principle also allows us to categorize mortality events into standing dead trees and broken/uprooted trees. The end result is monthly predictions of mortality events with a detection rate of 75%. The probability of detection increases with tree crown size, to above 90% when the mortality event is larger than 150 square meters, which makes the algorithm particularly well suited to study large tree mortality.

Early results show an increase in mortality across the Amazon Basin during the 2023-24 El Niño event and the potential of this method to study widespread effects of climatic changes at short temporal scales. Being able to detect large tree mortality fills an important gap in our knowledge of vegetation turnover and vulnerability, which underpins our understanding of tipping points in the Amazon and its resilience to climate change.