Nesha, Karimon; Herold, Martin; Sy, Veronique De; de Bruin, Sytze; Araza, Arnan; Málaga, Natalia; Gamarra, Javier G. P.; Hergoualc'h, Kristell; Pekkarinen, Anssi; Ramirez, Carla; Morales-Hidalgo, David; Tavani, Rebecca
Exploring characteristics of national forest inventories for integration with global space-based forest biomass data Journal Article
In: Science of The Total Environment, vol. 850, iss. 157788, 2022.
Abstract | Links | BibTeX | Tags: Aboveground biomass, CCI biomass, Creative Commons, Forest Resources Assessment, National Forest Inventory, NFI and space-based data integration, NFI plot design
@article{@article{NESHA2022157788,
title = {Exploring characteristics of national forest inventories for integration with global space-based forest biomass data},
author = {Karimon Nesha and Martin Herold and Veronique {De Sy} and Sytze {de Bruin} and Arnan Araza and Natalia Málaga and Javier G.P. Gamarra and Kristell Hergoualc'h and Anssi Pekkarinen and Carla Ramirez and David Morales-Hidalgo and Rebecca Tavani},
url = {https://www.sciencedirect.com/science/article/pii/S0048969722048872?via%3Dihub},
doi = {https://doi.org/10.1016/j.scitotenv.2022.157788},
year = {2022},
date = {2022-12-01},
journal = {Science of The Total Environment},
volume = {850},
issue = {157788},
abstract = {National forest inventories (NFIs) are a reliable source for national forest measurements. However, they are usually not developed for linking with remotely sensed (RS) biomass information. There are increasing needs and opportunities to facilitate this link towards better global and national biomass estimation. Thus, it is important to study and understand NFI characteristics relating to their integration with space-based products; in particular for the tropics where NFIs are quite recent, less frequent, and partially incomplete in several countries. Here, we (1) assessed NFIs in terms of their availability, temporal distribution, and extent in 236 countries from FAO's Global Forest Resources Assessment (FRA) 2020; (2) compared national forest biomass estimates in 2018 from FRA and global space-based Climate Change Initiative (CCI) product in 182 countries considering NFI availability and temporality; and (3) analyzed the latest NFI design characteristics in 46 tropical countries relating to their integration with space-based biomass datasets. We observed significant NFI availability globally and multiple NFIs were mostly found in temperate and boreal countries while most of the single NFI countries (94 %) were in the tropics. The latest NFIs were more recent in the tropics and many countries (35) implemented NFIs from 2016 onwards. The increasing availability and update of NFIs create new opportunities for integration with space-based data at the national level. This is supported by the agreement we found between country biomass estimates for 2018 from FRA and CCI product, with a significantly higher correlation in countries with recent NFIs. We observed that NFI designs varied greatly in tropical countries. For example, the size of the plots ranged from 0.01 to 1 ha and more than three-quarters of the countries had smaller plots of ≤0.25 ha. The existing NFI designs could pose specific challenges for statistical integration with RS data in the tropics. Future NFI and space-based efforts should aim towards a more integrated approach taking advantage of both data streams to improve national estimates and help future data harmonization efforts. Regular NFI efforts can be expanded with the inclusion of some super-site plots to enhance data integration with currently available space-based applications. Issues related to cost implications versus improvements in the accuracy, timeliness, and sustainability of national forest biomass estimation should be further explored.},
keywords = {Aboveground biomass, CCI biomass, Creative Commons, Forest Resources Assessment, National Forest Inventory, NFI and space-based data integration, NFI plot design},
pubstate = {published},
tppubtype = {article}
}
Milenkovic, Milutin
Big EO Data Processing Approaches for Monitoring Tropical Forest Disturbance with Sentinel-1 Data Presentation
15.11.2022.
Abstract | BibTeX | Tags: Earth Observations, Presentation, Sentinel-1, tropical forests
@misc{nokey,
title = {Big EO Data Processing Approaches for Monitoring Tropical Forest Disturbance with Sentinel-1 Data},
author = {Milutin Milenkovic},
year = {2022},
date = {2022-11-15},
abstract = {An increase in the disturbance of tropical forests, mainly caused by human-driven deforestation, drouths, and forest fires, has been constantly reported over the last decades. Those stressors are pushing tropical ecosystems to their limits, and it is urgently needed to understand how far the ecosystems are from their point of no return. Novel global and free Earth Observation (EO) data, such as Sentinel-1 satellite radar images from the EU’s Copernicus programme, have high spatial (20 m) and temporal (6 or more days) resolution, providing an excellent possibility for monitoring tropical ecosystem change. However, those terabytes of data also require novel processing ways. In this talk, I will present three different big data approaches utilized to detect Sentinel-1 signal disturbances and recoveries over tropical forests in the Amazon and Congo basins. The presentation will give a user perspective and experience on the Sentinel-1 processing on local servers, the ESA openEO platform, and a national computing infrastructure used within the C-SCALE project. A novel EU-funded project on big EO data processing, the open Earth monitor cyberinfrastructure, will also be introduced.
Event: Enhancing Satellite Imagery and Geospatial Information Capabilities in Support of Nuclear Safeguards
Place: UN, IAEA Headquarters, Vienna, Austria
Date: 15–17 November 2022},
keywords = {Earth Observations, Presentation, Sentinel-1, tropical forests},
pubstate = {published},
tppubtype = {presentation}
}
Event: Enhancing Satellite Imagery and Geospatial Information Capabilities in Support of Nuclear Safeguards
Place: UN, IAEA Headquarters, Vienna, Austria
Date: 15–17 November 2022
Mîrț, Andrei; Reiche, Johannes; Verbesselt, Jan; Herold, Martin
A Downsampling Method Addressing the Modifiable Areal Unit Problem in Remote Sensing Journal Article
In: Remote Sensing, vol. 14, iss. 22, no. 5538, 2022.
Abstract | Links | BibTeX | Tags: big data, downsampling, modifiable areal unit problem, multscale, remote sensing
@article{nokey,
title = {A Downsampling Method Addressing the Modifiable Areal Unit Problem in Remote Sensing},
author = {Andrei Mîrț and Johannes Reiche and Jan Verbesselt and Martin Herold},
url = {https://www.mdpi.com/2072-4292/14/21/5538},
doi = {https://doi.org/10.3390/rs14215538},
year = {2022},
date = {2022-11-03},
journal = {Remote Sensing},
volume = {14},
number = {5538},
issue = {22},
abstract = {Handling multiple scales efficiently is one avenue for processing big remote sensing imagery data. Unfortunately, imagery is also affected by the infamous modifiable areal unit problem, which creates unpredictable errors at different scales. We developed a downsampling method that attempts to keep the data distribution in a downsampled image constant, reducing the modifiable areal unit problem. We tested our method against classic downsampling methods (mean, central pixel selection, random) under a range of typical remote sensing scenarios. Under our experimental conditions, our downsampling method consistently outperformed the classical downsampling methods within a 95% confidence level. The downsampling method can be used in most typical situations where downsampling is needed, but it is likely to shine when used as a pyramid building policy in geocomputing platforms, such as Google Earth Engine.},
keywords = {big data, downsampling, modifiable areal unit problem, multscale, remote sensing},
pubstate = {published},
tppubtype = {article}
}
Masolele, Robert N.; Sy, Veronique De; Marcos, Diego; Verbesselt, Jan; Gieseke, Fabian; Mulatu, Kalkidan Ayele; Moges, Yitebitu; Sebrala, Heiru; Martius, Christopher; Herold, Martin
Using high-resolution imagery and deep learning to classify land-use following deforestation: a case study in Ethiopia Journal Article
In: GIScience & Remote Sensing, vol. 59, no. 1, pp. 1446-1472, 2022.
Abstract | Links | BibTeX | Tags: Attention U-Net, deep learning, deforestation drivers, Planet-NFCI, remote sensing
@article{@article{doi:10.1080/15481603.2022.2115619,
title = {Using high-resolution imagery and deep learning to classify land-use following deforestation: a case study in Ethiopia},
author = {Robert N. Masolele and Veronique De Sy and Diego Marcos and Jan Verbesselt and Fabian Gieseke and Kalkidan Ayele Mulatu and Yitebitu Moges and Heiru Sebrala and Christopher Martius and Martin Herold},
url = {https://doi.org/10.1080/15481603.2022.2115619
},
doi = {https://doi.org/10.1080/15481603.2022.2115619},
year = {2022},
date = {2022-09-07},
urldate = {2022-09-07},
journal = {GIScience & Remote Sensing},
volume = {59},
number = {1},
pages = {1446-1472},
abstract = {National-scale assessments of post-deforestation land-use are crucial for decreasing deforestation and forest degradation-related emissions. In this research, we assess the potential of different satellite data modalities (single-date, multi-date, multi-resolution, and an ensemble of multi-sensor images) for classifying land-use following deforestation in Ethiopia using the U-Net deep neural network architecture enhanced with attention. We performed the analysis on satellite image data retrieved across Ethiopia from freely available Landsat-8, Sentinel-2 and Planet-NICFI satellite data. The experiments aimed at an analysis of (a) single-date images from individual sensors to account for the differences in spatial resolution between image sensors in detecting land-uses, (b) ensembles of multiple images from different sensors (Planet-NICFI/Sentinel-2/Landsat-8) with different spatial resolutions, (c) the use of multi-date data to account for the contribution of temporal information in detecting land-uses, and, finally, (d) the identification of regional differences in terms of land-use following deforestation in Ethiopia. We hypothesize that choosing the right satellite imagery (sensor) type is crucial for the task. Based on a comprehensive visually interpreted reference dataset of 11 types of post-deforestation land-uses, we find that either detailed spatial patterns (single-date Planet-NICFI) or detailed temporal patterns (multi-date Sentinel-2, Landsat-8) are required for identifying land-use following deforestation, while medium-resolution single-date imagery is not sufficient to achieve high classification accuracy. We also find that adding soft-attention to the standard U-Net improved the classification accuracy, especially for small-scale land-uses. The models and products presented in this work can be used as a powerful data resource for governmental and forest monitoring agencies to design and monitor deforestation mitigation measures and data-driven land-use policy.},
keywords = {Attention U-Net, deep learning, deforestation drivers, Planet-NFCI, remote sensing},
pubstate = {published},
tppubtype = {article}
}