Slagter, Bart; Reiche, Johannes; Marcos, Diego; Mullissa, Adugna; Lossou, Etse; Peña-Claros, Marielos; Herold, Martin
Monitoring direct drivers of small-scale tropical forest disturbance in near real-time with Sentinel-1 and -2 data Journal Article
In: Remote Sensing of Environment, vol. 295, pp. 113655, 2023, ISSN: 0034-4257.
Abstract | Links | BibTeX | Tags: deep learning, Deforestation, Driver attribution, Forest degradation, Near real-time monitoring, Small-scale forest disturbance, Smallholder agriculture, tropical forests
@article{nokey,
title = {Monitoring direct drivers of small-scale tropical forest disturbance in near real-time with Sentinel-1 and -2 data},
author = {Bart Slagter and Johannes Reiche and Diego Marcos and Adugna Mullissa and Etse Lossou and Marielos Peña-Claros and Martin Herold},
url = {https://www.sciencedirect.com/science/article/pii/S0034425723002067},
doi = {10.1016/j.rse.2023.113655},
issn = {0034-4257},
year = {2023},
date = {2023-09-01},
journal = {Remote Sensing of Environment},
volume = {295},
pages = {113655},
abstract = {Advancements in satellite-based forest monitoring increasingly enable the near real-time detection of small-scale tropical forest disturbances. However, there is an urgent need to enhance such monitoring methods with automated direct driver attributions to detected disturbances. This would provide important additional information to make forest disturbance alerts more actionable and useful for uptake by different stakeholders. In this study, we demonstrate spatially explicit and near real-time methods to monitor direct drivers of small-scale tropical forest disturbance across a range of tropical forest conditions in Suriname, the Republic of the Congo and the Democratic Republic of the Congo. We trained a convolutional neural network with Sentinel-1 and Sentinel-2 data to continuously classify newly detected RAdar for Detecting Deforestation (RADD) alerts as smallholder agriculture, road development, selective logging, mining or other. Different monitoring scenarios were evaluated based on varying sensor combinations, post-disturbance time periods and confidence levels. In general, the use of Sentinel-2 data was found to be most accurate for driver classifications, especially with data composited over a period of 4 to 6 months after the disturbance detection. Sentinel-1 data showed to be valuable for more rapid classifications of specific drivers, especially in areas with persistent cloud cover. Throughout all monitoring scenarios, smallholder agriculture was classified most accurately, while road development, selective logging and mining were more challenging to distinguish. An accuracy assessment throughout the full extent of our study regions revealed a Macro-F1 score of 0.861 and an Overall Accuracy of 0.897 for the best performing model, based on the use of 6-month post-disturbance Sentinel-2 composites. Finally, we addressed three specific monitoring use cases that relate to rapid law enforcement against illegal activities, ecological impact assessments and timely carbon emission reporting, by optimizing the trade-off in classification timeliness and confidence to reach required accuracies. Our findings demonstrate the strong capacities of high spatiotemporal resolution satellite data for monitoring direct drivers of small-scale forest disturbance, considering different user interests. The produced forest disturbance driver maps can be accessed via: https://bartslagter94.users.earthengine.app/view/forest-disturbance-drivers.},
keywords = {deep learning, Deforestation, Driver attribution, Forest degradation, Near real-time monitoring, Small-scale forest disturbance, Smallholder agriculture, tropical forests},
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}
}