Fig 2 - uploaded by Teresa De Marzo
Content may be subject to copyright.
Three examples of land-use practices leading to forest disturbances, as seen in very-high-resolution satellite images in Google Earth (left), as well as in time series of Tasselled Cap Wetness (TCW) medoids calculated over May-July (right; original annual values in green, values fitted with LandTrendr in yellow). (A) logging for charcoal production (note that the charcoal kiln is also visible); (B) selective logging of valuable tree species; (C) fire. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Source publication
Forest loss in the tropics affects large areas, but whereas full forest conversions are routinely assessed, forest degradation patters remain often unclear. This is particularly so for the world’s tropical dry forests, where remote sensing of forest disturbances is challenging due to high canopy complexity, strong phenology and climate variability,...
Context in source publication
Context 1
... evaluated how different parameters changed the trajectory fitting for these exemplary regions, and chose the parameter combination that across samples visually resulted in the best time series segmentations (Table 1). Thus, this segmentation procedure resulted in a distinct set of disturbance metrics for each of the nine annual time series (in Fig. 2 examples of the TCW segmentation). We selected the following metrics describing the segment with the highest magnitude (in the direction of forest loss): (1) the index value for the year before the beginning of the disturbance (hereafter: prevalue), (2) the delta of the values of the index between the beginning and end of the ...
Similar publications
As the largest source of uncertainty in carbon cycle studies, accurate quantification of gross primary productivity (GPP) is critical for the global carbon budget in the context of global climate change. Numerous vegetation indices (VIs) based on satellite data have participated in the construction of GPP models. However, the relative performance o...
Citations
... Integrating livestock data, such as the vaccination data we used, with satellite time series (Marzo et al 2021) could provide a more comprehensive understanding of livestock grazing impact on woodlands (Nanni et al 2024, Peri et al 2024. Similarly, better data on woodland smallholders could help to consider these actors and their relation to woodlands in sustainability planning (Del Giorgio et al 2021, Levers et al 2021, Pratzer et al 2024. ...
Livestock grazing is a key land use globally, with major environmental impacts, yet the spatial footprint of grazing remains elusive, particularly at broad scales. Here, we combine livestock system indicators based on remote sensing and livestock vaccination data with a biophysical grass growth model to assess forage production, livestock carrying capacity, and grazing pressure on rangelands in the South American Dry Chaco. Specifically, we assess how considering different livestock systems (e.g. fattening in confinement, grazing with supplementary feeding, woodland grazing) changes estimations of grazing pressure. Our results highlight an average carrying capacity of 0.48 animal units equivalents (AUEs) per hectare for the Chaco (0.72 for pastures, 0.43 for natural grasslands, 0.37 for woodlands). Regional livestock requirements ranged between 0.02–6.43 AUE ha⁻¹, with cattle dominating livestock requirements (91.6% of total AUE). Considering livestock systems with different production intensities markedly altered the rangeland carrying capacity and degradation estimations. For example, considering confinements and supplementary feeding drastically reduced the pasture area with potential overgrazing, from about 58 000 km² to <19 000 km² (i.e. 13.5% vs 5.7% of the total rangeland area). Conversely, considering the typically unaccounted-for cattle of woodland smallholders markedly increased the potentially degraded woodland area, from 3.2% (∼1000 km²) to 12.1% (3700 km²) of the total woodland area. Our work shows how ignoring production intensity can bias grazing pressure estimations and, therefore, conclusions about rangeland degradation connected to livestock production. Mapping indicators characterizing the intensity of livestock systems thus provide opportunities to understand better grazing impacts and guide efforts towards more sustainable livestock production.
... Recent advancements in satellite imaging, algorithms, and computing now enable highresolution, large-scale mapping of forest disturbances [8]. While most studies focus on boreal, temperate, and tropical wet forests, the mapping of forest alterations in subtropical semi-arid regions has only recently gained attention [11,12]. To accurately assess forest characteristics in bio-climatic regions with strong seasonality, RS techniques can be combined with approaches that capture spatial and temporal vegetation patterns [13]. ...
... Previous studies in the Chaco region have examined the effects of anthropogenic and natural disturbances on plant community phenology using RS data [11,40,41]. Most of these studies have focused on the Normalized Difference Vegetation Index (NDVI), ...
Anthropogenic alteration of tropical and subtropical forests is a major driver of biodiversity loss; notably, the Chaco Forest, which is the largest dry forest in the Americas, is among the most impacted regions. Sustainable forest management, a key objective of the UN’s 15th Sustainable Development Goal (SDG), underscores the need for advanced monitoring tools. This study integrates Sentinel-2 remote sensing (RS) spectral indices with field data to analyze forests under varying management regimes and levels of alteration in a representative area of the Chaco region (Chancaní Provincial Reserve and surrounding areas of the West Arid Chaco). Forest structure types and conservation levels were linked to monthly spectral index behavior using linear mixed models. Spectral indices such as the BI (Brightness Index), NDWIGao (Normalized Difference Water Index), and MCARISent (Modified Chlorophyll Absorption in Reflectance Index) effectively differentiated forest stands by conservation status and structural alteration. This combined RS and field data approach proved highly effective for detecting and characterizing forests with diverse conservation and sustainability conditions. The methodology demonstrates significant potential as a reliable RS-based tool for monitoring forest health and supporting progress toward SDG targets, particularly in regions like the Chaco Forest, which face extensive anthropogenic pressures.
... Climatic conditions are linked to stand dieback, and remote sensing data can be used to assess the physiological state of plants and their changes, with drought being the most important factor affecting stand decline (Huang et al., 2019). This implies the potential for mapping analyses with climate data, which was applied by De Marzo et al. (2021) to produce a multi-temporal disturbance map with an OA of 79 %, where significant increases in dry years were observed. The use of well-performing algorithms is crucial to achieve high accuracy. ...
Climate change is increasing the frequency of extreme events, including those in forests. One of the major drivers of forest change in Europe is the bark beetle, which causes large-scale annual changes in spruce forest areas. Mountain forests are particularly vulnerable as changes occur rapidly and require long-term monitoring of ongoing ecological changes. For this purpose, a 10-year time series of Sentinel-2 optical satellite data fused with Sentinel-1 radar and topographic derivatives was applied to the natural forests of the Tatra Mountains in Central Europe. Based on machine learning algorithms and iterative methods, overall classification accuracies of 0.94–0.96 and snags with an F1-score of 0.81–0.98 were achieved. The highest spruce mortality rate was observed in 2018, with extensive snag areas persisting until 2024. This study revealed that smaller infestation patches (< 0.1 ha) consistently dominated the landscape, peaking in 2018, whereas larger patches (> 0.5 ha) showed a declining trend, particularly after 2020. The variable importance analysis revealed that topographic factors are critical for predicting forest disturbance patterns. Elevation emerged as the most significant predictor with a Mean Decrease in Accuracy ranging from 95 to 150, followed by slope and aspect. Snag occurrence was strongly influenced by elevation, ranging from 700 to 1700 m a.s.l., with the median elevation increasing from 1150 m in 2015 to 1400 m in 2024. The slope also played an important role, with the median slopes for snag occurrences ranging from 15° to 25°, indicating a tendency for mortality on moderate inclines, although mortality on steeper slopes (up to 50°) was occasionally observed, particularly in 2017 and 2023. Regarding the slope orientation, the southeastern and eastern aspects consistently experienced higher proportions of spruce mortality (particularly between 2017 and 2021). A strong correlation between spruce mortality and temperature-related variables was identified, particularly degree days above 5 °C and 8.3 °C during key months (April, June, and July). Median yearly air temperature showed a correlation, whereas precipitation-related variables, including the Standardised Precipitation Evapotranspiration Index (SPEI), exhibited negative correlations, particularly the SPEI 01 median. These findings improve the understanding of long-term forest changes caused by disturbances and provide key insights for the data-driven management of protected forests in a changing climate.
... However, whether the seasonal and spectral indicators chosen in this study work better in other locations requires additional investigation. A growing number of studies have shown that the transferability of a single-detection algorithm is unstable, and it is still a challenge to improve this [45][46][47]. ...
... It should be noted that our research focuses on the extraction of most disturbance events from long-term Landsat imagery. However, due to image spatial and temporal resolution limitations, scattered small disturbance events, such as selective cutting and thin-ning in forest tending projects, may be missed, and estimates of disturbance area may be conservative [46]. Furthermore, knowing the motivations behind forest disturbance events and the conditions of forest restoration following disturbance would aid in understanding forest change processes [56]. ...
Cold-temperate forests (CTFs) are not only an important source of wood but also provide significant carbon storage in China. However, under the increasing pressure of human activities and climate change, CTFs are experiencing severe disturbances, such as logging, fires, and pest infestations, leading to evident degradation trends. Though these disturbances impact both regional and global carbon budgets and their assessments, the disturbance patterns in CTFs in northern China remain poorly understood. In this paper, the Genhe forest area, which is a typical CTF region located in the Inner Mongolia Autonomous Region, Northeast China (with an area of about 2.001 × 10⁴ km²), was selected as the study area. Based on Landsat historical archived data on the Google Earth Engine (GEE) platform, we used the continuous change detection and classification (CCDC) algorithm and considered seasonal features to detect forest disturbances over nearly 30 years. First, we created six inter-annual time series seasonal vegetation index datasets to map forest coverage using the maximum between-class variance algorithm (OTSU). Second, we used the CCDC algorithm to extract disturbance information. Finally, by using the ECMWF climate reanalysis dataset, MODIS C6, the snow phenology dataset, and forestry department records, we evaluated how disturbances relate to climate and human activities. The results showed that the disturbance map generated using summer (June–August) imagery and the enhanced vegetation index (EVI) had the highest overall accuracy (88%). Forests have been disturbed to the extent of 12.65% (2137.31 km²) over the last 30 years, and the disturbed area generally showed a trend toward reduction, especially after commercial logging activities were banned in 2015. However, there was an unusual increase in the number of disturbed areas in 2002 and 2003 due to large fires. The monitoring of potential widespread forest disturbance due to extreme drought and fire events in the context of climate change should be strengthened in the future, and preventive and salvage measures should be taken in a timely manner. Our results demonstrate that CTF disturbance can be robustly mapped by using the CCDC algorithm based on Landsat time series seasonal imagery in areas with complex meteorological conditions and spatial heterogeneity, which is essential for understanding forest change processes.
... Despite recent efforts to control land-use change in the Gran Chaco (Piquer-Rodríguez et al. 2015; Morea 2017), deforestation has continued as well as the transition to intensive types of production like soybean monoculture and feedlots (Baumann et al. 2017;De Marzo et al. 2021). Nothing suggests that this trend will change (Redclift and Sage 1998;Hubacek et al. 2017). ...
Communicating climate change projections to diverse stakeholders and addressing their concerns is crucial for fostering effective climate adaptation. This paper explores the use of storyline projections as an intermediate technology that bridges the gap between climate science and local knowledge in the Pilcomayo basin. Through fieldwork and interviews with different stakeholders, key environmental concerns influenced by climate change were identified. Traditional approaches to produce regional climate information based on projections often lack relevance to local communities and fail to address their concerns explicitly. By means of storylines approach to evaluate climate projections and by differentiating between upper and middle-lower basin regions and focusing on dry (winter) and rainy (summer) seasons, three qualitatively different storylines of plausible precipitation and temperature changes were identified and related to the main potential risks. By integrating these climate results with local knowledge, a summary of the social and environmental impacts related to each storyline was produced, resulting in three narrated plausible scenarios for future environmental change. The analysis revealed that climate change significantly influences existing issues and activities in the region. Projected trends indicate a shift towards warmer and drier conditions, with uncertainties mainly surrounding summer rainfall, which impacts the probability of increased flooding and river course changes, two of the most concerning issues in the region. These findings serve as a foundation for problem-specific investigations and contribute to informed decision-making for regional climate adaptation. Finally, we highlight the importance of considering local concerns when developing climate change projections and adaptation strategies.
... As the most powerful cloud computing platform (Azedou et al., 2022;Singh & Pandey, 2021), GEE offers the advantage to use temporal segmentation models that group observations into larger representative time periods or "segments", allowing for more effective detection of forest changes, thus providing an opportunity for large-scale application of vegetation change detection (Long et al., 2021). As an accurate long-term change detection model Marzo et al., 2021), LT aims to provide information on three elements: magnitude of change, duration of change and the year in which the disturbance or land cover recovery occurs . The findings of this work will prove the effectiveness of the creation of protected areas in terms of reducing the degradation of natural resources and for ensuring sustainable use of biodiversity. ...
Arid regions worldwide are facing environmental deterioration due to the climate change impacts. Assessing the spatial and temporal patterns of vegetation change and their relation with human activities is essential to address the challenges posed by this climate emergency. In this research, cloud computing techniques based on Google Earth Engine (GEE) allowed to identify the dynamics of vegetation change in Talassemtane National Park (TNP) using spectral-temporal segmentation algorithm LandTrendr (LT) for the period 1990–2020. NDVI was chosen as an appropriate index to detect vegetation trends, and yearly reflectance values were computed after cloud and shadow masking. Results showed over the past 30 years that the TNP has undergone a vegetation change of 16,329 ha where 61.02% corresponds to vegetation disturbance and 39.58% corresponds to vegetation gain. The main forested areas of the TNP showed significantly higher rates of vegetation increase compared to areas close to villages and cultivation areas with a respective percentage of 21.4% and 53.9% for vegetation disturbance versus 52% and 36.2% for vegetation gain. Computed years of change showed that the history of vegetation patterns has witnessed two main periods: a period that ends around 2007 when vegetation loss rates exceed vegetation gain rates and a second period beginning in 2008 (four years after the creation of the TNP) where vegetation gain exceeded vegetation loss. Hence, the recovery of vegetation within the protected areas is successful, which shows the effectiveness of these biodiversity conservation strategies.
... This makes the landscape prone to fires. Although fires are more frequent in highland ecosystems (upper watershed), extensive and severe fires may occur in the western plains (e.g., ≈100,000 ha in 1994) (Alinari et al. 2015;De Marzo et al. 2021). When the intensity, frequency, and/or extent increases, fire becomes a disturbance factor that triggers ecosystem degradation, for example, the erosion of the biophysical legacy such as soil loss and local species extinction (Kitzberger et al. 2016). ...
Agroforestry landscapes provide a variety of ecosystem goods and services at both the farm and landscape levels. They also host thousands of rural people whose livelihoods depend on the forest. Forest sustainable management is needed for farmer’s development. This is complex because it implies the integration of biological and socio-productive diversity with spatiotemporal dynamic interventions. In this chapter, we propose to reduce the vulnerability of agroforestry systems to climate change through resilience management of social-ecological systems (SES) at the landscape scale. Specifically, we examine key properties of farm-level SES components, exemplify how they collectively interconnect at the landscape scale, and analyze the benefits of resolving social-ecological conflicts at the landscape scale. We include a case study documenting adaptation measures to climate change for rural families whose livelihoods depend on the forest. These innovations included improved rainwater harvesting and conservation, resource use efficiency and soil conservation, agroecological diversification, and socioeconomic organization. The promotion of the adaptation of rural families to global change allows families to remain inhabiting their lands. Rural emigration is associated with high values of unsatisfied basic needs. The innovations proposed in this chapter are indirectly associated with global change mitigation (fixing and/or maintaining a high amount of carbon in the soil and in agroforestry biomass). We conclude that we must consider the adaptation and mitigation capacity of socio-ecosystems at the farm and landscape scale to find solutions to the challenges of global change, namely, anthropogenic pressure and climate change. This reinforces the socio-ecological resilience of the entire forest landscape by maintaining ecosystem services (support and regulation services) and improving rural and urban population livelihoods.
... Algorithm ensembles are built by aggregating results from multiple change detection algorithms that are executed in parallel (Cohen et al. 2020;Healey et al. 2018;Hislop et al. 2019). Conversely, multispectral ensembles Marzo et al. 2021;Schultz et al. 2016) involve the analysis of several bands through parallel runs of a univariate algorithm, which outputs are then dissolved into a single result using a supervised classifier, e.g. random forest. ...
Time series analysis of medium-resolution multispectral satellite imagery is critical to investigate forest disturbance dynamics at the landscape scale. In particular, the spatial, temporal, and radiometric consistency of Landsat time series data provides unprecedented insight into past disturbances that occurred over the last four decades. Several Landsat time series-based algorithms have been developed to automate the detection of forest disturbances. However, automated detection of non-stand-replacing disturbances based on Landsat time series remains a challenging task due to the difficulty of effectively separating them from spectral noise. Here, we present the High-dimensional detection of Landscape Dynamics (HILANDYN) algorithm, which exploits spatial and spectral information provided by Landsat time series to detect forest disturbance dynamics retrospectively. A novel and unsupervised procedure for changepoint detection in high-dimensional time series allows HILANDYN to perform the temporal segmentation of inter-annual time series into linear trends. The algorithm embeds a noise filter to remove spurious changepoints caused by residual spectral noise in the time series. We tested HILANDYN to detect disturbances that occurred in the forests of the European Alps over a period of 39 years, i.e. between 1984 and 2022, and evaluated its accuracy using a validation dataset of 3000 plots randomly located inside and outside the disturbed patches. We compared HILANDYN with the Bayesian Estimator of Abrupt change, Seasonality, and Trend (BEAST), which is a well-established and high-performing time series-based algorithm for changepoint detection. The quantitative results highlighted that the number of bands, i.e. original Landsat bands and spectral indices, included in the high-dimensional time series and the threshold controlling the significance of changepoints strongly influenced the user’s accuracy (UA). Conversely, changes in the combinations of bands primarily affected the producer’s accuracy (PA). HILANDYN achieved an F1 score of 0.801, which increased to 0.833 when we activated the noise filter, allowing the algorithm to balance UA (83.1%) and PA (83.5%). The qualitative results showed that disturbed forest patches detected by HILANDYN were characterized by a high spatio-temporal consistency, regardless of the disturbance severity. Furthermore, our algorithm was able to detect forest patches associated with secondary disturbances, such as salvage logging, that occur in close succession with respect to the primary event. The comparison with BEAST evidenced a similar sensitivity of the algorithms to non-stand-replacing events, as both achieved comparable PA. However, BEAST struggled to balance UA and PA when using a single parameter set, achieving a maximum F1 score of 0.717. Moreover, the computational efficiency of BEAST in processing high-dimensional time series was very limited due to its univariate nature based on the Bayesian approach. The adaptability of HILANDYN to detect a wide range of disturbance severities using a single parameter set and its computational efficiency in handling high-dimensional time series promotes its scalability to large study areas characterized by heterogeneous ecological conditions.
... Finally, integrated and synthesized evidence is depicted in balloon/triangle plots and impact-influence diagrams, where dissimilar stakeholders' vulnerabilities are visualized. 2019, de la Sancha et al., 2021), caused by multiple interacting drivers (De Marzo et al., 2021). Second, stakeholders are diverse and heterogeneous (Baldi et al., 2015;Mastrangelo et al., 2019b), owing to a process of modern commodity frontier expansion over indigenous and peasant territories (Aguiar et al., 2022). ...
... However, the introduction of the free and open data policy for Landsat in 2008 sparked methodological advances for burned area mapping and monitoring using medium spatial resolution sensors. This has led to a growing interest in time-series analysis methods that utilize large stacks of images (Chuvieco et al., 2019;De Marzo et al., 2021). ...
