The ablation zones of debris-covered glaciers in Himalaya exhibit heterogeneous processes and melt patterns. Although sub-debris melt is measured at ablation stakes, the high variability of debris thickness necessitates distributed melt measurements at the glacier scale. Focusing on Annapurna III Glacier, we used uncrewed aerial system (UAS) photogrammetry to estimate total volume loss and slope-perpendicular glacier melt between May and November 2019 using flow-corrected point clouds. Results indicated the average elevation change was −1.10 ± 0.19 m, while the mean melt was −0.87 m w.e., equating to a mean melt rate of −0.47 cm w.e. d⁻¹. However, the spatial pattern was highly variable due to complex local processes necessitating future study over short intervals. The evaluation of specific areas showed the interplay of debris thickness variability, subseasonal debris redistribution, supraglacial channel reconfiguration and the imprint of relict ice cliffs in leading to contemporary melt rates. Ice cliffs had higher melt distances (mean −3.9 ± 0.19 m) compared to non-cliff areas (mean −0.75 ± 0.19 m) and were the predominant control on the spatial patterns of seasonal melt rates. Crucially, the definition of ice cliff areas from thinning data has a profound impact on derived melt rates and melt enhancement. Our study demonstrates the possibility and utility of deriving fully-distributed slope-perpendicular melt measurements.

Understanding ecological patterns and response to climate change requires unbiased data on species distribution. This can be challenging, especially in biodiverse but extreme environments like the Himalaya. This study presents the results of the first ever application of Unmanned Aerial Systems (UAS) imagery for species-level mapping of vegetation in the Himalaya following a hierarchical Geographic Object Based Image Analysis (GEOBIA) method. The first level of classification separated green vegetated objects from the rest with overall accuracy of 95%. At the second level, seven cover types were identified (including four woody vegetation species). For this, the suitability of various spectral, shape and textural features were tested for classifying them using an ensemble decision tree algorithm. Spectral features alone yielded ~70% accuracy (kappa 0.66) whereas adding textural and shape features marginally improved the accuracy (73%) but at the cost of a substantial increase in processing time. Contrast in plant morphological traits was the key to distinguishing nearby stands as different species. Hence, broad-leaved versus fine needle leaved vegetation were mapped more accurately than structurally similar classes such as Rhododendron anthopogon versus non-photosynthetic vegetation. Results highlight the potential and limitations of the suggested UAS-GEOBIA approach for detailed mapping of plant communities and suggests future research directions.

If he were living today, Alexander von Humboldt would be using current technology to evaluate change in the Andes. Inspired by von Humboldt’s scientific legacy and the 2019 celebrations of his influence, we utilize a Moderate Resolution Imaging Spectroradiometer (MODIS)time-series vegetation index to ask questions of landscape change. Specifically, we use an 18-year record of Normalized Difference Vegetation Index (NDVI) data as a proxy to evaluate landscape change in Peru, which is well known for its high biological and ecological diversity. Continent-level evaluations of Latin America have shown sites with a positive trend in NDVI, or “greening” and “browning”, a negative trend in NDVI that suggests biophysical or human-caused reductions in vegetation. Our overall hypothesis was that the major biomes in Peru would show similar NDVI change patterns. To test our expectations, we analyzed the NDVI time-series with Thiel-Sen regression and evaluated Peru overall, by protected area status, by biome, and by biome and elevation. Across Peru overall, there was a general greening trend. By protected area status, surprisingly, the majority of greening occurred outside protected areas. The trends were different by biome, but there were hotspots of greening in the Amazon, Andean Highlands, and Drylands where greening dominated. In the Tropical Subtropical Dry Broadleaf Forest biome, greening and browning signals were mixed. Greening trends varied across the elevation gradient, switching from greening, to browning, and then back to greening as elevation increased. By biome and elevation, the results were variable. We further explored biome-specific drivers of greening and browning drawing on high-resolution imagery, the literature, and field expertise, much as we imagine von Humboldt might have approached similar questions of landscape dynamism.

Aquatic invasive plants (AIPs) are a global threat to local biodiversity due to their rapid adaptation to the new environments. Lythrum salicaria, commonly known as purple loosestrife, is a predominant AIP in the upper Midwestern region of the United States and has been designated as a deadly threat to the wetlands of this region. Accurate estimation of its current extent is a top priority, but regular monitoring is limited due to cost-, labor-, and time-intensive field surveys. Therefore, the goal of the present study is to accurately detect purple loosestrife from very high-resolution UAV imagery using deep neural network-based models. As a case study, this study implemented U-Net and LinkNet models with ResNet-152 encoder in the wetlands of the upper Mississippi River situated in La Crosse County, Wisconsin. The results showed that both models produced 88–94% training accuracy and performed better in landscapes that were occupied by smaller, disaggregated, and more equitably distributed purple loosestrife. Furthermore, the study adopted a transfer learning approach to implement a trained purple loosestrife model of the first study site and implemented it for the second study site. The results showed that the pre-trained model implementation generated better accuracy in less than half the time of the original model. Therefore, the transfer learning approach, if adapted efficiently, can be highly beneficial for continuous monitoring of purple loosestrife and strategic planning for application of direct biocontrol measures.

The relative importance of various drivers of fire regimes in savanna ecosystems can be location-specific. We utilised satellite-derived time-series burned area (2001–13) to examine how spatiotemporal variations in burned area and fire frequency were determined by rainfall, vegetation morphology and land use in semiarid savanna. Mean precipitation of the rainy season (Nov–Apr) had a strong and positive relationship with burned area in the following dry season (variance explained 63%), with the relationship being strongest inside protected areas (variance explained 73%). Burned area and fire frequency were higher in vegetation types with higher herbaceous cover, indicating a causal link between herbaceous load and fire. Among land use, fire frequency was highest in protected areas and lowest in farms and ranches. Spatial models (generalised linear models with Poisson and negative binomial distribution) accounting for spatial autocorrelation showed that land-use classes and vegetation types together explained approximately half of the deviance in null model (48%). Existence of fences and boreholes resulted in finer-scale spatial differences in fire frequency. There was minimal dependence of vegetation types on land-use classes in determining fire frequency (interaction between the two predictors was minimal). These results have significant implications for understanding drivers of fire activity in savanna ecosystems.