Refine
Has Fulltext
- no (3) (remove)
Document Type
- Article (3) (remove)
Language
- English (3)
Is part of the Bibliography
- yes (3)
Keywords
- Random Forest (3) (remove)
Institute
As the recent permafrost thawing of northern Asia proceeds due to anthropogenic climate change, precise and detailed palaeoecological records from past warm periods are essential to anticipate the extent of future permafrost variations. Here, based on the modern relationship between permafrost and vegetation (represented by pollen assemblages), we trained a Random Forest model using pollen and permafrost data and verified its reliability to reconstruct the history of permafrost in northern Asia during the Holocene. An early Holocene (12-8 cal ka BP) strong thawing trend, a middle-to-late Holocene (8-2 cal ka BP) relatively slow thawing trend, and a late Holocene freezing trend of permafrost in northern Asia are consistent with climatic proxies such as summer solar radiation and Northern Hemisphere temperature. The extensive distribution of permafrost in northern Asia inhibited the spread of evergreen coniferous trees during the early Holocene warming and might have decelerated the enhancement of the East Asian summer monsoon (EASM) by altering hydrological processes and albedo. Based on these findings, we suggest that studies of the EASM should consider more the state of permafrost and vegetation in northern Asia, which are often overlooked and may have a profound impact on climate change in this region.
Several thousands of moraine-dammed and supraglacial lakes spread over the Hindu Kush Himalayan (HKH) region, and some have grown rapidly in past decades due to glacier retreat. The sudden emptying of these lakes releases large volumes of water and sediment in destructive glacial lake outburst floods (GLOFs), one of the most publicised natural hazards to the rapidly growing Himalayan population. Despite the growing number and size of glacial lakes, the frequency of documented GLOFs is remarkably constant. We explore this possible reporting bias and offer a new processing chain for establishing a more complete Himalayan GLOF inventory. We make use of the full seasonal archive of Landsat images between 1988 and 2016, and track automatically where GLOFs left shrinking water bodies, and tails of sediment at high elevations. We trained a Random Forest classifier to generate fuzzy land cover maps for 2491 images, achieving overall accuracies of 91%. We developed a likelihood-based change point technique to estimate the timing of GLOFs at the pixel scale. Our method objectively detected ten out of eleven documented GLOFs, and another ten lakes that gave rise to previously unreported GLOFs. We thus nearly doubled the existing GLOF record for a study area covering similar to 10% of the HKH region. Remaining challenges for automatically detecting GLOFs include image insufficiently accurate co-registration, misclassifications in the land cover maps and image noise from clouds, shadows or ice. Yet our processing chain is robust and has the potential for being applied on the greater HKH and mountain ranges elsewhere, opening the door for objectively expanding the knowledge base on GLOF activity over the past three decades.
Overland flow is an important hydrological pathway in many forests of the humid tropics. Its generation is subject to topographic controls at differing spatial scales. Our objective was to identify such controls on the occurrence of overland flow in a lowland tropical rainforest. To this end, we installed 95 overland flow detectors (OFDs) in four nested subcatchments of the Lutzito catchment on Barro Colorado Island, Panama, and monitored the frequency of overland flow occurrence during 18 rainfall events at each OFD location temporal frequency. For each such location, we derived three non-digital terrain attributes and 17 digital ones, of which 15 were based on Digital Elevation Models (DEMs) of three different resolutions. These attributes then served as input into a Random Forest ensemble tree model to elucidate the importance and partial and joint dependencies of topographic controls for overland flow occurrence.
Lutzito features a high median temporal frequency in overland flow occurrence of 0.421 among OFD locations. However, spatial temporal frequencies of overland flow occurrence vary strongly among these locations and the subcatchments of Lutzito catchment. This variability is best explained by (1) microtopography, (2) coarse terrain sloping and (3) various measures of distance-to-channel, with the contribution of all other terrain attributes being small. Microtopographic features such as concentrated flowlines and wash areas produce highest temporal frequencies, whereas the occurrence of overland flow drops sharply for flow distances and terrain sloping beyond certain threshold values.
Our study contributes to understanding both the spatial controls on overland flow generation and the limitations of terrain attributes for the spatially explicit prediction of overland flow frequencies.