Figure 38 - uploaded by Jan Hjort
Content may be subject to copyright.
Ca. 20-40 cm high active earth hummocks (69°34'10"N, 26°14'55"E / 370 m a.s.l. / 9 th of July 2002). Dwarf birches (Betula nana) are common in the hummock fi elds.
Source publication
Determination of the environmental factors controlling earth surface processes and landform patterns is one of the central themes in physical geography. However, the identification of the main drivers of the geomorphological phenomena is often challenging. Novel spatial analysis and modelling methods could provide new insights into the process-envi...
Similar publications
Fluvial processes play a crucial role in landscape development in periglacial regions. Geomorphological studies have yielded contrasting inferences about the importance of various environmental factors to drainage density. In this paper, 12 explanatory variables, based on relief, vegetation, soil, and rock type, were used to explain the controlling...
Citations
... However, the allocation of active or relict status is often only a first approximation (cf. Hjort 2006), because recognising the extent of modern activity in the face of environmental change is problematic. Some stratigraphic contexts enable the reconstruction of the changing frequency of events since at least the early Holocene. ...
Periglaciation in Scandinavia is reviewed with a focus on active and relict landforms, frost processes, permafrost distribution and landform age in plateau, steepland, low-gradient, and glacier-foreland landscapes. Scandinavian periglacial landscapes are conceptualized as complex palimpsests in a continually changing environment. Distinctive aspects in the context of Europe are emphasized and some important general periglacial research problems are highlighted. Many types of landforms are relict, at least in part, despite extensive areas where permafrost exists today, and the even more widespread occurrence of seasonal frost. Elements of the plateau landscapes and the coastal strandflat are of pre-Pleistocene age, although they owe much of their present character to frost weathering processes. Some landform types, such as blockfields, tors and cryoplanation terraces, originated before the Last (Weichselian) Glaciation and were preserved beneath Pleistocene cold-based ice sheets. Many large rock-slope failures, rock glaciers, large-scale sorted circles, raised coastal rock platforms and inland parabolic dunes are Late Weichselian or relict paraglacial features. Currently active landforms and processes, such as debris flows, solifluction, slope wash, palsas, earth hummocks, small rock-slope failures, snow-avalanche landforms and dune re-activation, continue to experience significant changes in activity in response to Holocene and anthropocene climatic variations and resulting shifts in the altitudinal limits of permafrost and seasonal frost. Altitudinal zonation of periglacial landforms is problematic because of difficulties in separating active from relict forms. Active landforms on glacier forelands, such as sorted patterned ground, may provide modern analogues for interpreting larger-scale relict features. Scandinavian periglacial landscapes have played and will likely continue to occupy an indispensable and influential position in the development of knowledge and understanding of the periglacial landscapes of Europe.
... It allows the use of both dependent and independent variables as continuous data type as well as categorical. Another advantage of GLM is that the dependent variable can have a different distribution than normal, by using various function related to the type of distribution (Guisan et al., 2002;Hjort, 2006). The general formula of the GLM is (Atkinson et al., 1998): Y = a + bX1 + cX2 + … + mXn, where Y is the dependent variable, X1, X2, Xn are the independent variables, and a, b, c, m are the coefficients. ...
... For the categorical data type, the ANOVA test was used. If two predictors were correlated, only one of them, the one with a higher geomorphological significance, was considered here (Hjort, 2006). Also, the variables that show the same characteristics (e.g. ...
The paper aims to determine to what extent the size of the rock glaciers (RG) in the Southern Carpathians (Romania) is influenced by their contributing area (CA) parameters. Simple linear regression (LR) and generalized linear models (GLM) were used to meet this goal, considering as independent variables the main morphometric characteristics of the contributing area. The LR coefficients revealed that the most influential variables were the width (R2=0.57) and the size of the CA (R²=0.51). Based on the best GLM results the size of the rock glaciers can be statistically explained quite well (R²=0.58) by a combination of three variables: CA length, CA width, and the minimum altitude of the CA. Rock glaciers are thus complex landforms resulting from a combination of many variables (climatic, topographic and geologic) including contributing area parameters. Both LR and GLM analysis revealed that the size of the rock glaciers can only be partly explained by the characteristics of the CA. The study revealed that GLM are powerful analytical tools which give reasonable results when analysing the role of rock glaciers developmental controls. © 2017 University of Craiova, Faculty of Social Sciences, Department of Geography. All rights reserved.
... This study explores which environmental factors control recent process activity of cryoturbation and solifluction features (activity modelling, i.e. determination of the differences between active and inactive sites) in an extensively studied subarctic area in Finnish Lapland (Hjort, 2006;Hjort et al., 2007). It investigates geospatial data using three different statistical techniques, namely, the generalised additive model (GAM), the boosted regression tree (BRT) and hierarchical partitioning (HP). ...
... The topography of the area is characterised by gently sloping hills and treeless fells (elevation range = 195-641 m asl) ( Figure 2). Till is the most common surficial ground material deposit, and peat, silt, sand and gravel deposits are common in valleys between the fells (Hjort, 2006). The climate is subarctic with a mean annual air temperature (MAAT) of -1.7°C (-14.8°C in January, 13.0°C in July) and mean annual precipitation of 414 mm (1971-2000Drebs et al., 2002). ...
... Alpine vegetation covers two-thirds of the area and is dominated by Empetrum and B. nana types of heaths. Further details about the study area are given by Hjort (2006). ...
Environmental factors that affect the activity-inactivity variation of periglacial features may differ from those factors that control the distributional patterns of active features. To explore this potential difference, a statistically based modelling approach and comprehensive data on active and inactive cryoturbation and solifluction features from a subarctic area of Finnish Lapland are investigated at a landscape scale. In the cryoturbation modelling, vegetation abundance is the most important environmental variable explaining both the activity-inactivity variation and the distribution of active sites. The next most important variables are soil moisture and (micro)climatological conditions in the activity modelling, and slope angle and ground material in the distribution modelling. For solifluction, the key variables determining the activity-inactivity variation are mean annual air temperature and mean maximum snow depth, whereas vegetation abundance and slope angle control the distribution of active sites. Comparison between the environmental conditions of active and inactive periglacial features may provide new insights into activity-environment relationships, which in turn are valuable when the effects of climate change on periglacial processes are explored. Copyright © 2014 John Wiley & Sons, Ltd.
... The area is characterized mostly by open uplands with forests of subalpine mountain birch Betula pubescens ssp. czerepanovii, shallow peat supporting mires, and gently sloping glacially sculptured fells (Hjort, 2006). Geologically, the study area belongs to the Pre-Cambrian c. 1.9 billion-year-old granulite complex (Meriläinen, 1976). ...
... Etzelmüller et al., 2001;Ridefelt et al., 2010). The occurrence of solifluction was determined using landforms considered as clear indicators of solifluction operation over a considerable period of time and/or at present (Tolgensbakk and Sollid, 1983;Flakstad et al., 1985;Sollid, 1987, 1988;Tolgensbakk et al., 2000;Hjort, 2006). ...
... In the Paistunturit study area, the most common landform types are sorted solifluction sheets and streams and non-sorted terraces (Hjort, 2006). Lobe-like landforms are relatively rare in the area. ...
... GDMs are empirical models relating field observations to explanatory variables (i.e., predictor variables and environmental variables), based on statistically or theoretically derived response surfaces. Geomorphic data can be simple presence, presence-absence, or abundance (e.g., cover and activity) observations on landforms, processes, or feature assemblages (Guzzetti et al., 1999;Hjort, 2006). Environmental variables can be acquired from various sources, commonly from GI and RS data (Moore et al., 1991;Etzelmü ller et al., 2001). ...
... Moreover, they can be used to model feature assemblages (e.g., geodiversity and geomorphic process units) and to identify the shapes of the geomorphic process-environment relationships. In general, GDMs provide a mathematical basis for the interpretation of relationships between response and explanatory variables and Figure 1 Important predictors (i.e., explanatory variables) for the distribution of cryoturbation features were drawn from a set of potentially important predictors (Hjort, 2006). The analyses were conducted in northern Finland at a mesoscale resolution (grid cell size ¼ 500 Â 500 m). ...
... However, there is a paucity of examples in geomorphology. To our knowledge, the only examples are in modeling the distributions of periglacial landforms and processes (Hjort andMarmion, 2008, 2009;Marmion et al., 2008Marmion et al., , 2009Luoto et al., 2010). ...
Statistically-based geomorphic distribution modeling (GDM) has become popular among geoscientists as an efficient approach for analysis and prediction. Here, we provide a cross-section of the concept of GDM. First, we introduce the main steps in the GDM process. Second, we provide an overview of statistical techniques, which have shown to be promising in geomorphic modeling. Third, we draw attention to important advantages and pitfalls of GDM. Finally, we highlight some future challenges in the application of the GDM approach. The general aim is to aid the geomorphic community to gain novel insights into Earth surface process-environment relationships using the concept of GDM.
... Gruber and Hoelzle, 2001;Janke, 2005;Ridefelt et al., 2008), as well Geomorphology 155-156 (2012) as periglacial landforms and processes (e.g. Etzelmüller et al., 2001;Luoto and Hjort, 2004;Hjort, 2006). This paper focuses on the spatial analysis of the distribution of hummocky terrains on Deception Island in order to define the factors controlling their distribution. ...
... The DEM was used to derive the predictive (independent) variables of elevation, aspect, slope, wetness index and total curvature. Elevation and slope are parameters usually used in geomorphological modelling in periglacial environments (Hjort, 2006). Elevation controls air temperature, snow and solar radiation, while slope also controls geomorphic and hydrological processes (Selby, 1993). ...
... Aspect controls potential radiation and snow distribution (Barry, 1992), while in many areas it also plays a role in active layer thickness. Total slope curvature has been suggested by several authors to be the best parameter to calculate terrain convexities or concavities, which relate to accumulation vs. erosion of sediment, and also to accumulation of water, snow, as well as of cold air (Gallant and Wilson, 2000;Hjort, 2006). ...
The hummocky terrains of Deception Island (Antarctic Peninsula) are continuous surfaces with decimetre to metre wide and decimetre depth bumps located mainly in the lower section of sloping lapilli and scoria terrains. A detailed study site between Cerro Caliente and Crater Lake was selected for the detailed mapping of hummocky terrains and for modelling their spatial distribution according to controlling geographical factors. A model of the susceptibility of occurrence of the hummocky terrains was created using the information value method, together with five independent variables: elevation, slope, global summer radiation, total curvature and lithology. Success and prediction rate curves were used for model validation and the Area Under the Curve index was used to quantify the levels of performance and prediction. The results were of high quality with a success rate of 88% and a prediction rate of 78%. The classes of the independent variables with more relevance in the occurrence of hummocky terrains were: elevation between 20–30 m and 60–70 m; concave or rectilinear/flat areas; slopes between 8 and 12º; tuff cones and maar deposits and global summer radiation between 1.8 and 2.0 TJm− 2. The good quality of the modelling results supports its use for assessing the future potential for formation of new hummocky terrain areas, or even to estimate the spatial distribution of buried ice within the permafrost environment of Deception Island.
... Valleys between the fells with silty soil are the sites of peat, sand, and gravel deposits. The silty sediments deposited at the bottoms of temporary, ice-dammed lakes during the deglaciation period are covered predominantly by organic material (Hjort, 2006). ...
... czerepanovii (Orlova) Hämet-Ahti] and alpine heaths, and small scattered scots pine (Pinus sylvestris) forests characterize the lower elevations. The forests cover about 37% of the study area (Hjort, 2006). Mires, which cover 10% of the study area, belong to the palsa and subalpine types (Luoto and Seppälä, 2002). ...
... The periglacial features were mapped in the fi eld using black-and-white aerial photographs (1 : 31 000) and a global positioning system (GPS) device (Garmin eTrex personal navigator with an accuracy of about 10 m; Garmin, Olathe, KS, USA) (Hjort, 2006). The activity of the features was defi ned based on the observations of lichen cover on the stones and blocks, rock weathering, frost heaving as well as of general soil and vegetation disturbance (Washburn, 1979;Hjort and Luoto, 2009). ...
Robust models of geomorphic process–environment relationships are important to advance theoretical knowledge of geomorphic systems. Here, we examined a generalized additive modeling (GAM) based approach to provide new theoretical insights into process–environment relationships. More precisely, we (i) simulated the shapes of the relationships between geomorphic processes and environmental variables based on GAM and (ii) compared the shapes of the simulated response curves to (a) the hypothetical curves based on theory and (b) the response curves produced by generalized linear modeling (GLM). Hitherto, GLM was the most common technique to study the relationships between environmental gradients and geomorphic processes. The study is based on empirical cryoturbation and solifluction data and environmental variables from subarctic Finland. Our results showed that non-linear relationships were more common than linear responses and the simulated GAM based response curves coincided more closely with the hypothetical response curves than did the response curves derived from GLM. The simulated response curves showed high potential in geomorphic hypothesis testing. In conclusion, our findings indicate that careful examination of the response curves may provide new insights into theoretical debates in the earth sciences. Copyright © 2010 John Wiley and Sons, Ltd.
... Additionally, the investigator focused only on the geophysical variables to guarantee independence of the botanical and geomorphological data. The cover of the active geomorphological processes in the sampling sites (cryoturbation, fluvial activity, nivation, solifluction, weathering;Hjort, 2006) that showed current activity was estimated as a percentage following the methodology of Hjort & Luoto (2009). This geomorphological disturbance measure thus combines many physical disturbance factors potentially causing biomass loss, excluding biotic disturbances caused by grazing animals. ...
Aim We test how productivity, disturbance rate, plant functional composition and species richness gradients control changes in the composition of high-latitude vegetation during recent climatic warming.
Location Northern Fennoscandia, Europe.
Methods We resampled tree line ecotone vegetation sites sampled 26 years earlier. To quantify compositional changes, we used generalized linear models to test relationships between compositional changes and environmental gradients.
Results Compositional changes in species abundances are positively related to the normalized difference vegetation index (NDVI)-based estimate of productivity gradient and to geomorphological disturbance. Competitive species in fertile sites show the greatest changes in abundance, opposed to negligible changes in infertile sites. Change in species richness is negatively related to initial richness, whereas geomorphological disturbance has positive effects on change in richness. Few lowland species have moved towards higher elevations.
Main conclusions The sensitivity of vegetation to climate change depends on a complex interplay between productivity, physical and biotic disturbances, plant functional composition and richness. Our results suggest that vegetation on productive sites, such as herb-rich deciduous forests at low altitudes, is more sensitive to climate warming than alpine tundra vegetation where grazing may have strong buffering effects. Geomorphological disturbance promotes vegetation change under climatic warming, whereas high diversity has a stabilizing effect.
... Mires belong to the palsa and subalpine mire types (Luoto and Seppä lä , 2002). A more detailed description of the study region can be obtained from Hjort (2006). ...
... The landforms were mapped and converted to gridbased modelling data in a four-step process (Hjort, 2006): ...
Currently, statistical models, which relate the spatial distribution of
landforms and processes with environmental conditions, are widely used
in geomorphological mapping. However, because models are the result of
both simplification and imperfect representation of reality, model
predictions may be tainted by errors. In this study, we evaluated the
ability of four statistical consensus methods (namely Median(All),
Mean(All), Best and Weighted Average) to improve the accuracy of
predictions based on models. The spatial distributions of twelve
geomorphological features were recorded at a 500 x 500 m resolution in
subarctic Finland. Nine topographical, surficial ground material and
land cover variables were used to predict the distribution of
geomorphological features using eight widely used "single-modelling"
techniques [including Generalized Linear Models (GLMs), Generalized
Boosting Method (GBM), Generalized Additive Models (GAMs),
Classification Tree Analysis (CTA), Artificial Neural Network (ANN),
Multivariate Adaptive Regression Splines (MARS), Mixture Discriminant
Analysis (MDA) and Random Forest (RF)]. The outputs of the single-models
(i.e. probability values of occurrence) were combined using consensus
algorithms. The accuracy of the predictions was evaluated on spatially
independent data by the area under the curve (AUC) of a receiver
operating characteristic (ROC) plot. The mean AUC values of the eight
single-models varied between 0.711 (CTA) and 0.755 GAM), whereas mean
AUC values of consensus methods ranged from 0.752 [Median(All)] to 0.784
[Mean(All)]. Consensus methods based on the average function were the
most efficient to improve the accuracy of the predictions. For eleven
geomorphological features out of twelve, Weighted Average and Mean(All)
provided more accurate predictions than those based on the best
single-modelling technique (GAM). The results of this study suggest that
consensus methods are powerful to increase the accuracy of predictive
models. These methods should be used more often in applied and
theoretical geomorphology.
... Important environmental determinants for the studied patterned ground features are slope gradient, soil moisture, grain size distribution and vegetation cover (Washburn, 1979;Matsuoka, 2001;Luoto and Hjort, 2004). Sorted patterned ground is clearly defined on the upper fell areas with relatively high slope angle, typically on till soils, whereas non-sorted patterned ground is more common on valleys with higher soil moisture and fine-scale concave topography (Hjort, 2006). The spatial pattern of sorted patterned ground reflects the distribution of the upper fell areas, while the non-sorted patterned ground is more dispersed in the study area, indicating the structure of the valleys and depressions of the area (Luoto and Hjort, 2005). ...
... The periglacial features used in the analysis were mapped in situ during the spatially comprehensive field surveys by utilizing pre-mapping results, black-and-white aerial photographs (1:31,000), and a Global Positioning System (GPS) device (Garmin eTrex personal navigator with an accuracy of about 10 m; Garmin, Olathe, KS, USA) during the summer 2002 (Hjort, 2006). All surveys were performed by the same geomorphologist (JH) to exclude variation resulting from differences between observers. ...
... All surveys were performed by the same geomorphologist (JH) to exclude variation resulting from differences between observers. Because of the extensive study area (100 km 2 ), study aims (focus on the distribution of active features, not on activity itself) and modelling resolution (1-ha), we used a visual method to gain an estimate of the activity (active/inactive) of the mapped features (Hjort, 2006). ...
Maps of earth surface processes and the potential distribution of landforms make an important contribution to theoretical and applied geomorphology. Because decision making often depends on information based on spatial models, there is a great need to develop methodology to evaluate the spatial uncertainty resulting from those models. In this study we developed a new method to produce maps of the uncertainty of predictions provided by ten state-of-the-art modelling techniques for sorted (SP) and non-sorted (NSP) patterned ground in subarctic Finland at a 1.0-ha resolution. Six uncertainty classes represent the modelling agreement between the different modelling techniques. The resulting uncertainty maps reflect the reliability of the estimates for the studied periglacial landforms in the modelled area. Our results showed a significant negative correlation between the degree of uncertainty and the accuracy of the modelling techniques. On average, when all ten models agreed, the mean area under the curve (AUC) values were 0.904 (NSP) and 0.896 (SP), these values decreased to 0.416 (NSP) and 0.518 (SP), respectively, when only five models agreed. Mapping of the uncertainty of predictions in geomorphology can help scientists to improve the reliability of their data and modelling results. The predictive maps can be interpreted simultaneously with the uncertainty information, improving understanding of the potential pitfalls of the modelling.
