Effects of sample size on the accuracy of geomorphological models

Commonly, the most costly part of geomorphological distribution modelling studies is gathering the data. Thus, guidance for researchers concerning the quantity of field data needed would be extremely practical. This paper scrutinises the relationship between the sample size (the number of observations varied from 20 to 600) and the predictive ability of the generalized linear model (GLM), generalized additive model (GAM), generalized boosting method (GBM) and artificial neural network (ANN) in two data settings, i.e., independent and split-sample approaches. The study was performed using empirical data of periglacial processes from an area of 600 km² in northernmost Finland at grid resolutions of 1 ha (100 × 100 m) and 25 ha (500 × 500 m). A rather sharp increase in the predictive ability of the models was observed when the number of observations increased from 20 to 100, and the level of robust predictions was reached with 200 observations. The result indicates that no more than a few hundred observations are needed in geomorphological distribution modelling at a medium scale resolution (ca. 0.01–1 km²).     

Elementary forms for land surface segmentation: The theoretical basis of terrain analysis and geomorphological mapping

Land surface morphology is fundamental to geomorphological mapping and many GIS applications. Review and comparison of various approaches to segmentation of the land surface reveals common features, and permits development of a broad theoretical basis for segmentation and for characterization of segments and their boundaries. Within the context of defining landform units that maximise internal homogeneity and external differences, this paper introduces the concept of elementary forms (segments, units) defined by constant values of fundamental morphometric properties and limited by discontinuities of the properties. The basic system of form-defining properties represents altitude and its derivatives, constant values of which provide elementary forms with various types of homogeneity. Every geometric type of elementary form can be characterized by a defining function, which is a specific case of the general polynomial fitted function. Various types of boundary discontinuity and their connections and transformations into other types of morphological unit boundaries are analysed.The wealth of types of elementary forms and their boundaries is potentially unbounded and thus is sufficient to cover the real variety of landforms. Elementary forms in the basic set proposed here have clear potential for genetic and dynamic interpretation. A brief worked example documents the possibility of analytical computation of various models of ideal elementary forms for particular segments of landform. Ideal elementary forms can be considered as attractors, to which the affinity of surface segments can be measured by multivariate statistical methods. The use of the concept of elementary forms in landscape segmentation is promising and it could be adapted for elementary segmentation of various other spatial fields.     

A representativeness-directed approach to mitigate spatial bias in VGI for the predictive mapping of geographic phenomena

Volunteered geographic information (VGI) contains valuable field observations that represent the spatial distribution of geographic phenomena. As such, it has the potential to provide regularly updated low-cost field samples for predictively mapping the spatial variations of geographic phenomena. The predictive mapping of geographic phenomena often requires representative samples for high mapping accuracy, but samples consisting of VGI observations are often not representative as they concentrate on specific geographic areas (i.e. spatial bias) due to the opportunistic nature of voluntary observation efforts. In this article, we propose a representativeness-directed approach to mitigate spatial bias in VGI for predictive mapping. The proposed approach defines and quantifies sample representativeness by comparing the probability distributions of sample locations and the mapping area in the environmental covariate space. Spatial bias is mitigated by weighting the sample locations to maximize their representativeness. The approach is evaluated using species habit suitability mapping as a case study. The results show that the accuracy of predictive mapping using weighted sample locations is higher than using unweighted sample locations. A positive relationship between sample representativeness and mapping accuracy is also observed, suggesting that sample representativeness is a valid indicator of predictive mapping accuracy. This approach mitigates spatial bias in VGI to improve predictive mapping accuracy.     

Evaluation of approaches for forest cover estimation in the Pacific Northwest, USA, using remote sensing

The transformation of land cover, in particular coniferous forest, constitutes one of the most notable agents of regional-to-global-scale environmental change. Remote sensing provides an excellent opportunity for providing forest cover information at appropriate spatial and temporal scales. The optimal exploitation of remote sensing relies on the link between known forest cover and the remotely sensed dataset. This paper explores the accuracy of three methods – vegetation indices, regression analysis and neural networks – for estimating coniferous forest cover across the United States Pacific Northwest. All methods achieved a similar accuracy of forest cover estimation. However, in view of the benefits and limitations of each, the neural network approach is recommended for future consideration.     

Evaluation of vegetative criteria for near-surface groundwater detection using multispectral mapping and GIS techniques in semi-arid Botswana

The critical need to consider all options in the search for groundwater in semi-arid areas has promoted work on the possible association of near-surface groundwater and vegetation characteristics using a combination of remote-sensing data and geographic information systems (GIS) techniques. Two vegetative criteria (dense woody cover and abundance of deep-rooting species) are identified as being indicative of near-surface groundwater, and their spatial distribution is tested against the location of aquifers in southeast Botswana. Vegetative criteria classes were combined in a GIS environment with the distribution of geomorphic units and bedrock geology to determine the degree of coincidence with assumed or known aquifers. Results indicate that the distribution of dense woody vegetation as mapped from Thematic Mapper imagery has some potential in identifying especially surficial but also bedrock near-surface groundwater sources in mostly naturally vegetated semi-arid areas. Dense woody cover classes tend to select aquifers in topographically higher areas while classes comprising some deep-rooting species tend to select low-lying aquifers such as those occurring in fossil valleys. Deep-rooting species, however, are less successful as a vegetative criterion. Although various technical refinements are suggested, this work shows that vegetative criteria mapping can however be used in conjunction with conventional geological/geophysical techniques to enhance the prospects for groundwater location in relatively undisturbed semi-arid areas.