Uncertainty propagation in vegetation distribution models based on ensemble classifiers

Ensemble learning techniques are increasingly applied for species and vegetation distribution modelling, often resulting in more accurate predictions. At the same time, uncertainty assessment of distribution models is gaining attention. In this study, Random Forests, an ensemble learning technique, is selected for vegetation distribution modelling based on environmental variables. The impact of two important sources of uncertainty, that is the uncertainty on spatial interpolation of environmental variables and the uncertainty on species clustering into vegetation types, is quantified based on sequential Gaussian simulation and pseudo-randomization tests, respectively. An empirical assessment of the uncertainty propagation to the distribution modelling results indicated a gradual decrease in performance with increasing input uncertainty. The test set error ranged from 30.83% to 52.63% and from 30.83% to 83.62%, when the uncertainty ranges on spatial interpolation and on vegetation clustering, respectively, were fully covered. Shannon’s entropy, which is proposed as a measure for uncertainty of ensemble predictions, revealed a similar increasing trend in prediction uncertainty. The implications of these results in an empirical distribution modelling framework are further discussed with respect to monitoring setup, spatial interpolation and species clustering.     

我国的城市发展与大气污染

为了表明我国城市发展与大气污染的关系,以翔实的统计数据,采用两种分类的方法,研究和讨论了不同人口规模、不同功能类型的城市大气污染特点,并从防治大气污染的角度阐明了城市应定性质、定规模、定方向地协调发展。      

Spatial variability and temporal stability of throughfall deposition under beech (Fagus sylvatica L.) in relationship to canopy structure

Although the spatial variability of throughfall (TF) in forest ecosystems can have important ecological implications, little is known about the driving factors of within-stand TF variability, particularly in deciduous forests. While the spatial variability of TF water amount and H+ deposition under a dominant beech (Fagus sylvatica L.) tree was significantly higher in the leafed period than in the leafless period, the spatial TF deposition patterns of most major ions were similar in both periods. The semiannual TF depositions of all ions other than H+ were significantly positively correlated (r=0.68-0.90, p<0.05) with canopy structure above sample locations throughout the entire year. The amounts of TF water and H+ deposition during the leafed period were negatively correlated with branch cover. We conclude that the spatial heterogeneity of ion deposition under beech was significantly affected by leaves in the growing period and by branches in non-foliated conditions.     

Assessment of Temporal and Spatial Variation of Nitrate Removal in Riparian Zones

The aim of this study was to monitor long-term temporal and spatial groundwater NO₃^- removal efficiencies in different riparian zones via a limited number of sampling wells. Groundwater NO₃^- concentrations were measured fortnightly or monthly over a period of two years using transects of ground water sampling wells. Depending on the level of the NO₃^- load (up to 120mgNL^-1 at the input side of the riparian zone a distance of 10 to 30m was needed to remove NO₃^- from the groundwater below 11.3mgNL^-1. Considering all seasons, the mixed vegetation and grass riparian site succeeded to remove groundwater NO₃^- efficiently (92—100% within a distance of 30m. The forested riparian zone removed 72—90% of the total NO₃^- input within a distance of 30m. Evidence emerged that NO₃^- could also be removed actively at depths up to 2m, due to the presence of organically enriched layers of alluvial deposits or roots. Our four dimensional approach (three dimensional space and time), in combination with a limited number of sampling wells, was shown to be a useful monitoring tool to assess the variability of NO₃^- removal in riparian zones.