This study evaluates and compares the performance of four machine learning classifiers—support vector machine (SVM), normal Bayes (NB), classification and regression tree (CART) and K nearest neighbor (KNN)—to classify very high resolution images, using an object-based classification procedure. In particular, we investigated how tuning parameters affect the classification accuracy with different training sample sizes. We found that: (1) SVM and NB were superior to CART and KNN, and both could achieve high classification accuracy (>90%); (2) the setting of tuning parameters greatly affected classification accuracy, particularly for the most commonly-used SVM classifier; the optimal values of tuning parameters might vary slightly with the size of training samples; (3) the size of training sample also greatly affected the classification accuracy, when the size of training sample was less than 125. Increasing the size of training samples generally led to the increase of classification accuracies for all four classifiers. In addition, NB and KNN were more sensitive to the sample sizes. This research provides insights into the selection of classifiers and the size of training samples. It also highlights the importance of the appropriate setting of tuning parameters for different machine learning classifiers and provides useful information for optimizing these parameters.

Landscape changes associated with urbanization can lead to many serious ecological and environmental problems. Quantifying the vertical structure of the urban landscape and its change is important to understand its social and ecological impacts, but previous studies mainly focus on urban horizontal expansion and its impacts on land cover/land use change. This papers focuses on the residential landscape to investigate how the vertical dimension of the urban landscape (i.e., building height) change through time, and how such change is related to changes in the horizontal dimension of the landscape, using Beijing, the capital of China, as a case study. We quantified the expansion of the residential neighborhoods from 1949 to 2009, and changes in vegetation coverage, building density, and building height within these neighborhoods, using 1 m spatial resolution imagery. One-way ANOVA and correlation analysis were used to examine the relationships of building height to vegetation coverage and building density. We found: (1) The residential areas expanded rapidly and were dominated by outward growth, with much less within-city infilling. The growth rate varied greatly through time, first increasing from 1949–2004 and then decreasing from 2005–2009. The expansion direction of newly built residential neighborhoods shifted from west to north in a clockwise direction. (2) The vertical structure of residential neighborhoods changed with time and varied in space, forming a “low-high” pattern from urban central areas to the urban edges within the 5th ring road of Beijing. (3) The residential neighborhoods built in different time periods had significant differences in vegetation coverage, building density, and building height. The residential neighborhoods built in more recent years tended to have taller buildings, lower building density and lower vegetation coverage.

The thermal environment in residential areas is directly related to the living quality of residents. Therefore, it is important to understand thermal heterogeneity and ways to regulate temperature in residential neighborhoods. We investigated the spatial heterogeneity and temporal dynamics of air temperatures in 20 residential neighborhoods within the 5th ring road of Beijing, China. We further explored how the variations in air temperature were related to the patterns of the surrounding greenspace at different scales. We found that: (1) large air temperature differences existed among residential neighborhoods, with hourly maximum differences in air temperature reaching 5.30 °C on hot summer days; (2) not only the percentage but also the spatial configuration (e.g., edge density) of greenspace affected the local air temperature; and (3) the effects of spatial greenspace patterns on air temperature were scale dependent and varied by season. For example, increasing the proportion of greenspace in surrounding areas within a 100-m radius and increasing the edge density within radii from 500 to 1000 m could lower air temperatures in summer but not affect air temperatures in winter. In addition, decreasing the edge density of greenspaces within a 100-m radius of the surrounding areas would lead to an increase in air temperature in winter but not affect the air temperature in summer. These results extend our understanding of thermal environments and their relationships with greenspace patterns at the microscale (i.e., residential neighborhoods). They also provide useful information for urban planners to optimize greenspace patterns under better thermal conditions at the neighborhood scale.