Feasibility of urban bird occurrence and nest amount evaluation by the street-view image virtual survey.

Bird observation mainly relies on field surveys, which are time-consuming and laborious. In this study, we explored using street-view images in the virtual survey of urban birds and nests. Using the coastal city of Qingdao as the study area, 47 201 seamless spherical photos at 2741 sites were collected using the Baidu street-view (BSV) map. Single-rater-all photo checks and seven-rater-metapopulation checks were used to find inter-rater repeatability, the best viewing layer for BSV collection, and possible environments affecting the results. We also collected community science data for comparison. The BSV time machine was used to assess the temporal dynamics. Kappa square test, generalized linear model, redundancy ordination and ArcMap were used in the analysis. Different rater repeatability was 79.1% in nest evaluations and 46.9% in bird occurrence. A re-check of the different-rating photos can increase them to 92% and 70%. Seven-rater statistics showed that more than 5% sampling ratio could produce a non-significant different bird and nest percentage of the whole data, and the higher sampling ratio could reduce the variation. The middle-viewing layer survey alone could produce 93% precision of the nest checks by saving 2/3 of the time used; in birds, selecting middle and upper-view photos could find 97% of bird occurrences. In the spatial distribution, the nest's hotspot areas from this method were much greater than the community science bird-watching sites. The BSV time machine made it possible to re-check nests in the same sites but challenging the re-check of bird occurrences. The nests and birds can be observed more in the leafless season, on wide, traffic-dense coastal streets with complex vertical structures of trees, and in the gaps of tall buildings dominated by road forests. Our results indicate that BSV photos could be used to virtually evaluate bird occurrence and nests from their numbers, spatial distribution and temporal dynamics. This method provides a pre-experimental and informative supplement to large-scale bird occurrence and nest abundance surveys in urban environments.

Geo-climates and street developments shape urban tree characteristics: A street-view inventory analysis of over 200,000 trees of 11 metropolises in China.

Street trees play an important role in the city, but large-scale, multi-city inventory data are very limited to date, which can help to define geo-climatic and social development influences on urban forest characteristics. In this paper we speculate that at national level, geocliamtes and street development shape the different street tree characteristics, and large scale street View images (SVIs)-measurements favor the identification of factors responsible for the street tree variations in China. By collecting urban trees from 11 metropolises through SVIs method, an inventory of urban trees in China, including 201,942 trees at 9807 sites, was obtained from a latitude gradient from tropical 18oN to cold-temperate 45oN. Individual tree size-related growth status, tree-shrub-herb-related vertical structure, tree species identity, and street condition and street development (total 20 social development parameters) in the inventory is recorded. We analyzed trends and factors influencing street trees characteristics through latitudinal variation, distribution, linear regression, redundancy (RDA) ordination, and inter-city comparisons. The results showed that 1) with latitude increased, DBH and CPS linearly decreased, together with more highly dense forests (>100 trees/100 m street segment) observed. Latitude independence was in TH and forest vertical structural complexity. 2) All tree size data were in the log-normal distribution pattern when the two-parameter model was used and was best fitted by the Johnson distribution pattern when the >2-parameter model was used. 3) Tree growth status showed strong latitude dependency (R2 > 0.4, p < 0.05), with latitude increase accompanied by a higher percentage of trees with poor growth status (diebacks, dead trees, etc.). 4) The top abundant trees were Populus spp., Cinnamomum camphora, Salix spp., Platanus acerifolia, Ficus macrocarpa (5.5 %-14.6 %), and the arbor-shrub-herb three-layer structured forests took 52.3 % of total sites. With latitude rise, increasing abundance of Populus spp., Salix spp., elm, and pine but decreasing abundance of the unrecognizable tree groups were found (p < 0.05). 5) We also constructed a street tree comprehensive index based on their potential for providing services to citizen from the inventory data and found it was negatively related to latitudes. RDA ordination showed that geo-climatic conditions (49 %-61.5 %) and social developments (21.4 %-52.7 %) were almost equally responsible for tree size, growth status, and vertical structural variations, while road width (lane number of the street) was the most potent predictor (coefficient > 2.0 %, p < 0.01) for these variations. Our study can benefit the national-level management of urban forests and inventory-based various ecological service precise evaluation.

Big-sized trees and higher species diversity improve water holding capacities of forests in northeast China.

High water-holding forests are essential for adapting to drought climates under global warming, and a central issue is which type of forests could conserve more water in the ecosystem. This paper explores how forest structure, plant diversity, and soil physics impact forest water-holding capacities. We investigated 720 sampling plots by measuring water-holding capacities from 1440 soil and litter samples, 8400 leaves, and 1680 branches and surveying 18,054 trees in total (28 species). Water-holding capacities were measured as four soil indices (Maxwc, maximum water-holding capacity; Fcwc, field water-holding capacity; Cpwc, soil capillary water-holding capacity; Ncpwc, non-capillary water-holding capacity), two litter metrics (Maxwcl, maximum water-holding capacity of litters; Ewcl, effective water-holding capacity of litters), and canopy interception (C, the sum of estimated water interception of all branches and leaves of all tree species in the plot). We found that water-holding capacity in the big-sized tree plots was 4-25 % higher in the litters, 54-64 % in the canopy, and 6-37 % in the soils than in the small-sized plots. The higher species richness increased all soil water-holding capacities compared to the lowest richness plot. Higher Simpson and Shannon-Wiener plots had 10-27 % higher Ewcl and C than the lowest plots. Bulk density had the strongest negative relations with Maxwc, Cpwc, and Fcwc, whereas field soil water content positively affected them. Soil physics, forest structure, and plant diversity explained 90.5 %, 5.9 %, and 0.2 % of the water-holding variation, respectively. Tree sizes increased C, Ncpwc, Ewcl directly (p < 0.05), and richness increased Ewcl directly (p < 0.05). However, the direct effects from the uniform angle index (tree distribution evenness) were balanced by their indirect effect from soil physics. Our findings highlighted that the mixed forests with big-sized trees and rich species could effectively improve the water-holding capacities of the ecosystem.

Big data-based urban greenness in Chinese megalopolises and possible contribution to air quality control.

Urban greenness is essential for people's daily lives, while its contribution to air quality control is unclear. In this study, Streetview big data of urban greenness and air quality data (Air Quality Index, PM2.5, PM10, SO2, NO2, O3, CO) from 206 monitoring stations from 27 provincial capital cities in China were analyzed. The national averages for the sky, ground and middle-level (shrub and short trees) view greenness were 5.4%, 5.5%, and 15.4%, respectively, and the sky:ground:middle ratio was 2:2:6. Street-view/bird-view greenness ratio averaged at 1.1. Large inter-city variations were observed in all the greenness parameters, and the weak associations between all street-view parameters and bird-eye greenspace percentage (21%-73%) indicate their representatives of different aspects of green infrastructures. All air quality parameters were higher in winter than in summer, except O3. Over 90% of air quality variation could be explained by socioeconomics and geoclimates, suggesting that air quality control in China should first reduce efflux from social economics, while geoclimatic-oriented ventilation facilitation design is also critical. For different air quality components, greenness had most significant associations with NO2, O3 and CO, and street-view/bird-view ratio was the most powerful indicator of all greenness parameters. Pooled-data analysis at national level showed that street-view greenness was responsible for 2.3% of the air quality variations in the summer and 3.6% in the winter; however, when separated into different regions (North-South China; East-West China), the explaining power increased up to 16.2%. Increased NO2 was accompanied with decreased O3, indicating NO titration effect. The higher O3 aligned with the higher street-view greenness, showing the greenness-related precursor risk for O3 pollution. Our study manifested that big internet data could identify the association of greenness and air pollution from street view scale, which can favor urban greenness management and evaluation in other regions where street-view data are available.