Browsing by Author "Zhang, Bai"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    A Comparative Assessment of Geostatistical, Machine Learning, and Hybrid Approaches for Mapping Topsoil Organic Carbon Content
    (MDPI, 2019-04) Chen, Lin; Ren, Chunying; Li, Lin; Wang, Yeqiao; Zhang, Bai; Wang, Zongming; Li, Linfeng; Earth Sciences, School of Science
    Accurate digital soil mapping (DSM) of soil organic carbon (SOC) is still a challenging subject because of its spatial variability and dependency. This study is aimed at comparing six typical methods in three types of DSM techniques for SOC mapping in an area surrounding Changchun in Northeast China. The methods include ordinary kriging (OK) and geographically weighted regression (GWR) from geostatistics, support vector machines for regression (SVR) and artificial neural networks (ANN) from machine learning, and geographically weighted regression kriging (GWRK) and artificial neural networks kriging (ANNK) from hybrid approaches. The hybrid approaches, in particular, integrated the GWR from geostatistics and ANN from machine learning with the estimation of residuals by ordinary kriging, respectively. Environmental variables, including soil properties, climatic, topographic, and remote sensing data, were used for modeling. The mapping results of SOC content from different models were validated by independent testing data based on values of the mean error, root mean squared error and coefficient of determination. The prediction maps depicted spatial variation and patterns of SOC content of the study area. The results showed the accuracy ranking of the compared methods in decreasing order was ANNK, SVR, ANN, GWRK, OK, and GWR. Two-step hybrid approaches performed better than the corresponding individual models, and non-linear models performed better than the linear models. When considering the uncertainty and efficiency, ML and two-step approach are more suitable than geostatistics in regional landscapes with the high heterogeneity. The study concludes that ANNK is a promising approach for mapping SOC content at a local scale.
  • Thumbnail Image
    Item
    Mapping Forest Cover in Northeast China from Chinese HJ-1 Satellite Data Using an Object-Based Algorithm
    (MDPI, 2018-12-16) Ren, Chunying; Zhang, Bai; Wang, Zongming; Li, Lin; Jia, Mingming; Earth Sciences, School of Science
    Forest plays a significant role in the global carbon budget and ecological processes. The precise mapping of forest cover can help significantly reduce uncertainties in the estimation of terrestrial carbon balance. A reliable and operational method is necessary for a rapid regional forest mapping. In this study, the goal relies on mapping forest and subcategories in Northeast China through the use of high spatio-temporal resolution HJ-1 imagery and time series vegetation indices within the context of an object-based image analysis and decision tree classification. Multi-temporal HJ-1 images obtained in a single year provide an opportunity to acquire phenology information. By analyzing the difference of spectral and phenology information between forest and non-forest, forest subcategories, decision trees using threshold values were finally proposed. The resultant forest map has a high overall accuracy of 0.91 ± 0.01 with a 95% confidence interval, based on the validation using ground truth data from field surveys. The forest map extracted from HJ-1 imagery was compared with two existing global land cover datasets: GlobCover 2009 and MCD12Q1 2009. The HJ-1-based forest area is larger than that of MCD12Q1 and GlobCover and more closely resembles the national statistics data on forest area, which accounts for more than 40% of the total area of the Northeast China. The spatial disagreement primarily occurs in the northern part of the Daxing'an Mountains, Sanjiang Plain and the southwestern part of the Songliao Plain. The compared result also indicated that the forest subcategories information from global land cover products may introduce large uncertainties for ecological modeling and these should be cautiously used in various ecological models. Given the higher spatial and temporal resolution, HJ-1-based forest products could be very useful as input to biogeochemical models (particularly carbon cycle models) that require accurate and updated estimates of forest area and type.
  • Thumbnail Image
    Item
    Variations in the light absorption coefficients of phytoplankton, non-algal particles and dissolved organic matter in reservoirs across China
    (Elsevier, 2021-10) Shang, Yingxin; Jacinthe, Pierre-Andre; Li, Lin; Wen, Zhidan; Liu, Ge; Lyu, Lili; Fang, Chong; Zhang, Bai; Hou, Junbin; Song, Kaishan; Earth Sciences, School of Science
    Reservoirs were critical sources of drinking water for many large cities around the world, but progress in the development of large-scale monitoring protocols to obtain timely information about water quality had been hampered by the complex nature of inland waters and the various optical conditions exhibited by these aquatic ecosystems. In this study, we systematically investigated the absorption coefficient of different optically-active constituents (OACs) in 120 reservoirs of different trophic states across five eco-regions in China. The relationships were found between phytoplankton absorption coefficient at 675 nm (aph (675)) and Chlorophyll a (Chla) concentration in different regions (R2:0.60-0.82). The non-algal particle (NAP) absorption coefficient (aNAP) showed an increasing trend for reservoirs with trophic states. Significant correlation (p < 0.05) was observed between chromophoric dissolved organic matter (CDOM) absorption and water chemical parameters. The influencing factors for contributing the relative proportion of OACs absorption including the hydrological factors and water quality factors were analyzed. The non-water absorption budget from our data showed the variations of the dominant absorption types which underscored the need to develop and parameterize region-specific bio-optical models for large-scale assessment in water reservoirs.