Bai Zhang

Northeast Institute of Geography and Agroecology, 臺中市, Taiwan, Taiwan

Are you Bai Zhang?

Claim your profile

Publications (92)61.03 Total impact


  • No preview · Dataset · Nov 2015
  • [Show abstract] [Hide abstract]
    ABSTRACT: Chromophoric dissolved organic matter (CDOM), the light absorbing fraction of dissolved organic carbon (DOC), together with phytoplankton and total suspended matter are the main optically active components could be retrieved by remote sensing data. Generally, different composition of DOC and CDOM corresponds to different water surface reflectance. Absorption properties of CDOM and retrieval models for CDOM and DOC were investigated with data from potable reservoirs located in the central of Jilin Province. Water sampling field surveys were conducted on 15, 16 and 19 of September 2012 across the Shitoukoumen, Erlonghu and Xilicheng reservoirs, respectively. Both empirical regression (single band model and band ratio model) and partial least squares coupled with back-propagation artificial neural models (PLSBPNN) were established to estimate CDOM absorption coefficient at 355 nm [aCDOM(355)] and DOC concentration with in situ measured remote sensing reflectance. It was found that the band ratio models and PLSBPNN model performed well for estimating DOC concentration while the band ratio models yielded the best result in retrieval CDOM. Moreover, all the three models performed better on the DOC concentration estimation than the performance on aCDOM(355). Band ratio models outperformed (R 2 = 0.55) other models for estimating CDOM absorption coefficient, while PLSBPNN model outperformed other models with respect to DOC estimation (R 2 = 0.93). High spectral slope values indicated that CDOM in the potable waters primarily comprised low molecular weight organic substances; while sources of DOC were mainly coming from exogenous input, which was the main reason lead to the difference of model performances on DOC and aCDOM(355) estimation. The algorithms developed in this study is needed to be tested and refined with more in situ spectral data, also future work is still needed to be undertaken for characterizing the dynamic of the potable water quality with remotely sensed imagery.
    No preview · Article · Jul 2015 · Journal of the Indian Society of Remote Sensing
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The Nenjiang River Basin is an important foodstuff base and eco-environmental fragile area in Northeast China. With the rapid rise in human population, human-induced changes in land use/land cover form an important component of regional environment and ecosystem service change. At the local and regional level, the ecosystem service concept can act as a decision support tool for a stakeholder to reach sustainable land use management. However, the prevailing ecosystem service evaluation would produce a biggish warp when it is applied to concrete area. So, it is essential to evaluate ecosystem service change according to the local actuality. According to 1:250,000 land use/land cover maps of China and the adjusted equivalent value per unit area of ecosystem services in the Nenjiang River Basin, we evaluated the ecosystem service change of the river basin from 1980 to 2005. The forest and wetland, which are mainly located in the upstream mountainous area of the Nenjiang River Basin, were the two valuable land cover types, accounting for more than three quarters of the total ecosystem service value of the river basin. As for individual ecosystem service, besides the food production, all of the ecosystem service values declined from 1980 to 2005. The total decline of 2.43 billion USD was mainly due to the cultivation of grassland (14.34 % of the area in 1980) and wetland (4.62 % of the area in 1980) in the downstream plain. Due to the increase in population and the concomitant requirement of grain, the inconsistency between decision-making at the macro-level, and the objective of agricultural production at the micro-level, cultivated land was increased through zealous reclamation of grassland, marginal woodland, and even fallow land. Tremendous land use/land cover changes had caused great damages to the ecological environment such as land degradation and ecosystem service recession. So, the policies of the Grain for Green and Construction of Ecological Province projects should be well-implemented to optimize land use/land cover.
    Preview · Article · Jun 2015
  • [Show abstract] [Hide abstract]
    ABSTRACT: Absorption and fluorescence properties of chromophoric dissolved organic matter (CDOM) in rivers across the Liaohe River Delta, a large estuary located in the southern region of northeast China, were investigated using spectroscopy and fluorescence to analyze CDOM characteristics, composition and sources in winter (January), spring (April and May) and autumn (September) 2013. Results indicated that CDOM absorption of ice samples was lower compared to water samples. CDOM absorption also showed significant spatial variation but not temporal variation. In contrast, dissolved organic carbon (DOC) concentrations showed obvious temporal characteristics. A stronger linear relationship was found between CDOM absorption and DOC concentration in winter (water, R 2 = 0.95, p < 0.001; ice, R 2 = 0.85, p < 0.001) compared to samples from other seasons (April, R 2 = 0.51, p < 0.01; May, R 2 = 0.34, p < 0.05; September, R 2 = 0.45, p < 0.01). CDOM fluorescence varied over a large range across seasons, with the highest levels observed in January at the Xisha River (XSR). Moreover, strong linear relationships were also observed between CDOM absorption and fluorescence intensity at 355 nm [Fn(355)] in January (R 2 = 0.87, p < 0.001), May (R 2 = 0.76, p < 0.001) and September (R 2 = 0.94, p < 0.001). Also based on study findings, CDOM fluorophores identified by 3-D excitation-emission matrices (EEMs) illustrated that the rivers across the Liaohe Delta were seriously polluted by anthropogenic disturbances, exhibiting strong protein-like fluorescence of CDOM in the water column. In addition, the results also confirm that absorption and EEMs would be useful tools for tracing the sources and characteristics of CDOM and monitoring riverine water quality.
    No preview · Article · Jan 2015 · Wetlands
  • Source

    Full-text · Dataset · Apr 2014
  • [Show abstract] [Hide abstract]
    ABSTRACT: Remote sensing image fusion is an effective way to extract a large volume of data from multi-source images. However, traditional image fusion methods cannot meet the requirements of applications because they can lose spatial information or distort spectral characteristics. In this paper, a new wavelet method based on a local algorithm is presented. The proposed method fuses multi-spectral (MS) and panchromatic (PAN) images to improve spatial information and preserve spectral characteristics. The main advantage of the new fusion method is the exploitation of the dependency between neighboring pixels. SPOT5 MS and PAN images were employed to execute the fusion methods. To compare with the new method, the principal component analysis (PCA), wavelet transformation, and PCA-based wavelet (PCA+W) image fusion methods were selected. Qualitative and quantitative analyses and classification accuracy assessment were conducted to evaluate the performance of the fusion methods. The results demonstrate that the new wavelet method based on a local algorithm is better than traditional image fusion methods. The new fusion method can achieve a wide range of balance between high spatial resolution retention and spectral characteristic preservation; thus, the new method is suitable for different applications.
    No preview · Article · Oct 2013 · Computers & Geosciences
  • Ying Liu · Bai Zhang · Li-min Wang · Nan Wang
    [Show abstract] [Hide abstract]
    ABSTRACT: Support vector machines (SVM) are nowadays receiving increasing attention in remote sensing applications although this technique is very sensitive to the parameters setting and training set definition. Self-training is an effective semisupervised method, which can reduce the effort needed to prepare the training set by training the model with a small number of labeled examples and an additional set of unlabeled examples. In this study, a novel semisupervised SVM model that uses self-training approach is proposed to address the problem of remote sensing land cover classification. The key characteristics of this approach are that (1) the self-adaptive mutation particle swarm optimization algorithm is introduced to get the optimum parameters that improve the generalization performance of the SVM classifier, and (2) the Gustafson-Kessel fuzzy clustering algorithm is proposed for the selection of unlabeled points to reduce the impact of ineffective labels. The effectiveness of the proposed technique is evaluated firstly with samples from remote sensing data and then by identifying different land cover regions in the remote sensing imagery. Experimental results show that accuracy level is increased by applying this learning scheme, which results in the smallest generalization error compared with the other schemes.
    No preview · Article · Sep 2013 · Computers & Geosciences
  • Source
    Tian-tian Shao · Kai-shan Song · Bai Zhang
    [Show abstract] [Hide abstract]
    ABSTRACT: Chlorophyll is one of the most important pigments for plant photosynthesis as a proxy of health of vegetation and photosynthetic capacity. In this study, estimation of chlorophyll content of corn of western regions of Jilin Province is conducted based on both the measured hyperspectral data and the environmental satellite hyperspectral image data (HJ-HSI). Four kinds of vegetation indices including NDVI, MSR, MCAVI/OSAVI and TCAVI/OSAVI, are used for the inversion of chlorophyll. Among these vegetation indices, NDVI performs the best in estimating chlorophyll, followed by MSR and TCAVI/OSAVI. Otherwise, using HSI data based on vegetation indices in modeling performs not very well but the verification accuracy is rather high (R2>0.9). And the slopes of validated models are all less than 0.4, which indicate that these estimates are lower than the measured results.
    Preview · Article · Aug 2013
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this study, the Surface Energy Balance Algorithms for Land (SEBAL) model and Moderate Resolution Imaging Spectroradiometer (MODIS) products from Terra satellite were combined with meteorological data to estimate evapotranspiration (ET) over the Sanjiang Plain, Northeast China. Land cover/land use was classified by using a recursive partitioning and regression tree with MODIS Normalized Difference Vegetation Index (NDVI) time series data, which were reconstructed based on the Savitzky-Golay filtering approach. The MODIS product Quality Assessment Science Data Sets (QA-SDS) was analyzed and all scenes with valid data covering more than 75% of the Sanjiang Plain were selected for the SEBAL modeling. This provided 12 overpasses during 184-day growing season from May 1st to October 31st, 2006. Daily ET estimated by the SEBAL model was misestimaed at the range of −11.29% to 27.57% compared with that measured by Eddy Covariance system (10.52% on average). The validation results show that seasonal ET from the SEBAL model is comparable to that from ground observation within 8.86% of deviation. Our results reveal that the time series daily ET of different land cover/use increases from vegetation on-going until June or July and then decreases as vegetation senesced. Seasonal ET is lower in dry farmland (average (Ave): 491 mm) and paddy field (Ave: 522 mm) and increases in wetlands to more than 586 mm. As expected, higher seasonal ET values are observed for the Xingkai Lake in the southeastern part of the Sanjiang Plain (Ave: 823 mm), broadleaf forest (Ave: 666 mm) and mixed wood (Ave: 622 mm) in the southern/western Sanjiang Plain. The ET estimation with SEBAL using MODIS products can provide decision support for operational water management issues.
    Full-text · Article · Feb 2013 · Chinese Geographical Science
  • Weijie Liu · Bai Zhang
    [Show abstract] [Hide abstract]
    ABSTRACT: This article is an elementary exploration and practice of applying the theory and method of ecological footprint to assess regional ecological security. The ecological footprint from 1996 to 2008 and ecological security of Jilin Province of China were calculated and analyzed based on the theory of ecological footprint. Furthermore, the concept and model of ecological footprint pressure index (EFPI) as well as the grade system of regional ecological security was put forward based on ecological footprint so as to analyze the regional ecological security, and some measures to slow down the growth of the ecological footprint and maintain the regional ecological security were put forward at last. The following results were obtained. (1) It is feasible to analyze regional ecological security based on EFPI, which can reflect the degree of ecological security from two respects of carrying capacity and pressure. (2) Among 13 years from 1996 to 2008, the per capita ecological capacity in three regions appears to be a reducing tendency; the per capita ecological footprint shows an increasing tendency.
    No preview · Conference Paper · Jun 2012
  • Ying Liu · Bai Zhang · Lihua Huang · Limin Wang
    [Show abstract] [Hide abstract]
    ABSTRACT: Nowadays, support vector machines (SVM) are receiving increasing attention in land cover/use classification although one of the major drawbacks of the technique is the kernel function selection and its parameters setting. In this paper, a novel SVM parameters optimization method based on self- adaptive mutation particle swarm optimizer (SAMPSO-SVM) is proposed to improve the generalization performance of the SVM classifier. The SAMPSO algorithm, which is based on the variance of the population's fitness, can break away the local optimum by the operation of self-adaptive mutation. Accordingly, very high classification accuracy will be achieved with the best value of the parameters of SVM, which have been searched using SAMPSO. In order to verify the validity of this SAMPSO-SVM method, a remote sensing land use/cover classification model is constructed using multi-spectral Landsat-5 TM data. In particular, they are organized so as to test the sensitivity of the SAMPSO-SVM model and that of the other reference classifiers used for comparison, i.e. maximum likelihood classifier (MLC), SVM classifier and standard PSO algorithm for SVM parameters optimization model (PSO-SVM). On an average, the SAMPSO-SVM model yielded an overall accuracy of 93.59% against 83.92% for maximum likelihood classier and outperformed PSO-SVM classier in terms of overall accuracy (by about 2%). The obtained results clearly confirm the effectiveness and robustness of the SAMPSO-SVM approach to the remote sensing land use/cover classification.
    No preview · Article · Apr 2012 · Journal of Food Agriculture and Environment
  • [Show abstract] [Hide abstract]
    ABSTRACT: Haze exerts a large effect on visibility reduction and has serious impacts on air quality and human health. Understanding the sources and transport of haze is of importance to improve the regional air quality and evaluate its health effects. In this study, we investigated a typical haze episode that occurred in northeast (NE) China during 4-6 November 2010 by analyzing the ground PM10 measurements from 11 monitoring sites, aerosol Lidar observations, synoptic charts, MODIS satellite imageries, and back trajectories. Our analyses suggest that the regional haze formed in the North China Plain (NCP) under stagnant conditions can be transported to NE China in ~1-3 days across Bohai Bay and Liaodong Bay - a typical transport pathway associated with the topography of northern China. The haze episode appeared to evolve progressively from southwest to northeast in the region of NE China, in agreement with the appearance of PM10 peak values, wind patterns, MODIS images and the back trajectories of air masses. Due to the haze impact, NE China showed significantly elevated particulate matter pollution by a factor of ~4-6 with the peak concentrations reaching ~410 μg m-3. The results together indicate that the regional transport from the NCP has a significant contribution to the PM pollution in NE China, thus efforts to control the source emissions over the NCP would be effective to improve the air quality in NE China.
    No preview · Article · Jan 2012 · Scientific online letters on the atmosphere: SOLA
  • [Show abstract] [Hide abstract]
    ABSTRACT: Marshes in the Sanjiang Plain of Northeast China have undergone dramatic loss and fragmentation over the past decades. This paper analyzed the loss and fragmentation of these marshes for the period 1954–2005 using historical land-cover information and remote sensing data. In 1954, marshes covered one-third of the total land area but have decreased by 77% over the 50year period. Results showed two distinct periods of impact: 1954–1986 and 1987–2005. In the earlier period, the number of marsh patches fell from 4,799 to 1,476 (−69.2%), and total marsh area decreased from 35,270km2 to 13,893km2 (−60.6%). In the latter period, the number of marshes declined from 1,476 to 1,037 (−29.7%), and the total area decreased from 13,893km2 to 8,100km2 (−41.7%). The rapid decrease in the number and area of marshes during 1954–1986 was largely attributed to extensive agricultural reclamation under the “Food First” agricultural policy. This resulted in many negative ecological consequences. In contrast, the slower reduction of marsh areas during 1987–2005 was due to the implementation of governmental policies for protecting and restoring marshes. Increasing air temperature would otherwise have enhanced crop yields and stimulated the conversion of marsh into crops. KeywordsAgriculture–Climatic changes–Marsh loss–Policy–Remote sensing
    No preview · Article · Oct 2011 · Wetlands
  • [Show abstract] [Hide abstract]
    ABSTRACT: The eddy covariance technique provides continuous measurements of plot-level net ecosystem carbon exchange (NEE) across a wide range of vegetation types. However, these NEE estimates only represent fluxes at the tower footprint scale. To quantify the NEE over regions or continents, flux tower measurements need to be up-scaled to large areas. In the present study, we propose a new NEE model solely based on Moderate Resolution Imaging Spectroradiometer data, including enhanced vegetation index (EVI), land surface water index (LWSI), land surface temperature (LST), and Terra nighttime LST′. Site-specific data from the deciduous-dominated Harvard Forest flux site were used. Analysis covered six years (2001–2006) of CO2 flux data. The data of the first four years were used for model building and the rest as validation set. Compared with the model based solely on EVI, we also introduced LST and LSWI into the new model. The results showed that this method could further improve the precision (R2 and RMSE reached 0.857 and 1.273, respectively) and generally capture the expected seasonal patterns of NEE.
    No preview · Article · Oct 2011 · Ecological Engineering
  • [Show abstract] [Hide abstract]
    ABSTRACT: The Songnen Plain is an important commodity grain product base in China for which a spatiotemporal pattern of actual evapotranspiration (ETa) would provide critical important information to evaluate crop growth status and water use efficiency. ETa over the Songnen Plain in the 2008 growing season (from May to September) was mapped using the moderate resolution imaging spectroradiometer time-series products based on the surface energy balance algorithm for land model and the Penman-Monteith equation. The estimated ETa was validated using eddy covariance surface data. The calculated and observed ETa values were highly consistent with a total difference of 18.26% in the whole growing season. Therefore, the ETa retrieval method based on remote sensing technology could satisfy the requirements for regional ETa estimation over the Songnen Plain. The total ETa over the Songnen Plain in the 2008 growing season ranged from 182.7 to 1002.4 mm, and the average value for the whole study area was 591.1 +/- 122.2 mm (standard deviation). ETa exhibited obvious spatial variation, gradually increasing from low values in the southwest to higher values in the east and northeast. Monthly ETa varied with meteorological conditions, land covers, root-zone soil moisture, and vegetation phenology. Higher monthly ETa values appeared in June, July, and August with a maximum value of 139.5 mm observed in July. The average monthly ETa for water-body, woodland, and wetland was much higher than cropland and grassland during the growing season. Grassland obtained the lowest monthly ETa due to the scarcity of rainfall and lower groundwater level.
    No preview · Article · Jul 2011 · Journal of Applied Remote Sensing
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: It is difficult to analyse soil properties quantitatively with multispectral remote sensing data. An alternative solution is to determine the main spectral characteristic control points of soil hyperspectral reflectance curves by sensitivity analysis methods. Hyperspectral reflectance is simulated using the control points based on multispectral reflectance collected from satellites in this study. The laboratory hyperspectral reflectance and its continuum-removed curve of Phaeozem and the parent material (PM) of samples collected from Heilongjiang Province, China were analysed, and the spectral characteristic control points determined. Hyperspectral simulating linear and quadratic models based on laboratory reflectance were then built. Results show that montmorillonite and illite are the dominant minerals in Phaeozem PM. Organic matter content determines the spectral characteristics of Phaeozem and makes it suitable for reflectance simulation in the spectrum range of 1000 nm and less; the higher the organic matter content the greater the spectral absorption area. There are two absorption valleys at 500 and 660 nm, which determine the spectral characteristic control points of Phaeozem between 450 and 930 nm, namely 450, 500, 590, 660 and 930 nm. Both the linear and quadratic simulation models built with the characteristic control points accurately describe Phaeozem reflectance, which proves that the characteristic control points are selected reasonably and representatively. The hyperspectral simulation method based on multispectral reflectance closely represents the characteristics of Phaeozem hyperspectral reflectance, partly removes noise and improves the precision of predicting organic matter content. Therefore the method is feasible and useful for data compression of Phaeozem hyperspectral reflectance, soil and vegetation indices building, and quantitative remote sensing in the Phaeozem Zone, northeast China.
    Full-text · Article · Jul 2011 · International Journal of Remote Sensing
  • [Show abstract] [Hide abstract]
    ABSTRACT: It is important to better understand the spatial distribution of non-point source pollution and related influencing factors in China. On the bases of field sampling, using GIS techniques, this paper analyzed the spatial characteristics of non- point pollution and its relation to land use structure, taking Muling river sub-basin in Sanjiang Plain as the study area. The result shows that the value of TSS, TN, and TP exceed the country criteria and the discrepancy in space is little. In the sub- basin which are controlled mainly by a single land use type, surface water quality of woodland controlled sub-basin is better than that of cropland controlled sub-basin. The concentration of pollutants in combined structure (cropland, woodland and grassland) is highest in three land use structures. This may be due to the complex surface background and other point source pollution in this study area.
    No preview · Article · Jun 2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Research on the optical characteristics of water color constituents in Chagan Lake of Jilin Province, Northeast China was carried out in order to investigate the variability of the spectra absorption parameters as inputs to bio-optical models and remote sensing algorithms for converting observed spectral signals into water quality information. Samples of total particulates, non-algal particles and colored dissolved organic matter (CDOM) were first prepared by quantitative filter technique (QFT) and then absorption coefficients of these color producing agents were determined by spectrophotometry. Spectral characteristics of absorption coefficients by total particulate matter, spectral specific absorption dependency on chlorophyll concentration (Chl-a) of phytoplankton, spectral absorption slopes variation for CDOM and non-algal particles and their corresponding reasons were examined and clarified over five months of 2009 and 2010 in this study. Results suggest that total particulate spectral absorption in Chagan Lake is mainly dominated by non-algal particles in most cases, but phytoplankton could be the dominant contributor when chlorophyll concentration is high (up to 84.48 mg/m3 in autumn 2010). The specific absorption coefficients of phytoplankton particulate (a*ph(λ)) dependency on Chl-a is significantly variable due to relative contributions of package effect and accessory pigments, and the parameters of power function are clearly biased on a long time span. The sources of variability in spectral absorption slopes of CDOM and non-algal particles are mainly attributed to the changing proportions of high molecular weight humic acids and mineral suspended sediments in waters, respectively. Keywordscolored dissolved organic matter–spectra absorption–water constituents–spectral slope–Chagan Lake
    Full-text · Article · Jun 2011 · Chinese Geographical Science
  • [Show abstract] [Hide abstract]
    ABSTRACT: Contaminants in the snow can be used to reflect regional and global environmental pollution caused by human activities. However, so far, the research on space-time monitoring of snow contamination concentration for a wide range or areas difficult for human to reach is very scarce. In the present paper, based on the simulated atmospheric deposition experiments, the spectroscopy technique method was applied to analyze the effect of different contamination concentration on the snow reflectance spectra. Then an evaluation of snow contamination concentration (SCC) retrieval methods was conducted using characteristic index method (SDI), principal component analysis (PCA), BP neural network and RBF neural network method, and the estimate effects of four methods were compared. The results showed that the neural network model combined with hyperspectral remote sensing data could estimate the SCC well.
    No preview · Article · May 2011 · Guang pu xue yu guang pu fen xi = Guang pu
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The concentrations of chlorophyll-a (Chl-a) and total suspended matter (TSM) are major water quality parameters that can be retrieved using remotely sensed data. Water sampling works were conducted on 15 July 2007 and 13 September 2008 concurrent with the Indian Remote-Sensing Satellite (IRS-P6) overpass of the Shitoukoumen Reservoir. Both empirical regression and back-propagation artificial neural network (ANN) models were established to estimate Chl-a and TSM concentration with both in situ and satellite-received radiances signals. It was found that empirical models performed well on the TSM concentration estimation with better accuracy (R (2) = 0.94, 0.91) than their performance on Chl-a concentration (R (2) = 0.62, 0.75) with IRS-P6 imagery data, and the models accuracy marginally improved with in situ spectra data. Our results indicated that the ANN model performed better for both Chl-a (R (2) = 0.91, 0.82) and TSM (R (2) = 0.98, 0.94) concentration estimation through in situ collected spectra; the same trend followed for IRS-P6 imagery data (R (2) = 0.75 and 0.90 for Chl-a; R (2) = 0.97 and 0.95 for TSM). The relative root mean square errors (RMSEs) from the empirical model for TSM (Chl-a) were less than 15% (respectively 27.2%) with both in situ and IRS-P6 imagery data, while the RMSEs were less than 7.5% (respectively 18.4%) from the ANN model. Future work still needs to be undertaken to derive the dynamic characteristic of Shitoukoumen Reservoir water quality with remotely sensed IRS-P6 or Landsat-TM data. The algorithms developed in this study will also need to be tested and refined with more imagery data acquisitions combined with in situ spectra data.
    Full-text · Article · Apr 2011 · Environmental Monitoring and Assessment

Publication Stats

635 Citations
61.03 Total Impact Points

Institutions

  • 2007-2013
    • Northeast Institute of Geography and Agroecology
      臺中市, Taiwan, Taiwan
  • 2005-2012
    • Chinese Academy of Sciences
      • Northeast Institute of Geography and Agroecology
      Peping, Beijing, China
  • 2011
    • National Space Science
      Peping, Beijing, China