Detection of sea ice in sediment laden water using MODIS in the Bohai Sea: a CART decision tree method

An inversion algorithm based on the Classification and Regression Tree (CART) has been developed to retrieve sea ice from Moderate Resolution Imaging Spectroradiometer (MODIS) images in the Bohai Sea where the sea water is characterized by a high concentration of suspended sediment in coastal areas. The inversion algorithm has been successfully applied to the sea-ice extraction from 2009 to 2012. The estimated sea ice is compared with previous studies and the comparison shows reasonable agreement. The model is further examined using sea-ice data from higher-spatial-resolution satellites, and the result indicates that the CART method is able to successfully retrieve sea ice in high sediment environments in the Bohai Sea. To comprehensively understand the working principles of the CART, a series of sensitivity studies to model input parameters such as sampling locations, the number of bands, and the effect of the thermal infrared band (TIB), was conducted. The sensitivity studies show that the CART method is easy to set up and the results are realistic. The TIB may play an important role in sea-ice inversion in turbid waters. The algorithm is also compared with a ratio-threshold segmentation (RTS) method, a common way to retrieve sea ice from satellite images in open oceans, and the comparison indicates that the algorithm developed in the present article is superior to the RTS method in high sediment environments.

[1]  J. Qi,et al.  An unstructured-grid, finite-volume sea ice model : development, validation, and application , 2011 .

[2]  Roberto Teti,et al.  Principal component analysis for feature extraction and NN pattern recognition in sensor monitoring of chip form during turning , 2014 .

[3]  Hua Su,et al.  Monitoring the Spatiotemporal Evolution of Sea Ice in the Bohai Sea in the 2009–2010 Winter Combining MODIS and Meteorological Data , 2011, Estuaries and Coasts.

[4]  Richard A. Olshen,et al.  CART: Classification and Regression Trees , 1984 .

[5]  Hua Su,et al.  Improving MODIS sea ice detectability using gray level co-occurrence matrix texture analysis method: A case study in the Bohai Sea , 2013 .

[6]  J. Levy,et al.  Using remote sensing to estimate sea ice thickness in the Bohai Sea, China based on ice type , 2009 .

[7]  Leen-Kiat Soh,et al.  A comprehensive, automated approach to determining sea ice thickness from SAR data , 1995, IEEE Trans. Geosci. Remote. Sens..

[8]  John R. Jensen,et al.  Introductory Digital Image Processing: A Remote Sensing Perspective , 1986 .

[9]  Menghua Wang,et al.  Sea ice properties in the Bohai Sea measured by MODIS-Aqua: 1. Satellite algorithm development , 2012 .

[10]  W. Rees.  Remote Sensing of Snow and Ice , 2005 .

[11]  John A. Richards,et al.  Remote Sensing Digital Image Analysis , 1986 .

[12]  Xiangming Zeng,et al.  Seasonal distribution of suspended sediment in the Bohai Sea, China , 2014 .

[13]  Chandra Kambhamettu,et al.  High resolution (400 m) motion characterization of sea ice using ERS-1 SAR imagery , 2008 .

[14]  Menghua Wang,et al.  Satellite views of the Bohai Sea, Yellow Sea, and East China Sea , 2012 .

[15]  M. Haller,et al.  Mapping sediment-laden sea ice in the Arctic using AVHRR remote-sensing data: Atmospheric correction and determination of reflectances as a function of ice type and sediment load , 2007 .

[16]  P. Eriksson,et al.  A method for observing compression in sea ice fields using IceCam , 2009 .

[17]  Chris Derksen,et al.  Development of a water clear of sea ice detection algorithm from enhanced SeaWinds/QuikSCAT and AMSR-E measurements , 2010 .

[18]  M. F. Meier,et al.  Remote sensing of snow and ice. , 1980 .

[19]  Chris Derksen,et al.  Extending the QuikSCAT record of seasonal melt–freeze transitions over Arctic sea ice using ASCAT , 2014 .

[20]  Hua Su,et al.  Using MODIS data to estimate sea ice thickness in the Bohai Sea (China) in the 2009-2010 winter , 2012 .

[21]  Menghua Wang,et al.  Sea ice properties in the Bohai Sea measured by MODIS-Aqua: 2. Study of sea ice seasonal and interannual variability , 2012 .

[22]  Paul M. Mather,et al.  An assessment of the effectiveness of decision tree methods for land cover classification , 2003 .

[23]  Kun Lei,et al.  Sediment transport off the Huanghe (Yellow River) delta and in the adjacent Bohai Sea in winter and seasonal comparison , 2011 .

[24]  R. Massom.  Satellite remote sensing of polar regions : applications, limitations and data availability , 1991 .

[25]  G. Milne,et al.  Calibrating a glaciological model of the Greenland ice sheet from the Last Glacial Maximum to present-day using field observations of relative sea level and ice extent , 2009 .

[26]  Wei-Yin Loh,et al.  Classification and regression trees , 2011, WIREs Data Mining Knowl. Discov..

[27]  A. Keen,et al.  A sensitivity study of the sea ice simulation in the global coupled climate model, HadGEM3 , 2014 .

[28]  Yang Shao,et al.  Comparison of support vector machine, neural network, and CART algorithms for the land-cover classification using limited training data points , 2012 .

[29]  Anna Barbati,et al.  Stochastic gradient boosting classification trees for forest fuel types mapping through airborne laser scanning and IRS LISS-III imagery , 2013, Int. J. Appl. Earth Obs. Geoinformation.

[30]  D. Steinberg.  CART: Classification and Regression Trees , 2009 .

[31]  Merryl Alber,et al.  Classification of salt marsh vegetation using edaphic and remote sensing-derived variables , 2014 .

[32]  Jose Raul Romo-Leon,et al.  Using remote sensing tools to assess land use transitions in unsustainable arid agro-ecosystems , 2014 .