Chin J Plant Ecol ›› 2008, Vol. 32 ›› Issue (4): 967-976.DOI: 10.3773/j.issn.1005-264x.2008.04.027

• Original article • Previous Articles    


BAI T. Jay, LIANG Ying-Quan   

  1. MDSM Data Analysis Services, LLC., 615 Joanne St. Fort Collins, CO 80524-3684, USA
  • Received:2007-12-10 Accepted:2008-03-11 Online:2008-07-30 Published:2008-07-30


Vegetation monitoring is important. This paper introduces a trend analysis method for vegetation sciences. A unit of homogeneous vegetation can be treated as a point, so it can have dynamic analysis. However, to carry enough information, this point has to be put into multi-variable space. Vegetation can be expressed as a position vector in multidimensional space. Vegetation is a resource competing system. All plant species compete for limited resources, and demands of all species can not exceed available resources. This can be expressed as the sum of the squares of all species equals one. Therefore, all the complementary plant species can be treated as mutually orthogonal. When using position vectors to represent vegetation, the magnitudes of the vectors carry information of the total biomass, while the directions of the vectors carry information of composition of vegetation; thus, the position vectors have to be standardized. Vegetation growth based on cell duplication is expressed as exponential growth. Changing trend is defined as present state over the past. The trend can be used to monitor the vegetation changes and to predict future states. Kalman filter is used to increase accuracy and lower monitoring cost.

Key words: vegetation monitoring, trend analysis, time series, multi-dimensional sphere model (MDSM), Kalman filter