In order to enhance the spectral characteristics of features for clustering, in the experiment of wetland extraction in
Sanjiang Plain, we use a series of approaches in preprocessing of the MODIS remote sensing data by considering
eliminating interference caused by other features. First, by analysis of the spectral characteristics of data, we choose a set
of multi-temporal and multi-spectral MODIS data in Sanjiang Plain for clustering. By building and applying mask, the
water areas and woodland vegetation can be eliminated from the image data. Second, by Enhanced Lee filtering and
Minimum Noise Fraction (MNF) transformation, the data can be denoised and the characteristics of wetland can be
enhanced obviously. After the preprocessing of data, the fuzzy c-means clustering algorithm optimized by particle
swarm algorithm (PSO-FCM) is utilized on the image data for the wetland extraction. The result of experiment shows
that the accuracy of wetland extraction by means of PSO-FCM algorithm is reasonable and effective.
Developments of raster data capture technologies and demands from application fields call for advanced raster data
analysis methods. Visual data mining that involves human's visual analytical capability in data analysis attracts attention
in recent years. Raster datasets usually have large amount of pixels, which may cause serious clotting problem in
visualization and thus challenges visual data mining. The research reported here mainly focuses on this problem and tries
to construct a hierarchical framework for visual data mining of raster data. In the hierarchical structure, the first level
uses volume rendering to visualize the whole raster dataset in attribute space, which can greatly reduce the impact of
clotting. To avoid the loss of subtle patterns, the second level makes use of parallel coordinates plot to reflect detailed
attribute information. This hierarchical structure ensures that both global and local patterns embedded in data can be
detected. In both levels, visualizations of attribute space are linked with that of geographic space. Software prototype
was developed and then applied to find small clusters that may relate to possible soil types. Case study result
demonstrated the effectiveness of this proposed approach.
As one of key factors which control the spatial soil variation in soil-landscape model, terrain information includes not only topographic attributes (such as slope gradient, curvature, etc.) but also information of slope positions. But the spatial gradation of slope positions is still not quantitatively considered in current predictive soil mapping and other related application areas. The issue of this paper is to make a primary discussion on the potential role of spatial gradation of slope positions in soil-landscape model. Taking a study area in Northeast China and a detailed taxonomy of slope position, this paper firstly utilized a fuzzy inference approach based on similarity to the typical locations to quantify the
spatial gradation between slope positions. Secondly, we took use of the soil-subgroup map in study area to analyze whether there is an evident relationship between soil distribution and spatial gradation of slope positions or not, by means of the statistics on the percentage of slope positions on where each soil subgroup in study area is distributed. The results show that the distributions between slope positions and soil subgroups are clearly correlative in the areas where fuzziness as one certain slope position is little. And the soil distribution shows obvious intergradation and uncertainty in
areas where the slope position is identified with much ambiguity. The evident relationship between soil distribution and spatial gradation of slope positions indicates that the quantitative information of spatial gradation of slope positions should be included into both soil-landscape model and its applications (e.g., digital soil mapping, etc.), for a better depiction on the co-variation between slope positions and soil type.
Conference Committee Involvement (1)
International Conference on Earth Observation Data Processing and Analysis
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.