| Ecological environment of agriculture and forestry (EEAF) provides safe products and high quality lives. The precise and digital management lays on the achievements of accurate detect, dynamic monitoring and scientific quality evaluation on EEAF by scientific monitoring and accurate observation, which are also the crucial issue to layout and control EEAF effectively. EEAF behaves obviously a spatial heterogeneity which involves a lot of data in various formats. Although great amounts of highly-resoluted basic raster and vector datasets of different scaled regions are provided by 3S, the sampling sites are quite limited. The analysis of EEAF status on various scales by limited sampling sites together with images and shapes is crucial to deploy monitoring and controlling. Geostatistics concludes an approach that estimates the values at unknown areas by sampling sites and achieve the transformation from points to the whole area. GIS integrated by spatial analysis, geostatistics and spatial database is an effective tool to analyze EEAF factors on various scales and the best platform to visualize the results of evaluation on EEAF.In this article, analysis and application researches on three scales of EEAF by advanced spatial database and GIS network, with fundamentals on geostatistics and spatial analysis, are discussed. As many changeable factors and spatial features were involved, spatial database with a massive capacity worked as the core to process, transform, analyze and save the huge amounts of data. GIS network made data accessible to different software to analyze and process. According to different objects, different methods of interpolation were adopted respectively: Inversed Distance Weight (IDW) was chosen when a general tendency was known; Universal Factor Kriging (UFK) was chosen when the variation was compound; Universal Regression Kriging (URK) was chosen when the variable was affected by both spatial and non-spatial factors.On a small scale, Shanghai Xinghuo Farm was chosen as an instance, to investigate its pollution status on heavy metals and pesticides. As the sources of heavy metals were discerned and disheveled, IDW was chosen to estimate and show the influence of industrial pollutants to the fields; the pollutions of pesticides were caused by agricultural operations. They were visualized by Web-GIS based on spatial database with satisfactory.On a middle scale, Chongming Island was chosen as an example, for the nutrients in soil over there are affected by geographic factors and long-turn human activities. The data were firstly transformed in the database to make them obey normal distributions, with outliers quarantined, and observe the basic features of their statistical distributions. In UFK multi-scaled interpolation, the Principle of Nugget Effect Minimization was adopted to discuss the search radius. Each available experimental model thus evaluated by their accuracy respectively. By the best model, precise estimations were obtained, according to which individual and comprehensive fertilities were evaluated by Membership Function and plant adaptability of several plants were evaluated by Dynamic Weighted Functions. The results were supported by other archives about this area, while the global trends and local details are precisely drawn by the multi-scaled interpolation, more reasonable, accurate and credible in real application on fertilization guide.On a large scale, Europe was chosen as a research area, estimating the leaf nutrients influenced by the habitation of vegetations. URK was adopted to analyze the reasons of spatial heterogeneity and map the variation of leaf minerals. Besides commonly used climatic factors, landscape ones were introduced as well, for example Urban Radiation (UR) and Forest Shield (FS), in the regression. The result proved that UR and FS can improve the accuracy of estimation and facilitate the division of variations. The hypotheses of First Order Reaction Kinetics and Log-Normal Distribution were demonstrated by significance tests. Data are processed as the follows: log-transformation, normalization, linearization, orthogonalization, regression and residual interpolation. The results were improved to show the actual distribution of leaf minerals by this novel approach. The trends of estimations are supported by the conclusions of former reports. The results are more intuitionistic and elaborate to present the relations between leaf minerals and the related surroundings. It was obviously observed, for instance, that N deposition raises leaf N and P in the north of Central Europe and K, P are leached by high precipitation on the mountains, which could hardly been discerned by traditional methods.In summary, spatial analysis and interpolation to EEFA on different scales by geostatistical analysis based on GIS were discussed in this article, and the methods were proved to be effective in improving the accuracy and resolution. Once more checking data were available, further explorations could be done: more relative factors could be introduced in spatial relativity analysis to make the results more convincing. |
PDF Full Text Request