The Building Of Forest Data Warehouse Based On Survey Data

With the rapid development of "digital forest", people have acquired a great deal of datum about forestry resources. Currently it has become a problem about how to obtain useful information and knowledge from those data and how to serve well to the forest management. Data warehouse and data mining can extract information from datum and then instruct forest production.According to the forestry governor's demand, this paper analyzed the characteristics of investigated forestry information and designed the forestry data warehouse by using concept model, logic model and physical model. In the concept model designing, this paper analyzed the manager's requirement and sequentially made information packets. This paper also used OLAP to design logic model and obtained dual granularity. RAID (Redundant Array of Inexpensive Disk) and BitMap index are also used to optimize the system .In physical model design, site quality evaluation model and predicted fir storage model in Fujian are presented in the data mining.Rough Set Theory was used in site quality evaluation, with the significances of attribution being used as factor of weight evaluation. We applied the total score principle to specify different appraisal standards. In order to solve the problem of slow speed when BP neural network is gaining knowledge in large sample, the model of fir storage in Fujian was designed with fuzzy Neural Network method and were programmed in three BP algorithm such as Bayesian Regularization method , Gradient Descent with Momentum Algorithm and Levenberg-Marquardt Optimization Algorithm by MATLAB so it can select partial sample from the original samples as new samples for study. This model consists of tree-height index, diameter and density. After many experiments and comparisons, Bayesian Regularization method is confirmed as the model for fir storage of prediction in Fujian province, with 15 as the node in middle transparent layer.At last, this paper proposed the Structure of forest data warehouse based on...