Weights Of Evidence Regression Model Based On ArcGIS And Python And Application On Seafloor Resource Assessment

The paper and study are parts of the"eleventh five-year"project"Processing Analysis Methods and System of Ocean Resources Environment Data and Information"(DYXM115-03-3-05)conducted by China Ocean Mineral Resources R&D Association. The main task objectives are as follows: analysis technology of spatial data and information based on Python and ArcGIS further development is discussed; the theory and method of weights of evidence and regression model is presented and tentatively applied to assessment of ocean mineral resources; integrated development technology of self-determined professional software modules based on GIS and Python is discussed; finally the methods are used in assessment of ocean mineral resources and an exemplary production is gained.Mineral resources assessment is a specific application of spatial data management and analysis. The application of GIS, the most powerful spatial data management and analysis technique, in mineral resources predictions, would greatly enhance the efficiency and accuracy of such assessments. ArcGIS, a powerful and easy to operate GIS software used in this paper can meet different needs of different users. And further Python is the first choice for Geoprocessing in ArcGIS, especially the version 9.2.This paper tends to analyze the data of polymetallic nodules about the east Pacific CC area, based on mathematical geology methods, combined with GIS technique and Python language, and to forecast the potential resources of polymetallic nodules through building a mathematical model of the study area.In the area of Survey of land mineral resources, the application of ore-forming geological prediction theory and mathematical methods to study the regional metallogenic regularity and favorable conditions for mineralization, the establishment of deposit prediction model, delineated mineralization prospect area, has always been an important metallogenic prediction ideas. Prediction and evaluation of marine mineral resources work has its own particularity Compared with land-based work. Mature theory of land-based mineral resources and technical ability to borrow to the ocean? How effective is their application? Methods need to be adjusted? These are the very Interested subject .There are not so much people to do enough work to answer these questions.The weight of evidence Model in terms of resources evaluation has been targeted forecasting methods.In this paper, linear regression model with the combination of the right to present evidence regression model, the model is a quantitative prediction method of applying the potential of mineral resources in the seabed manganese forecast work, achieved good results.The main research ideas for this paper are as follows:①Analysis of the characteristics of the target resource evaluation;②analysis of data characteristics;③the use of ArcGIS spatial data processing and management capabilities for data processing;④u sing ArcGIS spatial analysis function of the characteristics of evidence;⑤to build a weights of evidence regression model;⑥to develop a professional model using Python;⑦make use of the known distribution of resources as a training area, establish a prediction and evaluation resources model;⑧make characteristic layer of the forecast area into the evaluation model of resources, access to results;⑨using ArcGIS graphic input module to output the results;⑩to predict the reliability of the evaluation. Based on the above steps making use of ArcGIS and Python Integrated Development Environment to develop an information system which separated from the specific mandate.The main research methods of this paper are as follows: First, the distribution of ocean polymetallic nodule resources and their forming-controlling factors are analyzed based on previous studies. Due to the huge amount of data involved in mineral predications, ArcGIS and Python technique are used to develop needed software. And then evidence weight model, only applicable to location predication is mended to be more suitable to predict potential polymetallic nodule resources.Second, the collected data is processed and analyzed. The huge amount of data, with inconsistent accuracy, involves two-dimensional and multi-disciplinary information of East Pacific Ocean CC Area, with an area of ~14 million km2. As a result, fixed-grid methods are adopted to assign values of the units based on the powerful data processing and management module of ArcGIS. Third, Use GIS spatial analysis module to extract spatial characteristics of the grid cell assignment. Here the main use are statistical analysis of the grid unit and the logic layer method of calculation to predict the grid for the assignment, to obtain the necessary resources of calculation process of evaluating the characteristics of the data layer. Fourth, to build the Weights of evidence regression model. First of all, in the training area of the distribution of polymetallic nodule resources, according to the size of the volume of resources from small to large, system resources are divided into 5 categories by 5 points, each category of evidence to the right of access to five, according to the right to verify the evidence; and then five layers with evidence of the right resources to carry out linear regression calculated to obtain the right training zone regression model evidence; the final promotion of the model to predict the forecast area. Fifth, the results of the output. Use ArcGIS graphics, text, tables and other data output module to output the results. Sixth, the system integration. Return the mathematical model by Python programming language and integrated into ArcGIS software as its toolbox.The main results of this Paper are as follows: (1) explored the integration of Python and ArcGIS environment resource assessment system technology, which is the first time in China. (2) in the weight of evidence and the basis of linear regression of the right to build a regression model of the evidence and applied successfully to the ocean for polymetallic nodules in the evaluation of potential resources. (3) Python, ArcGIS and evidence the right to return to the organic integration of algorithms, developed a preliminary resource evaluation software. (4) estimated the CC zone the abundance of polymetallic nodules in the distribution of resources, in addition to and have found general agreement, but also predicted the potential of resources to find areas of polymetallic nodules.