Process of constructing predictive model for producing mineral potential map
Producing a mineral potential map showing mineralised zones with corresponding probabilities is the most important step in any mineral exploration program. All the subsequent studies, detailed exploration budget and planning depends upon the produced mineral potential map. Therefore, It is utmost important that the mineral potential map should be made with careful and deliberate efforts. With the advent of fast computing and advancement in the development in mathematical modelling and machine learning methodologies, now, it is possible to build an effective mineral potential map. In this article, a process of building a mathematical model to produce mineral potential zones has been explained.
Process of building predictive model for mineral zones
It all start with a mineral exploration strategies that decide which mineral and area to focus on. For instance, copper exploration in the Aravalli system of rajasthan. After a selection of a study area, a geological knowledge about the area is developed through collecting informatio from published work, geological reports, professional experience and knowledge, and geological field work and data collection. An exploration database needs to be prepared to provide all basic information such as record of field evidences such as observed field evidences supporting hydrothermal deposits type. The next step is to produce a conceptual model of metallogenesis or mineral deposits. There are many known models of metallogenesis such as porphyry type, supergene etc. In this step, it is identified that why and how? Mineralization took place in the area. For example, stratigraphic control, stratiform to strata-bound SEDEX type sedimentary hosted ore deposits of Rampura-Agucha or Stat-abound sedimentary hosted copper ore deposits of Copperbelt region of Zambia. Preparing a metallogenesis model is so important as it lets you develop recognition criteria for mineral deposits. Also this metallogenesis models tells what evidence to look for and what data to consider. After this step, recognition criteria are developed. For instance, the recognition criteria for strata-bound, sedimentary hosted ore deposits is as follows:
- Association with a particular type of lithologies.
- Association with a particular type of stratigraphic group such as lower roan group in Zambia or Dharwar group in India.
- Association with sedimentary environment.
- Association with magmatic volcanic rocks.
- Association with structural features such as dyke, fault systems, shear zones etc.
Once the recognition criteria has been decided, the next step is to consider list of databases to consider. For example, the structural features can be extracted from remote sensing images, aeromagnetic maps in the form of various derivative maps, Euler solutions or inversion results., structural geological maps also provide information that could be combined with the results of aeromagnetic data. Ground magnetic can also be used as it provides more detailed structural information that could be vital in improving the productivity of model. Similarly, detailed lithological and stratigraphic map provide lithological and structural information. Now a days geochemistry is also very popular in providing the lithological and other vital information, therefore it could also be considered as an input. A Schematic diagram for constructing a predictive mathematical or machine learning model is shown in the figure 2.
