MULTIVARIATE ANALYSIS USING GEOSTATISTICAL METHOD FOR LATERITIC NICKEL DEPOSIT IN GEE ISLAND, EAST HALMAHERA, NORTH MALUKU

• Main Author: SURYONO (NIM : 22107013), DWI
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/20046
• Institution: Institut Teknologi Bandung

Gee Island is one of mining concession area belongs to PT. Aneka Tambang, Tbk. which is administratively located in East Halmahera, North Maluku Province with <br /> <br /> <br /> area of 200 Ha. The methodology of research are conducted by literature study using secondary data of drilling exploration from PT. Aneka Tambang, Tbk. Data assay of lateritic nickel deposit is divided into three zones, i.e.: Limonite Ore Zone (LOZ), Low Saprolitic Ore Zone (LSOZ) and High Saprolitic Ore Zone (HSOZ) located in North Block and South Block. The variables used in modelling are Ni, Fe, Co, SiO2, CaO and MgO grades, and geometrical variables such as thickness, topographical slope, and Ni grade accumulation. Modelling is performed using multivariable geostatistical method. The data analysis involves statistical analysis, variogram and cross-variogram analysis, and geostatistical estimation by comparing the result of estimation and estimation variance of ordinary kriging (OK) and ordinary cokriging (COK). Structural analysis performed for multivariate data results the strong spatial correlation if the coefficient is > 0.75 and < -0.75, and weak spatial correlation if the coefficient is < 0.75 and > -0.75. Those coefficients are obtained from the spatial structures with proportional and non-proportional variogram and crossvariogram, cross-variogram with pure nugget fluctuated around 0 (null), and cross-variogram with pure nugget not fluctuated around 0 (null). The multivariate data with strong correlation either statistically or spatially at micro scale with range 0-25 m or spatially at local scale with range 25-100 m is generally found for the data pairs of Fe - MgO, Fe - SiO2, and Ni grade accumulation - thickness at all <br /> <br /> <br /> of lateritic zones both in North and South Blocks. <br /> <br /> <br /> Comparison of kriging and cokriging estimation variance for the variogram and cross-variogram with proportional and non-proportional structures showed that cokriging variance tends to be smaller than kriging variance if the nugget effect of cross-variogram is smaller than the nugget effect of variogram of primary variable. Configuration of data also have an effect on the estimation variance, where the more dense data (0-25 m) results the kriging variance smaller than cokriging variance, while the more wide apart of data (25-100 m) results the cokriging variance smaller than kriging variance. In the case of cross-variogram with pure nugget fluctuated around 0 (null), the kriging variance is equal to the cokriging variance, while in the case of cross-variogram with pure nugget not fluctuated around 0 (null), the kriging variance is always smaller than cokriging variance.