# Statistical interpretations of stream sediment geochemical data of Kabupaten Yapen Waropen, Irian Jaya, Indonesia and the Shag catchment, Otago, New Zealand, and modern models for threshold determination

### Year:

### Project type:

### Subject:

### Advisers:

### Abstract:

The thesis deals with the statistical interpretations of stream sediment geochemical data to show the potential areas for exploration of precious and base metals in Kabupaten Yapen Waropen of the Irian Jaya Province of Indonesia, and the •u.u ..... v .. of arsenic as an indicator element for detecting gold-tungsten mineralisation in Macraes area and the surrounding t. Apart from the statistical interpretations, this thesis introduces two new statistical models for threshold determinations. first method is a univariate method and was initially developed in this study. The second method is the application of the Jviahalanobis distance for a multivariate threshold determination.

The anomalous samples of Kabupaten Waropen have been determined by setting the threshold value equal to the mean plus ··two standard deviations. The statistical interpretations of the data indicate that the eastern part of Yapen Island is favorable for follow up exploration because most of the anomalies (Au, Ag, Cu, Pb, Zn, As and Bi) are grouped on the eastern side of the island. The anomalies are consistent with the presence of the Jobi Ophiolite Hydrothermal Breccia and caldera structure which are a good indicator for detecting epithermal ore deposits.

Three known elemental sources, gold-tungsten mineralisation in Macraes area, the Haast Schist bedrock and volcanic rocks of the Pigroot area, were used to study the downstream distribution patterns of arsenic and other associated elements in Deepdell Creek and Shag River. The study has established backgrounds of the elements, measuring different 'dimensions' of the elemental ·variation and elucidating the relationship between elemental variables in multivariate data using the principal component and canonical discriminant function analysis theories as well as examination of the downstream elemental mobilisation and ·correlation matrix to identify the interelement relationship. The downstream spatial trends of arsenic content in stream sediments and water of Oeepdell Creek are negatively correlated. This is a useful prospecting tool for gold-tungsten mineralisation. The ·principal component analysis indicates that the anomaly index model for detecting gold-tungsten mineralisation can be written as

GTM = {(0.87 As + 0.13 Cr + 0.13 Zn + 0.12 V+ 0.11 Ni + 0. 01 La) - (0.58 Cu + 0.15 Ba + 0. 12 Nb + 0.08 Se+ 0.03 Nd)}

where As, Cr, ..... , Nd here represent the concentration values from a set of data after they have been standardised to have zero means and unit standard deviations.

The GTM model is consistent with the observed geochemistry that As-bearing fluids will lead to the precipitation of goldtungsten minerals with depletion of Ba in the schist host rock.

The exploration implication of the GTM model is that, stream sediment samples that are close to or exceed the unity of GTM and contain multiple anomalies of As, Cr, Zn, Ni and La are useful properties for detecting gold-tungsten mineralisation. Arsenic is the real indicator for detecting gold-tungsten mineralisation as shown by positive high coefficient (0.87) of the GTM model, the elements are incidental.

The relative geochemical difference between Deepdell Creek and Shag River is shown in the canonical discriminant function analysis mainly by the ratio of As to Ba. High ratio of As to Ba in stream sediments of Deepdell Creek is a good indicator for detecting gold-tungsten mineralisation.

The different types of the statistical analyses that have been carried out to investigate different aspects of the same stream sediment geochemical data from the Shag catchment indicate that arsenic is the best pathfinder element for detecting goldtungsten mineralisation in stream sediment surveys in Macraes area and the surrounding district.

The new univariate statistical model for the threshold determination can be written as

GTh = exp ( z + 21 Y2 ) + 2- -J (cxp (2z + Y2 ){ exp (Y 2 ) - 1})

where z and Y here represent the pooled estimate of the sample mean and standard deviation of log normal distribution.

The initial application of the GTh model to the stream sediment geochemical data from Deepdell Creek, upper and lower Shag River, considered in this thesis, has yielded useful results. The study has showed that the GTh model is better than the arithmetic mean plus two standard deviations for establishing the geochemical threshold of log normal distribution. The comparative study was investigated by using statistical tests of significance of mean values in terms of the thresholds. The thresholds calculated by the two methods, were based on the sample randomisation and with respect to the normal and log normal distributions of the stream sediment geochemical data from Deepdell Creek, upper and lower Shag River.

The GTh model is comparable to the arithmetic mean plus two standard deviations, the two models always give similar threshold result for a normal distributed data. Control on the normality of a set of data can be investigated by the equality of threshold values produced by the two models.

The GTh model would give a precise threshold value compared to the arithmetic mean plus two standard deviations for a log normal distributed data because it uses information on the mean and standard deviation of log normal behaviour.

The GTh model is an alternative way of establishing the geochemical threshold for normal or log normal distributed data because it is robust with respect to normal or log normal distribution.

Mahalanobis distance, a multivariate statistical method, has been suggested as a criterion for detecting multivariate outliers. This method can be used for oG.tlining stream sediment geochemical anomalous data in mineral exploration. The main strength of Mahalanobis distance analysis lies on its simplicity to describe the data in terms of anomalous indices and anomalous variables. The initial application of Mahalanobis distance to Yapen, Miosnum and Ambai Islands stream sediment geochemical data considered in this study has yielded useful qualitative and quantitative results, but its significance should be further justified by geologists.

The anomalous samples of Kabupaten Waropen have been determined by setting the threshold value equal to the mean plus ··two standard deviations. The statistical interpretations of the data indicate that the eastern part of Yapen Island is favorable for follow up exploration because most of the anomalies (Au, Ag, Cu, Pb, Zn, As and Bi) are grouped on the eastern side of the island. The anomalies are consistent with the presence of the Jobi Ophiolite Hydrothermal Breccia and caldera structure which are a good indicator for detecting epithermal ore deposits.

Three known elemental sources, gold-tungsten mineralisation in Macraes area, the Haast Schist bedrock and volcanic rocks of the Pigroot area, were used to study the downstream distribution patterns of arsenic and other associated elements in Deepdell Creek and Shag River. The study has established backgrounds of the elements, measuring different 'dimensions' of the elemental ·variation and elucidating the relationship between elemental variables in multivariate data using the principal component and canonical discriminant function analysis theories as well as examination of the downstream elemental mobilisation and ·correlation matrix to identify the interelement relationship. The downstream spatial trends of arsenic content in stream sediments and water of Oeepdell Creek are negatively correlated. This is a useful prospecting tool for gold-tungsten mineralisation. The ·principal component analysis indicates that the anomaly index model for detecting gold-tungsten mineralisation can be written as

GTM = {(0.87 As + 0.13 Cr + 0.13 Zn + 0.12 V+ 0.11 Ni + 0. 01 La) - (0.58 Cu + 0.15 Ba + 0. 12 Nb + 0.08 Se+ 0.03 Nd)}

where As, Cr, ..... , Nd here represent the concentration values from a set of data after they have been standardised to have zero means and unit standard deviations.

The GTM model is consistent with the observed geochemistry that As-bearing fluids will lead to the precipitation of goldtungsten minerals with depletion of Ba in the schist host rock.

The exploration implication of the GTM model is that, stream sediment samples that are close to or exceed the unity of GTM and contain multiple anomalies of As, Cr, Zn, Ni and La are useful properties for detecting gold-tungsten mineralisation. Arsenic is the real indicator for detecting gold-tungsten mineralisation as shown by positive high coefficient (0.87) of the GTM model, the elements are incidental.

The relative geochemical difference between Deepdell Creek and Shag River is shown in the canonical discriminant function analysis mainly by the ratio of As to Ba. High ratio of As to Ba in stream sediments of Deepdell Creek is a good indicator for detecting gold-tungsten mineralisation.

The different types of the statistical analyses that have been carried out to investigate different aspects of the same stream sediment geochemical data from the Shag catchment indicate that arsenic is the best pathfinder element for detecting goldtungsten mineralisation in stream sediment surveys in Macraes area and the surrounding district.

The new univariate statistical model for the threshold determination can be written as

GTh = exp ( z + 21 Y2 ) + 2- -J (cxp (2z + Y2 ){ exp (Y 2 ) - 1})

where z and Y here represent the pooled estimate of the sample mean and standard deviation of log normal distribution.

The initial application of the GTh model to the stream sediment geochemical data from Deepdell Creek, upper and lower Shag River, considered in this thesis, has yielded useful results. The study has showed that the GTh model is better than the arithmetic mean plus two standard deviations for establishing the geochemical threshold of log normal distribution. The comparative study was investigated by using statistical tests of significance of mean values in terms of the thresholds. The thresholds calculated by the two methods, were based on the sample randomisation and with respect to the normal and log normal distributions of the stream sediment geochemical data from Deepdell Creek, upper and lower Shag River.

The GTh model is comparable to the arithmetic mean plus two standard deviations, the two models always give similar threshold result for a normal distributed data. Control on the normality of a set of data can be investigated by the equality of threshold values produced by the two models.

The GTh model would give a precise threshold value compared to the arithmetic mean plus two standard deviations for a log normal distributed data because it uses information on the mean and standard deviation of log normal behaviour.

The GTh model is an alternative way of establishing the geochemical threshold for normal or log normal distributed data because it is robust with respect to normal or log normal distribution.

Mahalanobis distance, a multivariate statistical method, has been suggested as a criterion for detecting multivariate outliers. This method can be used for oG.tlining stream sediment geochemical anomalous data in mineral exploration. The main strength of Mahalanobis distance analysis lies on its simplicity to describe the data in terms of anomalous indices and anomalous variables. The initial application of Mahalanobis distance to Yapen, Miosnum and Ambai Islands stream sediment geochemical data considered in this study has yielded useful qualitative and quantitative results, but its significance should be further justified by geologists.

### Thesis description:

viii, 168 p. : col. ill., maps ; 30 cm.

### Department:

### OU geology Identifier:

1992Karubaba

### OURArchive handle:

### OURArchive access level:

### Collection

### Citation

Karubaba, Jannes Johan John, 1953-, “Statistical interpretations of stream sediment geochemical data of Kabupaten Yapen Waropen, Irian Jaya, Indonesia and the Shag catchment, Otago, New Zealand, and modern models for threshold determination,”

*Otago Geology Theses*, accessed May 22, 2024, https://theses.otagogeology.org.nz/items/show/261.