Sedimentary Geology, 61 (1989) 135-150
135
Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
Textural procedures for the environmental discrimination of late Neogene coastal sand deposits, southwest Auckland, New Zealand S T E P H E N STOKES, C A M P B E L L S. N E L S O N and T E R R Y R. H E A L Y Department of Earth Sciences, University of Waikato, Private Bag, Hamilton 2001 (New Zealand) Received April 25, 1988; revised version accepted July 27. 1988
Abstract Stokes, S., Nelson, C.S. and Healy, T.R., 1989. Textural procedures for the environmental discrimination of late Neogene coastal sand deposits, southwest Auckland, New Zealand. Sediment. Geol., 61: 135-150. The coastal region of much of the northwestern North Island, New Zealand consists of late Neogene, poorly consolidated, coastal sand and tephra deposits of the Kaihu Group. Determination of the environments of formation of these deposits using standard grain size analytical approaches has met with limited success, both in this and previous studies. However, here we demonstrate that samples from known beach, dune and tephra facies can be differentiated in at least 90% of cases using stepwise and canonical discriminant analyses of the textural data. Classification models derived from the canonical analyses are used to distinguish the depositional facies of homogenised, subsurface drillhole samples collected at Taharoa, the site of a large, currently mined ironsand deposit in southwest Auckland. A SEM, process-based, interpretation of the quartz grain surface features in the sand deposits shows general agreement with the environments determined from the discriminant function analyses despite a considerable degree of post-depositional alteration.
Introduction
Numerous sedimentologic studies have endeavoured to use the size distributions of sedimentary particles to differentiate depositional environments (e.g., Mason and Folk, 1958; Klovan, 1966; Hodgson and Scott, 1970; Samsuddin, 1986). Most early workers attempted to distinguish depositional environments using bivariate plots of grain size statistics (e.g., Friedman, 1961; Moiola and Weiser, 1968), or by further combining the statistics to determine standardised indices considered to be environmentally unique (e.g., Sahu, 1964; Sevon, 1966). However, as Taira and Scholle (1979) have pointed out, these early studies did not fully consider the many location specific factors of deposition inherent in any sedimentary setting, including the integrated effects of previously active 0037-0738/89/$03.50
© 1989 Elsevier Science Publishers B.V.
processes that the sediments may have undergone and the controls on the availability of sediment to an environment. More recently, multivariate discriminant function analyses have been used in environmental discrimination (e.g., Moiola et al., 1974; Taira and Scholle, 1979; El-Ella and Coleman, 1985), as well as in many other geologic applications (e.g., King et al., 1982). To our knowledge, with the exception of an analysis of the Navajo Sandstone (Taira and Scholle, 1979), discriminant function procedures using sediment textural data have been applied solely to modern depositional settings. Like grain size analysis, the surface textural analysis of detrital quartz grains has also been used extensively as a means of identifying environments of deposition (e.g., Krinsley and Donahue, 1968; Rex et al., 1970; Coch and Krins-
136 • : ~ ~!:,.~.i ~I:; :.~!:~i,i ~ : :': :. ,"
~ ~ ':.~."
~' ; ~
.
.: ,
.'i.: ~:~:~/:~ .:..:~:i: ~.i
,',i'
.:::,.'. , " ;~
~ :...: ;i~:,~;:,
.,'.'!' ':?~ : ~
"
's...:i
.i..::.
.;':~. "i'. i i i : : : " :
© "...
Taharoa
". :
Mitlwat Stm -..:...1.: " L. Taharoa
Wainui Stm..-..~_.~.;.. - ..:
Mine head quarters
ptti *so
175°E
o80
N
-.
35 ° SNORTH ISLAND
~
Mesozoic basement rocks
[-~]
Swamp
'. Taharo Deposi
'~20.~ Contours (m)
-L.Rototapu '
o60 Drill hole (depth m) 0
1
I
I
•
. :?/i(.):i'i!4
~
"fasrBan
Sea
k~m Fig. 1. Locality map of the Taharoa sand deposit in North Island, New Zealand and the extent of the Kaihu Group (black in inset). Also shown are the topographic contours on the sand deposit, and the locations and depths to the bottom of drillholes from which homogenised samples were selected for analysis.
ley, 1971; Margolis and Kennett, 1971; Bull and Culver, 1979; Georgiev and Stoffers, 1980; Bull et al., 1986; Sharp and Gomez, 1986). Earlier approaches suggested the presence or absence of one or two textural features as being environmentally diagnostic (e.g., Krinsley and Donahue, 1968;
Margolis and Krinsley, 1974). This approach has been superseded by methods which quantify the abundances and relative chronology of numerous surface textural features (e.g., Cater, 1984). With this information, the process-related significance of the surface textures can be inferred, and the
137
environment(s) of deposition may be suggested on the basis of the dominant process(es) recognised (e.g., Al-Saleh and Khalaf, 1982; Cater, 1984). As part of a study to ascertain the subsurface extent of iron-bearing sands within late Neogene (taken to include Pliocene and Quaternary; Jenkins et al., 1985) coastal sand deposits at Taharoa, southwest Auckland, New Zealand (Fig. 1), an attempt has been made to determine the environments of deposition of drillhole samples homogenised by the drilling technique employed. Devoid of sedimentary structures, and generally unfossiliferous, textural criteria provided the only feasible basis for environmental discrimination. A dual approach was adopted, using multivariate discriminant functions of grain size data and information on the surface textures of detrital quartz grains.
The Taharoa sand deposit The Taharoa sand body (174°65'E, 38°20'S) occurs on the west coast of the North Island of New Zealand (Fig. 1) and is currently mined for titanomagnetite for use in steel production (Stokes et al., 1989). The deposit is 8 km long and 5 km wide, reaches a maximum depth of up to 120 m below sea level (Lawton, 1979), and is bounded to the north, south and east by Mesozoic greywacke basement rocks, and to the west by the Tasman Sea. At the surface the sand body comprises an extensive coastal dunefield, presently active dune types including barchan, parabolic and clifftop varieties (Pain, 1976). A series of lakes, formed in response to the impounding of water by encroaching sand dunes, occur along parts of the eastern margin of the deposit (Fig. 1). The deposits forming the Taharoa sand body belong in the late Neogene (Pliocene-Holocene) Kaihu Group (Fig. 2), which comprises mainly dune, beach and estuarine sedimentary facies occupying much of the coastal strip of northwestern North Island, New Zealand (see Fig. 1 inset) (Chappell, 1970), together with intercalated but subordinate tephric and lignite deposits. Last Glacial and Holocene sand units up to 80 m in thickness drape much of the "surface" at Taharoa, and are the currently mined portion of the de-
Group
Formation Mitiwai
Age Holocene
Waiau B K
Waiau A
A
Parawai
I
Nihinihi
H
Awhitu U
Ohuka
Late Piiocene
Kaawa
Fig. 2. Stratigraphic nomenclature and age of the Kaihu Group in the Southwest Auckland region of New Zealand, based on Kear (1965), Chappell (1970), Mildenhall (1975) and Pain (1976).
posit. The stratigraphy of deeper horizons at Taharoa is based solely on drillhole samples collected down to 90 m below sea level. Stokes (1987) has shown the sequence consists of up to eight unconformity-bound sand units, each of highly variable thickness (0 60 m), and separated by muddy, often carbonaceous interbeds of probable pedogenetic origin. The sand units correspond to formations within the Kaihu Group (Fig. 2).
Methods of study
Textural analysis of grain size data Field and laboratory procedures Drillholes are periodically sunk into the Taharoa sand deposit, using a reverse circulation technique, as part of on-going exploration and proving of reserves. This type of drilling produces bulk samples representative of 3 m downcore intervals. The 166 subsurface samples used in this study were split from these bulk samples. To set up the discriminant functions (see later) additional samples from known depositional environments were required. These comprised 26 field samples collected at or near Taharoa, together with 35 modern beach and dune samples from the same area analysed by Christie (1975). Details of the geographic and stratigraphic location of all sam-
138 ples are given in Stokes (1987) and are available upon request. Bulk sediment samples were cleaned in oxalic acid (Carver, 1971), wet sieved over a 4~ screen. and the percentages of sand, silt and clay determined following drying of the sand fraction ( < 4~) and pipette analysis (Folk, 1968) of the mud ( > 4~) fraction. Sand fractions were then dry sieved at 1 / 4 ~ intervals for seven minutes on a Rotap Sieve Shaker and both graphic and moment grain size parameters calculated following Folk (1966. 1968).
Discriminant function analysis Discriminant analysis is a multivariate classifying technique which differentiates distinct populations by combining a number of independent attributes (variables), maximising the ratio of the differences in means between groups to the variances within the groups (Dunteman, 1984). A priori knowledge of groupings (or depositional environments) for a set (or sets) of samples is required to develop such a discriminant function (a reference set). Subsequently, the derived function(s) can be used to assign an unknown set of samples to one of the discriminated populations. Mathematical details of the technique are given by Hand (1967), Afifi and Azen (1972) and Dunteman (1984), among others. Where the discriminant function analysis is one-dimensional (reference sets A and B in this study--see below), involving the separation of two groups only, a discriminant index, R 0 - - t h e point about which the groups can be separated - - m a y be calculated (see Hand, 1967). For the separation of three or more groups, discriminant indices must be defined by continuous functions, describing discriminant field boundaries in multidimensional space (reference set C in this study being two-dimensional--see below). The textural data for 61 samples from known dune, beach and tephra deposits within the Kaihu Group provide the database of samples from which three reference sets (A-C) were derived: reference set A (n = 35)--modern dune and beach samples analysed by Christie (1975);
reference set B (n = 54)--reference set A plus additional beach and dune samples collected in this study: reference set C (n = 6 1 ) - - r e f e r e n c e set B plus samples of exposed tephra deposits. These reference set samples were obtained from within the same group of sediments as the drillhole samples so as to minimise grain size variability induced by differing sedimentation histories of reference and unknown sample sets. Analysis of the reference sets results in the formulation of a number of discrimination rules by which the unknown set (drillhole) samples can be classified. The approach adopted in the selection of variables for inclusion in the discriminant functions follows that used by Taira and Scholle (1979) whereby all moment and graphic sedimentary statistics, some combinations of the statistics, and the character of the grain size distribution curves (both arithmetic and probability scaled) are initially used (Table 1). In addition, a principal components analysis was performed on reference sets A, B and C, the first 15 principal component scores of the samples being used as variables. Evaluation of the discrimination efficiency of individual variables within the discriminant model is indicated by the squared partial correlation criteria (R2), which measures the explanatory power of a variable in combination with other variables (Joyner, 1985). A second measure, the Wilks' Lambda Selection Criterion, is used as a measure of overall multivariate discrimination efficiency, lower values indicating better discrimination (Klecka, 1975). Two computer programs were used to determine the discriminant functions for the reference set samples. STEPDISC (Joyner, 1985), a stepwise discriminant analysis, was used to evaluate the discrimination efficiency of the reference set variables. The reference set data were then further analysed using only those variables that proved to be efficient discriminators. CANDISC (Joyner. 1985), a canonical discriminant function procedure, was then used to determine canonical discriminant coefficients, to which variable scores are combined and summed, hence numerically classifying depositional environments. A further program, written by the
139 TABLE 1
Quartz grain surface textural analysis
Grain size variables initially used in the discriminant analyses
Several ( 5 - 6 ) q u a r t z g r a i n s f r o m the 2 - 3 0 size
Xl = graphic mean X2 = median X3 - first percentile X4 D~5 X5 - D65 X6 - graphic sorting X7 = graphic skewness X8 = moment mean grain size X9 = moment sorting XI0 = moment kurtosis XI 1 = grain size range X12 = X2/Xll Xl3 - X1/X6 X14 = XI/X7 X15 = X1/X8 X] 6 = X6/X7 X17- X6/X8 X]8 - X7/X8 X19 = XS/X9 X20 - XS/XI0 X21 - moment skewness X22 = XS/X21 X23 = minimum grain size X24 = maximum grain size X25 = slope of curve between Do and D35 (arithmetic scale) X26 = slope of curve between D35 and Dso (arithmetic scale) X27 = slope of curve between Ds0 and D~5 (arithmetic scale) X28 = slope of curve between D~,s and Du~/ (arithmetic scale) X29 = slope of curve between D0 and DI~ (probability scale) XI0 = slope of curve between D35 and Ds0 (probability scale) X31 = slope of curve between Ds0 and D~5 (probability scale) X32 - slope of curve between D~,5 and D ~ (probability scale) X33 = kurtosis PRIN1 ) • principal components 1 to 15 PRIN15
f r a c t i o n of 24 s a m p l e s selected f r o m the grain size d a t a set were m o u n t e d o n t o 2.5 c m d i a m e t e r brass s t u b s u s i n g d o u b l e - s i d e d tape a n d o b s e r v e d u s i n g a JEOL JSM35 scanning electron microscope. The g r a i n s were p o s i t i v e l y i d e n t i f i e d as q u a r t z u s i n g a n attached
EDXA.
F o r each g r a i n the o c c u r r e n c e
a n d a b u n d a n c e of v a r i o u s m e c h a n i c a l a n d c h e m ical features ( T a b l e 2) were recorded, a n d p h o t o m i c r o g r a p h s p r e p a r e d of e x a m p l e s of the features. T h e a b u n d a n c e of each of the textures was n o t e d b y m a k i n g a visual e s t i m a t e of the p e r c e n t a g e of the grain surface area c o v e r e d b y the feature: a b u n d a n t ( > 50%); c o m m o n ( 2 0 - 5 0 % ) ; scarce ( < 20%); a b s e n t . F r o m this, a c h r o n o l o g i c s e q u e n c e of t r a n s p o r t , d e p o s i t i o n a l a n d d i a g e n e t i c processes was i n f e r r e d ( f r o m T a b l e 3) a n d s a m p l e d e p o s i t i o n a l e n v i r o n m e n t s suggested. T h e raw g r a i n size a n d surface textural d a t a f o r m i n g the basis of this p a p e r are a v a i l a b l e u p o n request. Results
Discrirninant function analysis Reference set analysis Bulk texture
of s a m p l e s
from
the reference
facies (beach, d u n e a n d tephric) d i s t i n g u i s h e d the beach/dune
s a m p l e s f r o m the t e p h r i c s a m p l e s o n
the basis of the s i g n i f i c a n t m u d (silt + clay) c o n tent of the latter, b u t failed to d i s c r i m i n a t e between the t o t a l l y s a n d - d o m i n a t e d b e a c h a n d d u n e s e d i m e n t s (Fig. 3A). A l s o the i n d i v i d u a l g r a i n size statistics for the r e f e r e n c e set s a m p l e s were n o t u n i q u e as to e n v i r o n m e n t (Fig. 3B); n e i t h e r were
s e n i o r a u t h o r , was used to c a l c u l a t e p r i n c i p a l c o m p o n e n t s a n d d i s c r i m i n a n t f u n c t i o n scores for d r i l l h o l e (unknown set) s a m p l e s , c o m b i n i n g selected v a r i a b l e s a n d their respective c a n o n i c a l coefficients. P r o j e c t i o n of s a m p l e d i s c r i m i n a n t f u n c t i o n scores o n t o d i s c r i m i n a n t f u n c t i o n axes d e t e r m i n e d for the reference s a m p l e set t h e n allows e n v i r o n m e n t a l classification of the d r i l l h o l e samples.
s t a n d a r d b i v a r i a t e p l o t s (Fig. 3C). Of the m a n y r e l a t i o n s h i p s e x a m i n e d , best s e p a r a t i o n o c c u r r e d o n the s o r t i n g versus m e a n grain size plot, especially b e t w e e n t e p h r i c a n d coastal ( b e a c h + d u n e ) samples, a l t h o u g h c o n s i d e r a b l e o v e r l a p of env i r o n m e n t s still o c c u r r e d . M u l t i v a r i a t e n u m e r i c a l analysis, u s i n g a disc r i m i n a n t f u n c t i o n a p p r o a c h , was t h e n u n d e r t a k e n o n the r e f e r e n c e d a t a b a s e s a m p l e s in three separate analyses. T w o of the a n a l y s e s c o n s i s t e d of o n l y b e a c h a n d d u n e s a m p l e s (i.e., l i n e a r d i s c r i m i -
140
TABLE 2 Classification of grain surface textural features after Pittman (1972), Krinsley and Doornkamp (1973), Higgs (1979) and Cater (1984) Feature
General description
1. Irregular pits 2. Conchoidal fractures 3. Stepped cleavage planes 4. Fracture plates 5. Parallel striations 6. Grinding features 7. V-shaped pits 8. Curved grooves 9. Meander ridges 10. Dish-shaped concavities 11. Upturned plates 12. Adhering particles 13. Breakage blocks 14. Triangular oriented etch pits 15. Grooves/anastomosing patterns 16. Surface solution/cracking 17. Dulled frosted surfaces 18. Silica pellicles/globules 19. Amorphous overgrowths 20. Euhedral overgrowths
Variety of surface depressions Clean conchoidal breakage Similar to 2 but parallel to curved Broken cleavage plates Parallel broken cleavage plates Clean imbricate grinding marks Triangular pits up to 5 # m wide Triangular pit associated scratches Intersections of conchoidal breaks Smooth curvilinear depressions Broken or precipitated parallel ridges, 0.5-10 # m long Attached fine material Irregular microblock textures Oriented triangular pits (dissolutional) Dissolutional depressions Variable dissolutional features Extensive surface smoothing (precipitated) Amorphous silica outgrowth (precipitated) Silica plastering features (precipitated) Crystalline silica outgrowth (precipitated)
nant analyses), and the third included tephric deposits. The three-fold approach was adopted in order to check the environmental classifications made (Beaudoin and King, 1986).
Table 4 lists the highly discriminating variables, and their associated statistics for each of the three reference sets. Principal component scores were consistently found to be efficient discriminators.
TABLE 3 The process-based grain surface textural classification system adopted for the Taharoa sand deposit (after Cater, 1984). Processes l - 8 concern pre-, post- or syn-depositional chemical activity, allowing inferences on secondary cycling of grains, and pre-depositional or pre-transportational histories; Processes 9-13 are more environment specific Processes
Texture or superposition of textures
Abrasion textures superimposed on solution features Abrasion both superimposed on, and superimposed by, solution Solution superimposed on abrasion textures Any solution feature without superposition Abrasion textures on precipitated silica Silica globules, silica flowers; precipitation before and after abrasion on the same grain Silica precipitation upon abrasion textures; euhedral outgrowths 7. Precipitation of silica after transport 8. Silica precipitation (timing unknown) Any precipitated silica without superposition Upturned silica plates and dish-shaped concavities 9. Non-cushioned, energetic impacts (eolian) 10. Fracturing (high energy) Conchoidal fractures, arcuate and linear steps, arc-stepped forms, meandering ridges Impact pits (V's > 5 #m), highly abraded edges, abundant grooves 11. High-energy subaqueous (cushioned) collisions 12. Moderate-energy subaqueous (cushioned) collisions Impact pits (V's < 5 #m) grooves and abraded edges present Few impact pits (V's < 5 tLm), no abraded edges or grooves 13. Low-energy subaqueous collisions
1. Solution of silica before transport 2. Solution of silica during transport 3. Solution of silica after transport 4. Silica solution (timing unknown) 5. Precipitation of silica before transport 6. Precipitation of silica during transport
141 A Beach
[.'
Dune
." ." . "..
"
"
Tephra I:".~--- . . . . . . c S i l ' r 0 25
• HI • .'HI
Sand Clay
[--~
16o
75
5'0 Percent
B
Beach
-e--
Dune
•
Tephra 0.4 C
*
j
-*-
• b 0.81.8 2:3 218 3.3-1.6-016 ÷0.4 .1.42.0 0:6 Sorting (~] Mean size [~] Skewness
•
I
"
3'.5 510 Kurtosis
6.5
1.0 + Beach O Dune ~k Tephra
I i
~
0
+ ~-
~+4v
*'**
O3
÷ 0.2
+
F
I
4 Mean size [~]
-0.7
0 Skewness
+0.7
Fig. 3. Bulk textural (A), univariate mean and range scores of some moment measure grain size statistical parameters (B), and examples of bivariate plots of moment statistics (C) of reference set samples from known depositional environments. Data from Christie (1975) and Stokes (1987).
In addition, m o m e n t and graphic mean, m o m e n t sorting, graphic mean divided by graphic sorting, and the slope of grain size curves between D o and D35 (probability s c a l e ) a n d between /)so and D65 (arithmetic scale) were sometimes efficient discriminators. Fig. 4 shows the degree of discrimination achieved. Discrimination indices (R0) were calculated for the two linear discriminant analyses (reference sets A and B; Fig. 4A, B), and yielded the following discrimination rules: IF sample canonical variate 1 score> 0.199 (R01~et AI) or > --0.036 (R0~e~m) THEN sample depositional environment = coastal dune ELSE sample deposition environment = beach• Although expressed in two dimensions, much of the variance (92%) of the data set C m a y be explained in terms of the first canonical discrimi-
nant function (Table 4). The degree of reference set sample discrimination is high, correct classification occurring for about 90% or more of the samples (Table 5), a considerable improvement over standard grain size measures (Fig. 3C).
Unknown set (drift hole sample) analysis Having developed a successful statistical model for discriminating samples from k n o w n environments it is possible to apply the canonical coefficients to the subsurface samples in order to attempt to distinguish (classify) their depositional settings. The discriminant function scores of the u n k n o w n set samples used are shown in Fig. 5. The degree of sample separation is generally good, and clusters of samples mimick those of the reference sets (Fig. 4).
142
TABLE 4 Summary of highly discriminating variables and the resulting canonical discriminant functions by reference set. Discrimination efficiency, determined by STEPDISC, is measured by minimisation of Wilks" Lambda selection criterion. The importance of each individual variable is indicated by its partial R 2, which measures the additional explanatory power derived by adding the variable to the specific model. A sample's discriminant function(s) score, Z , (in this case either 1 or 2), is the sum of the products of the selected sample variables and the respective standardised canonical discriminant function coefficients. The Mahalanobis squared distance statistic (,.k2) provides a comparative measure of the separation between groups Step
1 2 3 4 5
Variable entered
Number of variables
X29 PRIN 14 PRIN8 PRINll PRIN9
1 2 3 4 5
Wilks" Lambda
Partial R2
Standardised canonical discriminant functions
Function Z.
0.7509 0.6462 0.5468 0.4787 0.4377
(.4) Using refi'rence set A 1 I t).2491 1.1283 0.1394 - 0.6405 0.1538 0.6424 0.1246 0.5169 0.0857 0.4049 (B) Usmg reference set B 1 l
1 2 3 4 5 6 7 8
X8 PRIN5 PRIN14 PRINI5 PR1N12 PRIN9 X27 X8 (removed)
1 2 3 4 5 6 7 8
0.7398 0.6378 0.5401 0.4598 0.4049 0.3685 0.3419
0.2602 0.1379 0.1531 0.1488 0.1193 0.0899 0.0721
Eigenvalue
Percent of variance
Canonical correlation
1.2846
100
0.750
2.27
1.8852
100
0.808
2.72
0.6649 -0.6863 0.6530 0.5862 0.4187 1.0005
((') CSing reference set C 1
1 2 3 4 5 6 7 8 9
X8 X13 X1 X9 PRIN5 PRIN14 PRIN8 PRIN15 X27
1 2 3 4 5 6 7 8 9
0.4137 0.2751 0.2305 0.1984 0.1719 0.1499 0.1321 0.1193 0.1114
A few ( < 5%) o f t h e u n k n o w n
0.5863 0.3348 0.1624 0.1390 0.1338 0.1278 0.1182 0.0978 0.0662
samples showed
v e r y h i g h ( > 10) o r v e r y l o w ( < - 1 0 ) n a n t scores, b e y o n d
- 13.6717 -3.1317 19.0271 1.5427 1.0069 0.3007 0.5028 0.3078 -1.1334
discrimi-
the range of scores derived
2 15.5784 0.4802 12.8122 0.5558 -0.2290 0.6088 -0.0502 -0.1235 2.4350
1 2
5.2491 0.4369
~2
Beach
Beach Dune Tephra
2.6846 6.2613
92.32 7.68
0.917 0.551
Dune
Tephra
2.6846
6.2613 4.1207
4.1207
textural characteristics or were derived from depositional environments r e f e r e n c e set (e.g.,
not included within the
fluvial, e s t u a r i n e
or shallow
f r o m t h e r e f e r e n c e set. T h e s e h i g h - s c o r i n g s a m p l e s
m a r i n e ) a n d t h e i r n o n - c l a s s i f i c a t i o n is n o t c o n s i d -
always
e r e d to d e t r a c t f r o m t h e d i s c r i m i n a n t p r o c e d u r e .
had
dominated
high by
contents
halloysite
and
of
silt
and
smectite
clay,
(Stokes,
1987), i n d i c a t i n g a t e p h r i c o r i g i n (cf., Fig. 3A).
Quartz surface textures
Occasional samples remained unclassified following consideration
of independent
e v i d e n c e . It is
assumed that these samples either have anomalous
The nature and relative abundance
of the vari-
o u s s u r f a c e t e x t u r e s o n q u a r t z g r a i n s in s a m p l e s
143
14--
i
A
!
10: Ro value
8
Dune
i
Beach
N
4. N
I
2
2. i
i Ii
~
-4
0
-3
-2
-1
0 CVl
1
2
2
-2
+C
~-++
-!+ c,j > c
+++ +
0~
;
-24 -4
-1
8l
/
0 CV1 /
,1 -2
\
/+.o° &
, ¢'
+ Beach
2
\
/
¢+ +e- o / + +<~. o / / o +i
1
\ oo<~ S ° o o
°°
o ®
~ ~ , 0 CV1 o Dune
*
0
1
2
3
Ro value Beacr
4
-3
-2
-i
0
~.
1
2
\ * 2
\"\
*
\ 2
,J
'}t~ q,j.~ C
*
?.
~> e
/
~r,e +
q)~/ (9
,
'
,,
4
!
~
~il
,,,
/ ,d,
/%
Dune
* Tephra
<'~ ,~
qq
Beach
/
"~9 !P q)"~ "~
-2
0
Tephra
2
4
CV1 Fig. 5. Classification of the unknown set (drillhole) samples into inferred beach, dune or tephra facies based on transformations of their textural variables by the canonical coefficients for each of the three reference sets (Table 4, Fig. 4). Abbreviations defined in Fig. 4.
TABLE 5 Summary of the discriminant function classification efficiencies for each of the reference sets (n = number of samples) (n )
Discrimination beach
Set B
Set C
4
CV1
(canonical variates 1 and 2), are not as simple, being defined by continuous functions at group boundaries (marked here as perpendicular bisectors of the group means, shown by encircled symbols). N = number of samples.
Set A
3
Dune
2
only beach and dune samples, can be solved by only one canonical variate (CVI) resulting in their linear (1-dimensional) solution about the determined R o values. Classification rules for the third data set (C), projected in two-dimensions
Actual environment
4
4
Fig. 4. Discriminant function analysis summary of reference data sets (as defined in text). Data sets A and B, comprising
Reference sel
3
B
6
~
* \
1
CV1
N
-3
2
3
B N 46~j~2 [' XBeach ''/~~Dune~R%ixDune[~Beach
o
-3
I:
dune
tephra
% correct
beach
(22)
20
2
91
dune
(13)
0
13
100
beach
(23)
20
3
87
dune
(31)
4
27
87
beach dune
(20) (29)
18 4
2 25
0 0
90 86
tephra
(12)
0
0
12
100
144
TABLE 6 S u m m a r y of the occurrence of surface features on quartz sand grains from the T a h a r o a sand deposit Unit
Mitiwai
Waiau B Waiau A
Parawai
Nihinihi
Awhitu
Ohuka
Kaawa
Rd b
Rf c
Surface textural feature a 1
2
3
4
5
6
7
8
1
SA-SR
L
S-C
0
C
C
C
C
C
S
2
R
L
C
0-S
0
0-S
0
S
0-S
C
1
SR-R
M
S-C
0
C
C
S-C
S
S
S-C
1
R
L
C
0
0
0
0
0
0
S-C
2
SR-R
M
S
0-S
0-S
0-S
S-C
C
C
S-C
3
R
M-H
S-C
S
S
0
S
C
S
S-C
1
SR-R
M-L
S
0
0-S
0-S
S
S
S-C
C
2
R
M
0
0
0
S
C
S
S-C
C
3
SA
M-H
S
0-S
S
S
S
S
C
S-C
4
R-WR
M-L
S
0
0-S
0
S
0
S
0-S
1
R
M
0
0
0-S
0
0
S
0
0-S
2
R-WR
M-L
S
0
0
0
0-S
C
S
C S
3
SA-R
L-H
S
0
0
0
0
C
C
4
WR
L
C
0
0
0-S
S
S-C
0-S
C-A
5
R-SR
M-L
S
S-C
0-S
C
S
0-S
S
C
1
WR
L
S
0
0
0
0-S
0
S-C
S
2
R-WR
M-L
S
0
0
0
0
0
S-C
0-S
3
WR
L
S
0
0
0
0
0-S
S-C
0-S
4
R
L
S
0
0
0
0-S
0
0-S
0
5
R
L
S
0
0
0
0
C
S
S-C
1
SA-SR
M-H
0-S
0-S
0-S
0
0-S
0-S
S
S
2
SA-SR
M-L
0
S
0
0-S
S
S
C
S-C
3
SA-SR
L-H
S
0-S
0-S
0-S
S-C
C
C
S-C
1
SA-SR
M-H
S-C
0
0
0
0
0-S
S
C
a Surface textural features are those listed in Table 2 : 0 = absent, S = scarce ( < 10%), C = c o m m o n (10-50%), A = a b u n d a n t ( > 50%). b R o u n d n e s s : SA = s u b a n g u l a r , SR = s u b r o u n d e d , R = rounded, W R = well-rounded. " Relief: L = low, M = moderate, H = high.
are summarised in Table 6. The wide variety of features recognised is indicative of complex grain histories involving superposition of various mechanical and chemical processes (Table 3). In general, chemical processes dominate the surface features of most grains, those resulting from precipitation phenomena, such as grain plastering and overgrowth, completely dominating dissolution features, such as oriented etch pits or grooves. In particular, frosted grain surfaces are ubiquitous in all sand units, resulting from the post-depositional precipitation of silica and/or clay minerals on grains (Fig. 6A; Folk, 1978). A
variety of mechanisms have been suggested for the formation of eolian grain frosting, mostly from inland desert rather than coastal dune settings (e.g., Ricci Lucchi and Cassa, 1968; Margolis and Krinsley, 1971; Folk, 1978; Higgs, 1979). In the case of the Taharoa sands, groundwater acidification by intercalated peat layers (Barter, 1976), the impediment of groundwater flow by fines-rich laminae and paleosol horizons, and the abundance of metastable minerals in the deposits, were probably important factors in accelerating mineral decomposition and subsequently encouraging frosting. The dissolution of ferromagnesian grains,
145
Surface textural feature a 9
10
11
12
13
14
15
16
17
18
19
20
C 0-C
0-S S-C
S-C A
A S
C S
C 0
S-C 0
S S
A A
S S
C C
S C
C
S
S
A
C
C
0
S
A
S
S-C
S
0-S
C
A
C
0-S
S
S
C
A
0-S
A
S
S-C
0
C
C
S-C
C
0-S
0-S
S
A
S-C
S-C
0-S
C
C-A
S
0
0
0
S
A
C
C
0
C
C
S-C
A
S-C
C
S
0-S
A
C
S
0-S
S-C
S-C
0
C
S-C
S-C
S-C
0
A
C
S
S
C
S-C
S
C
S-C
C
0-S
0
A
C
S
S
0-S
0
C
C-A
0-S
0-S
S-C
0-S
A
S
0-S
0
S
C
C
0
S
0
0
S
C-A
S
S
0-S
S
C
A
S
0-S
S
0
C
A
S
0-S
S-C
0-S
0
S
S
S
S
S-C
C
C
S
0-S
S-C
C
C
A
C
S
S
S
S
A
0
S
0-S
S
S
A
C
S-C
S-C
0
C
A
S-C
A
0
S
0-S
C
A
S
0
S-C
0-S
A
S
S
0-S
S
0
C
A
S
0-S
C
0
A
S-C
S-C
0-S
S
0-S
C
A
0-S
0
S-C
0
A
0-S
S
S
0
0-S
C
A
S-C
S
S
C
C-A
C
S
0-S
S
C
C-A
S
S
0
0
0
A
S
C
0
0-S C
0 C
0 0
C-A A
S C
C C
0 0
S S
C-A A
0 S
0-S S
0-S S
S-('
S-C
0-S
0-S
C
C
0
S-C
A
S
S-C
S
S
S
0
S-C
S-C
S-C
S-C
S-C
A
C
S
S
common in the Taharoa sands (Stokes, 1987), may lead to ferric complex liberation and accelerated clay mineral formation (Pittman, 1972), which in turn encourages the precipitation of amorphous layers on grain surfaces (Folk, 1978; Walker, 1979). A variety of quartz crystal overgrowths are present on some grains. These include raised triangular and rhombic platforms (Fig. 6B), amorphous coatings (Fig. 6C) and occasional scarce fresh euhedral crystal overgrowths. Pittman (1972) proposed a genetic model of overgrowth development in which overgrowths start as numerous incipient crystals, eventually developing into overgrowths
with well defined crystal faces which may ultimately coalesce to form single euhedral crystals. The stratigraphic sequence at Taharoa does not fit clearly into Pittman's scheme because successively older units do not exhibit progressively more mature stages of overgrowth development. Rather than simply burial history or time, localised influences of ferruginous and/or clay mineral impurities, low pH groundwaters and grain size effects (fine sands being more susceptible than coarser sands to destruction of porosity; Mellon, 1964; Pittman, 1972), are suggested to have controlled the degree of overgrowth development.
146
147
Fig. 6. Scanning electron micrographs of quartz grain surface textures from the Taharoa sand deposit. Bracketed numbers relate to features given in Table 2. A. Frosted grain surface with some cracking (16, 17). B. Rhombic-shaped plastered overgrowths (20). C. Irregularly shaped grain displaying amorphous (19) and incipient crystal (20) overgrowths, particularly at grain edges and within concavities. D. Mechanically-formed, and chemically-enhanced, upturned plates (11) with superimposed adhering particles (12) and precipitated amorphous silica, clay minerals and iron oxides (19), determined using EDXA. E. Rounded, abraded grain exhibiting dish-shaped concavities (10), and scratches and irregular pits (1) which have subsequently been coated and infilled with material (12,19). F. Highly fractured, abraded grain displaying edge modification features, including conchoidal fracture (2), irregular pits (1), straight and curved scratches, and stepped cleavage planes (3). G. Clay-overcoated stepped cleavage planes (3). H. Irregular mechanical V-shaped pits (7), meander ridges (9), grooves (8) and oriented etch patterns (14).
The silica frosting masks to varying degrees many of the mechanical features on grains, which consequently tend to be rather subdued with only low to moderate relief. These include particularly upturned plates (Fig. 6D) and dish-shaped concavities (Fig. 6E), indicative of non-cushioned high energy eolian impacts (Kuenen, 1960; Krinsley and Trusty, 1985), and a variety of fracture-related features such as breakage blocks, parallel striations and fracture plates (Figs. 6F,G) typically the result of highly abrasive grain fracturing a n d / o r eolian action (Krinsley and Doornkamp, 1973). Less widespread, but evident in several samples, are V-shaped pits and meander ridges (Fig. 6H), suggestive of a history of subaqueous transport for the affected grains. Typical grain process histories for a selection of Taharoa samples are summarised in Table 7. While post-depositional silica and clay mineral precipitation (process 7, Table 3) is the dominant process, previous transport processes have seldom been completely overprinted and can still be deduced in
many cases. These are dominated by eolian (process 9) and eolian-related fracturing (process 10) processes. Some samples (e.g., Kaawa 1, Table 7) contain grains displaying both beach and dune surface textural features. The intermixture of surface textural affinities probably reflects sample mixing and contamination between the two environments (Margolis and Krinsley, 1971), or originates from differences in grain residence times within different environments causing variable degrees of feature production and paleofeature overprinting. Good correlation occurs between the depositional environments inferred from the surface textural process histories and those determined by discriminant function analyses of textural data (Table 7). In all cases, the most recent physical processes inferred to have been active upon the grain surface are in agreement with the inferred depositional environments, despite widespread post-depositional precipitation (process 7). This suggests that in the absence of other analytical
148
TABLE 7 S u m m a r y of s a m p l e process histories (see T a b l e 3) b a s e d on grain surface features. Also s h o w n are e n v i r o n m e n t s of deposition, i n d e p e n d e n t l y d e t e r m i n e d from the d i s c r i m i n a n t function analysis of textural d a t a Unit
Sequence of processes
Depositional
recognised a
environment/ process
Mitiwai
1 2
[11/12] ~ 9 ~ 7, 3 [11/12] ---, 9 --* 7, 3
dune dune
Waiau B
1
(10), 9 --, 7 *, 3
dune/paleosol
Waiau A
Parawai
Nihinihi
1
9, (10) ---, 7 *, 3
dune
2
(10), 9 ~ 7 *, 3
dune
3
(10), 9 ~ 7 *, 3
dune
1
9 -~ 7 *, 3
dune/paleosol
2
5 ~ [11/12] ---, 7*, 3
beach
3
5 -~ [12/13] ---, (3) ---, 10 --, 7 *, 3
dune
4
10, (11) -~ (9) -~ 7 " , 3
dune dune
1
10 ---, (9) ~ 7 *, 3
2
(5) ---, (9) ---, [11/12]
3
5 ~ [11/12] ~ 7 " , 3
dune
4
9 ~ 7", 3
dune
5
10 --, 7 *, 3 ~ (10)
dune/paleosol
1 2
9 ---, 7, 3 (5) ~ (9, 10) --* l l , 7 " , 3
beach beach
7 *, 3
Awhitu
Ohuka
Kaawa
beach
3
9 --, (3) ---, 7
dune
4
(9) ---, (11) ---, 7
beach
5
(11) ---. 9 --+ 7
dune
1
10 --+ 7 *, 3
tephra
2
10 ~ 7, 3 --* 12, 7
tephra
3
10 ---, 7, 3 --, (12)
tephra
1
(10) ---, [11/12] ~ 7 * (3)
beach/dune
or 5 + [12/13] ---, 7*, 3 a . : d o m i n a n c e of a process c o m p a r e d to others. [/]: either one or b o t h of the two processes were operative, distinction being uncertain. ( ): implies possible presence of process.
techniques, grain surface features could possibly be used in isolation as an environmental indicator. Conclusions
(1) Multivariate discriminant function analysis of grain size data can provide significant clues to
the environments of deposition of unknown sand samples. In order to use such discriminant functions an a priori group of samples (reference sets) from anticipated similar depositional settings is required to generate the classification procedures. A multiple approach, using different combinations of reference set data, may be used to allow checks on the classifications inferred. In this study, stepwise discriminant function analysis of the reference sets indicated no clear trend as to which variables are the best for environmental discrimination, with selected variables varying widely from set to set. This emphasises the importance of the composition of the reference set in the ultimate solution and supports an iterative approach to discriminant function analysis, subdividing a total data set into a number of varied reference sets. Reference set samples were consistently classified correctly about 90% or better of the time (Table 5). For the subsurface samples at Taharoa the environmental discrimination of the homogenised drillhole samples is enabling development of an environmental facies stratigraphy, with important repercussions for subsequent intrabasinal drillhole and interregional sequence correlation. (2) Factors which may lead to discriminant function misclassification in this type of application include pedogenetic alteration, origin of samples from depositional environments not included within the reference set and changes in sediment provenance. In our case these limitations have largely been overcome by considering ancillary, independent bulk textural ( s a n d / s i l t / clay) and compositional (% organic carbon, clay mineralogy) data. (3) The observation of surface textures on detrital quartz grains may also reveal important clues relating to processes of sediment deposition, to previous cycle grain histories, and to diagenetic and weathering processes affecting deposits. Despite rather extensive post-depositional surface precipitation on sand grains in the Taharoa deposit, grain transport process histories can be inferred from careful SEM study and are in general agreement with the environments determined from discriminant function analysis of the textural data. However, unlike the latter, the surface textural
149
analysis does not provide a simple quantitative environmental determination.
Acknowledgements The financial support provided by New Zealand Steel Mining Ltd. for this project is gratefully acknowledged. Anonymous reviewers are thanked for commenting on an earlier version of this manuscript. Mrs. S. Wright and Mrs. E. Norton are thanked for typing the manuscript.
References Afifi, A.A. and Azen, S.P., 1972. Statistical Analysis, a Computer Oriented Approach. Academic Press, New York, N.Y., 366 pp. AI-Saleh, S. and Khalaf, F.I., 1982. Surface textures of quartz grains from various recent sedimentary environments in Kuwait. J. Sediment. Petrol., 52: 215-225. Barter, T.P., 1976. The Kaihu Group, Plio-Quaternary of the Awhitu Peninsula, Southwest Auckland. Ph.D. thesis, University of Auckland, Auckland (unpublished). Beaudoin, A.B. and King, R.H., 1986. Using discriminant function analysis to identify Holocene tephras based on magnetite composition: a case study from the Sunwapta Pass area, Jasper National Park. Can. J. Earth Sci., 23: 804-812. Bull, P.A. and Culver, S.J., 1979. An application of scanning electron microscopy to the study of ancient sedimentary rock from Saionia Scarp, Sierra Leone. Palaeogeogr., Palaeoclimatol., Paleoecol., 26: 159-172. Bull, P.A., Whalley, W.B. and Magee, A.W., 1986. An annotated bibliography of environmental reconstruction by SEM 1962-1985. Br. Geomorphol. Res. Group, Tech. Bull., 35: 94 pp. Carver, R., 1971. Procedures in Sedimentary Petrology. Wiley and Sons, New York, N.Y., 372 pp. Cater, J.W.L., 1984. An application of scanning electron microscopy of quartz sand surface textures to the environmental diagnosis of Neogene carbonate sediments, Finestrat Basin, Southeast Spain. Sedimentology, 31: 717-732. Chappell, J., 1970. Quaternary geology of the South Auckland coastal region. J. R. Soc. N.Z., Earth Sci., 8: 133-153. Christie, A.B., 1975. Sedimentology of Pleistocene and Recent Ironsands, West Coast of the North Island, New Zealand. M.Sc. thesis, Victoria University of Wellington, Wellington (unpublished). Coch, N.K. and Krinsley, D.H., 1971. Comparison of stratigraphic and electron microscopic studies in Virginia Pleistocene coastal sediments. J. Geol., 79: 426-437. Dunteman, G.H., 1984. Introduction to Multivariate Statistics. Gage Publications, London, 237 pp. El-Ella, R.A. and Coleman, J.M., 1985. Discrimination be-
tween depositional environments using grain size analyses. Sedimentology, 32: 743-748. Folk, R.L., 1966. A review of grain-size parameters. Sedimentology, 6: 73-93. Folk, R.L., 1968. Petrology of Sedimentary Rocks. University of Texas Press, Austin, Texas, 99 pp. Folk, R.L., 1978. Angularity and silica coatings of Simpson Desert sand grains, Northern Territory, Australia. J. Sediment. Petrol., 4 8 : 6 1 1 - 6 2 4 . Folk, R.L. and Ward, W.C., 1957. Brazos River bar: a study in the significance of grain-size parameters. J. Sediment. Petrol., 27: 3-26. Friedman, G.M., 1961. Distinguishing between dune, beach and river sands from their textural characteristics. J. Sediment. Petrol., 31: 514-529. Georgiev, V.M. and Stoffers, P., 1980. Surface textures of quartz grains from late Pleistocene to Holocene sediments of the Persian G u l f / G u l f of O m a n - - a n application of the scanning electron microscope. Mar. Geol., 36:85 96. Hand, J.C., 1967. Differentiation of beach and dune sands using settling velocities of light and heavy minerals. J. Sediment. Petrol., 49: 599-610. Higgs, R., 1979. Quartz grain surface features of MesozoicCenozoic sands from western Labrador continental margins. J. Sediment. Petrol., 49: 599-610. Hodgson, A.V. and Scott, W.B., 1970. The identification of ancient beach sands by the combination of size analysis and electron microscopy. Sedimentology, 14: 67-75. Jenkins, D.J., Bowen, D.Q., Adams, C.G., Shackleton, N.J. and Brassell, S.C., 1985. The Neogene: Part 1. In: N.J. Snelling (Editor), The Chronology of the Geological Record. Br. Geol. Surv., Mem., 10: 199-210. Joyner, S.P., 1985. SAS User's Guide: Statistics. Version 5 Edition. SAS Institute, Cary, N.C., 956 pp. Kear, D. 1965. Geology of New Zealand's ironsand resources. 8th Commonw. Min. Metall. Congr., Australia/New Zealand, Pap., 219, 10 pp. King, R.H., Kingston, M.S. and Barnett, R.L., 1982. A numerical approach toward the classification of magnetites from tephras in southern Alberta. Can. J. Earth Sci., 19: 2012-2019. Klecka, W.R., 1975. Discriminant analysis. In: N.H. Nie, C.H. Hull, J.G. Jenkins, K. Steinbrenner and D.H. Bent (Editors), Statistical Package for the Social Sciences. McGrawHill, Toronto, Ont., pp. 434-467. Klovan, J.E., 1966. The use of factor analysis in determining depositional environments from grain-size distributions. J. Sediment. Petrol., 36: 115-125. Krinsley, D. and Donahue, J., 1968. Methods to study surface textures of sand g r a i n s - - a discussion. Sedimentology, 10: 209-216. Krinsley, D.H. and Doornkamp, J.C., 1973. Atlas of Quartz Sand Surface Textures. Cambridge University Press, Cambridge, 91 pp. Krinsley, D. and Trusty, P., 1985. Environmental interpretation of quartz grain surface textures. In: G.G. Zuffa (Edi-
150 tor), Provenance of Arenites. N A T O AS1 Series Vol. 148, Reidel, Dordrecht, pp. 213-229. Kuenen, P.H., 1960. Experimental abrasion, 4. Aeolian action. J. Geol., 68: 427-449. Lawton, D.C., 1979. Geophysical exploration of New Zealand ironsands. Ph.D. Thesis, University of Auckland, Auckland (unpublished). Margolis, S.V., 1968. Electron microscopy of chemical solution and mechanical abrasion features on quartz sand grains. Sediment. Geol., 2: 243-256. Margolis. S.V. and Kennett, J.P., 1971. Cenozoic paleo-history of Antarctica recorded in Subantarctic deep-sea cores. Am. J. Sci., 271: 1-36. Margolis, S.V. and Krinsley, D.H., 1971. Submicroscopic frosting on eolian and subaqueous quartz sand grains. Geol. Soc. Am. Bull,, 82: 3395-3406. Margolis, S.V. and Krinsley, D.H., 1974. Processes of formation and environmental occurrence of microfeatures on detrital quartz grains. Am. J. Sci., 274: 449-464. Mason, C.C. and Folk, R.L., 1958. Differentiation of beach, dune and aeolian flat environments by size analysis, Mustang Island, Texas. J. Sediment. Petrol., 28: 211-226. Mellon, G.B., 1964. Discriminatory analysis of calcite- and silicate-cemented phases of the Mountain Park Sandstone. J. Geol., 72: 786-809. Mildenhall, D.C. 1975. Lower Pleistocene palynomorphs from the Ohuka Carbonaceous Sandstone, Southwest Auckland, New Zealand. N . Z . J . Geol. Geophys., 18: 675-681. Moiola, R.J., and Weiser, D., 1968. Textural parameters: an evaluation. J. Sediment, Petrol., 38: 45-53. Moiola, R.J. Spencer, A.B. and Weiser, D., 1974. Differentiation of modern sand bodies by linear discriminant analysis. Trans, Gulf Coast Assoc. Geol. Soc., 24: 321-326. Pain, C.F., 1976. Late Quaternary dune sands and associated deposits near Aotea and Kawhia harbours, North Island, New Zealand. N . Z . J . Geol. Geophys., 19: 153-177. Pittman, E.D., 1972. Diagenesis of quartz in sandstones as revealed by scanning electron microscopy. J. Sediment. Petrol,, 42: 507-519.
Pye, K., 1983. Early post-depositional modification of aeolian dune sands. In: M.E. Brookfield and T.S. Ahlbrandt (Editors), Eolian Sediments and Processes. Developments in Sedimentology 38. Elsevier, Amsterdam, pp. 197-221. Rex, R.W., Margolis, S.V. and Murray, B. 1970. Possible interglacial dune sands from 300 m water depth in the Weddell Sea, Antarctica. Geol. Soc. Am. Bull., 81: 3465-3472. Ricci Lucchi, F. and Casa, G., 1968, Surface textures of desert quartz grains. A new attempt to explain the origin of desert frosting. G. Geol. (2), 1: 751-776. Sahu, B.K., 1964. Depositional mechanisms from the size analysis of clastic sediments. J. Sediment. Petrol. 34: 73-83. Samsuddin, M., 1986. Textural differentiation of the foreshore and breaker zone sediments on the northern Karala Coast, India. Sediment. Geol., 46: 135-145. Sevon, W.D., 1966. Distinction of New Zealand beach, dune and river sands by their grain size distribution characteristics. N . Z . J . Geol. Geophys., 9: 212-224. Sharp, M. and Gomez, B., 1986. Processes of debris comminution in the glacial environment and implications for quartz sand-grain micromorphology. Sediment. Geol., 46: 33-47. Stokes, S., 1987. Stratigraphy and Sedimentology of Late Neogene Coastal Deposits at Taharoa, Southwest Auckland, New Zealand. M.Sc. Thesis, University of Waikato, Hamilton (unpublished). Stokes, S., Nelson, C.S., Healy, T.R. and MacArthur, N.A., 1989. The Taharoa ironsand deposit. In: D. Kear (Editor), Economic Geology of New Zealand. Australian Institute of Mining and Metallurgy Monograph Series (in press). Taira, A. and Scholle, P.A., 1979. Discrimination of depositional environment using settling tube data. J. Sediment. Petrol., 49: 787-800. Walker, T.R. 1979. Red colour in dune sand. In: E.D. McKee (Editor), Study of Global Sand Seas. U.S. Geol. Surv., Prof. Pap., 1052: 61-81.