Rule-based spatial search

The application of applied artificial intelligence techniques to geographical data processing has attracted considerable interest in recent years as reflected in the research agendas of major GIS research institutions. This paper illustrates how a simple spatial problem, formalised using predicate calculus, may be readily implemented as a rule base in which the production rules represent Horn clauses. The theoretical significance of this exercise is the search for a high-level mathematical language capable of expressing, in a generic and implementation-independent syntax, the spatial relationships and queries typically handled by a GIS. Of more practical interest is the example it provides of mapping a conventional algorithm into a production rule knowledge structure.     

Detection of Lateglacial distal tephra layers in the Netherlands

Three distal tephra layers or cryptotephras have been detected within a sedimentary sequence from the Netherlands that spans the last glacial-interglacial transition. Geochemical analyses identify one as the Vedde Ash, which represents the southernmost discovery of this mid-Younger Dryas tephra so far. This tephra was found as a distinct horizon in three different cores sampled within the basin. The remaining two tephras have not been geochemically 'fingerprinted', partly due to low concentrations and uneven distributions of shards within the sequences sampled. Nevertheless, there is the potential for tracing these tephra layers throughout the Netherlands and into other parts of continental Europe. Accordingly, the possibilities for precise correlation of Dutch palaeoenvironmental records with other continental, marine and ice-core records from the North Atlantic region are highlighted.     

Are long subglacial bedforms indicative of fast ice flow?

It has been suggested that extremely long subglacial bedforms (e.g. attenuated drumlins and mega-scale glacial lineations) record former areas of fast-flowing ice and that bedform elongation ratio is a useful proxy for ice velocity. Despite the availability of much data pertaining to the measurement and analysis of subglacial bedforms, these assumptions have rarely been explicitly addressed in detail. In this paper, we demonstrate that long subglacial bedforms (length:width ratios ≥10:1) are indicative of fast ice flow. Using satellite imagery, we mapped over 8000 lineaments associated with a highly convergent flow pattern near Dubawnt Lake, District of Keewatin, Canada. This flow pattern is unusual in that it displays a large zone of convergence feeding into a main 'trunk' and then diverging towards the inferred ice margin. The 'bottleneck' pattern is taken to record an increase and subsequent decrease in ice velocity and we analysed transverse and longitudinal variations in bedform morphometry. The main trunk of the flow pattern (down-ice of the convergent zone) is characterized by mega-scale glacial lineations of great length (up to 13 km) and high elongation ratios (up to 43:1). The down-ice variations in elongation ratio reflect exactly what we would expect from a terrestrial ice stream whose velocity increases in the onset zone passes through a maximum in the main trunk and slows down as the ice diverges at the terminus. It is suggested that any unifying theory of drumlin formation must be able to account for the association between long subglacial bedforms and fast ice flow, although it is not assumed that fast ice flow always produces attenuated bedforms. A further implication of this work is that many more ice streams may be identified on the basis of attenuated subglacial bedforms, radically altering our views on the flow dynamics of former ice sheets.     

Laboratory sustained turbidity currents form elongate ridges at channel mouths

The turbulent flow structure, suspended sediment dynamics and deposits of experimental sustained turbidity currents exiting a channel across a break in slope into a wide tank are documented. The data shed light on the flow evolution and deposit geometry of analogous natural channel‐fed submarine fans. Flows generated in a 0·3 m wide, sloping (0°, 6°, 9° or 20°) channel crossed an angular slope break and spread onto a horizontal tank floor. Flow development comprised: (i) channelized phase (unsteady channel flow developing into steady channel flow); (ii) initial lateral expansion phase (unsteady‐spreading wall jet phase); (iii) constant lateral expansion phase (steady‐spreading wall jet phase); and (iv) rapid waning phase. Phases (i) and (iv) are similar to laterally constrained turbidity currents, but phases (ii) and (iii) are considerably different from such two‐dimensional currents. Steeper channel slopes produced greater flow velocities and turbulence intensities, but these effects diminished markedly with distance from the channel mouth. Flow velocity vectors in the tank had similar patterns for all channel slopes, with a central core of faster velocity and narrow vector dispersion and slower flow with larger dispersion at the jet margins. Suspended sediment concentrations were higher within flow heads and dense basal layers in flow bodies. Time‐averaged acoustic backscatter data showed vertical concentration gradients, confirmed by siphon samples. The deposits comprised a thick central ridge, of similar order width to the channel mouth, with abrupt margins and a surrounding, very thin, fan‐like sheet. The ridge was coarser grained and better sorted than the original sediment, with grain‐size fining downstream, particularly over the fan‐like sheet. The formation of a central ridge suggests that, in the tank, vertical turbulent momentum exchange is more significant for sediment dynamics than spanwise momentum exchange due to lateral expansion. The streamwise elongate geometry of the ridge contrasts with conventional fan‐like geometry found with surge‐type turbidity flows, a result that has widespread stratigraphic and economic implications.     

The Petrogenesis of the Kirwan Basalts of Dronning Maud Land, Antarctica

The 420 m thick sequence of Kirwan basalt crops out along thesouthernmost 50 km of the Kirwanveggen Escarpment (74?S, 6?W).There is little variation in major element chemistry of thesebasalts (SiO₂ 49?3–51?6 wt.%; MgO 5?1–6?6 wt.%),but the concentrations of certain incompatible elements (e.g.,Zr) vary by factors of approximately two or more. Most interelementplots show rather poor correlation (r<0?78), but rocks fromopposite ends of the data array can be related by 30% fractionationof plagioclase, clinopyroxene, olivine, and magnetite in theproportions 51:35:11:3. Plagioclase is much more abundant inphenocryst assemblages (85%) and it appears that selective transportof plagioclase to the surface occurred. The range in incompatible element concentrations cannot be explainedby crystal fractionation and is most probably a result of theparent liquids of these basalts being derived by slightly differentdegrees of partial melting of a common source, or alternativelyof open-system (RTF) magma processes. The strontium isotopedata for the freshest rocks (R₀=0?7049–0?7065) may beexplained by 7% contamination by crustal material with an R₀of 0?709 and bulk Sr of800 ppm, but there is little supportingevidence from other trace element variations for this hypothesis.Oxygen isotope determinations on whole-rock-plagioclase pairsshow that alteration has resulted in a 0?5%o shift in (¹⁸O.Alteration also appears to have resulted in a greater spreadof data, particularly for the LIL elements and Sr isotopes.The Sr and Nd isotopic composition of the suite is close tobulk Earth at 172 Ma and this, together with REE and other traceelement data, shows these basalts to be similar in compositionto the more primitive basalts among the Karoo basalt lavas.It is suggested that the Kirwan basalts were derived from asource which was similar to that of the southern Lebombo variantof the Sabie River Basalt Formation of the Karoo Volcanic Province.This part of the Karoo was closest to the Kirwanveggen beforethe break-up of Gondwanaland.     

The Generation of Continental Crust: An Integrated Study of Crust-Forming Processes in the Archaean of Zimbabwe

The Archaean craton of Zimbabwe includes two major episodesof crust generation at 3.5 and 2.9 Ga recorded in the emplacementof tonalite-gneiss granitoids. A total of 180 samples of representativegneisses and massive tonalites and sills has been collectedfrom three areas in the southern part of the craton, at Mashaba,Chingezi, and Shabani. These rocks have been analysed for major,trace, and rare earth elements to evaluate the effects of thefractional crystallization and partial melting processes inthe generation of this segment of Archaean crust. Three groups are distinguished on the basis of their major andtrace element contents, and they follow two main trends of differentiation:the sodic and the calc-alkaline (sensu stricto) trends. GroupI samples are tonalitic in composition and follow a sodic trendcharacterized by decreasing CaO/Na₂O ratios. Y and Sr behaveas compatible elements and are negatively correlated with Rb.REE patterns are moderately fractionated with La/Ybn=4–23.5.The characteristics of this group have been described only inthe Archaean craton from Swaziland. Group II is an intermediateGroup with a marked decrease in Na₂O/K₂O with increasing differentiation,similar to the Archaean tonalite-trondhjemite-granodiorite suitesfrom Finland or the Pilbara Block, Australia. Samples displaybiotite tonalite and trondhjemite compositions, and Y, Sr, andRb are all incompatible. The REE patterns are strongly fractionated,with La/Ybn=23–44, and with small positive or negativeEu anomalies, as observed in other Archaean tonalite-trondhjemites.Group III is composed mainly of trondhjemites and granites similarto many post-Archaean granitoids: they follow a calc-alkalinetrend (sensu stricto) with decreasing CaO/Na₂O and Na₂O/K₂O.Sr and Y are incompatible, whereas Rb increases with differentiation.REE patterns are variably fractionated, with La/Ybn=6–36,with high REE contents, and marked negative Eu anomalies. The above geochemical features are explained in a three-stagepetrogenetic model. The first stage consists of 6–20%melting of upper-mantle peridotite and the generation of tholeiiticbasalts, as observed in the associated greenstone belts. Thesecond stage involves 4–25% partial melting of metamorphosedbasalts with a Gt amphibolite (15–45% Pl + 30–50%Hb+2–35% Cpx+3– 15% Gt) residue resulting in theGroup I samples, under water-unsaturated conditions at intermediatepressure (16 kbar), or with an eclogite residue to generatethe parental magmas for the Group II rocks. The third stageis lowpressure fractional crystallization (<8 kbar) of liquidsgenerated during this second stage, leaving a 19–20% Qtz+36–42%Pl0–2% HbMt cumulate for the more evolved Group II samples,and 55% fractional crystallization of a 14% Qtz+37.6% Pl (An₂₆)3.3%Bt+0.1% Ilm0.8% Mt cumulate for Group III samples. The highlyfractionated REE patterns of the Group II rocks are inheritedfrom the second stage of partial melting of the metamorphosedbasalt source rocks with an eclogite residue. Thus Group IIand III initial liquids were generated through partial meltingof eclogite and Gt amphibolite, respectively. The genetic relationshipsbetween Group I sodic and Group III calc-alkaline suites areevaluated, with the latter resulting from various stages offractional crystallization processes of parental magmas withinthe sodic suite.     

Noble Metal Enrichment Processes in the Merensky Reef, Bushveld Complex

We have analysed sulphides, silicates, and chromites of theMerensky Reef for platinum-group elements (PGEs), Re and Auusing laser ablation-inductively coupled plasma mass spectrometryand synthetic pyrrhotite standards annealed with known quantitiesof noble metals. Os, Ir and Ru reside in solid solution in pyrrhotiteand pentlandite, Rh and part of the Reef’s Pd in pentlandite,whereas Pt, Au, Re and some Pd form discrete phases. Olivineand chromite, often suspected to carry Os, Ir and Ru, are PGEfree. All phases analysed contain noble metals as discrete micro-inclusionswith diameters typically <100 nm. Inclusions in sulphidescommonly have the element combinations Os–Ir–Ptand Pt–Pd–Au. Inclusions in olivine and chromiteare dominated by Pt ± Au–Pd. Few inclusion spectracan be related to discrete noble metal phases, and few inclusionshave formed by sub-solidus exsolution. Rather, some PGE inclusions,notably those in olivine and chromite, are early-magmatic nuggetstrapped when their host phases crystallized. We suggest thatthe silicate melt layer that preceded the Merensky Reef wasPGE oversaturated at early cumulus times. Experiments combinedwith available sulphide–silicate partition coefficientssuggest that a silicate melt in equilibrium with a sulphidemelt containing the PGE spectrum of the Merensky ore would indeedbe oversaturated with respect to the least soluble noble metals.Sulphide melt apparently played little role in enriching thenoble metals in the Merensky Reef; rather, its role was to immobilizea pre-existing in situ stratiform PGE anomaly in the liquid-stratifiedmagma chamber. KEY WORDS: Bushveld Complex; Merensky Reef; laser-ablation ICP-MS; platinum-group mineralization     

Oxygen and Hydrogen Isotope Stratigraphy of the Rustenburg Layered Suite, Bushveld Complex: Constraints on Crustal Contamination

New ¹⁸O values for plagioclase, pyroxene and olivine, and limitedwhole-rock D values are presented for samples from the RustenburgLayered Suite of the Bushveld Complex, South Africa. In combinationwith existing data, these provide a much more complete compositeO-isotope stratigraphy for the intrusion. Throughout the layeredsuite, mineral ¹⁸O values indicate that the magmas from whichthey crystallized had ¹⁸O values that were about 7·1,that is, 1·4 higher than expected for mantle-derivedmagmas, suggesting extensive crustal contamination. More limitedH-isotope data suggest that the OH present within whole rocks,regardless of the degree of alteration, is of magmatic originand not an alteration phenomenon. There appears to be no systematicchange in ¹⁸O value with stratigraphic height and this requiresthe contamination to have taken place in a ‘staging chamber’before emplacement of the magma(s) into the present chamber.Large amounts (30–40%) of contamination by the lower tomiddle crust are needed to explain these ¹⁸O values, which isin general agreement with previous estimates based on Sr- andNd-isotope data. Alternatively, smaller amounts of contamination(20%) by sedimentary rocks, or their partial melts, representedby the country rock can explain the data, but it is not apparenthow such material could have been present at the depth of the‘staging chamber’ in the lower to middle crust. KEY WORDS: Bushveld Complex; Rustenburg Layered Suite; oxygen isotopes; hydrogen isotopes; crustal contamination