Learning in neural spatial interaction models: A statistical perspective

 In this paper we view learning as an unconstrained non-linear minimization problem in which the objective function is defined by the negative log-likelihood function and the search space by the parameter space of an origin constrained product unit neural spatial interaction model. We consider Alopex based global search, as opposed to local search based upon backpropagation of gradient descents, each in combination with the bootstrapping pairs approach to solve the maximum likelihood learning problem. Interregional telecommunication traffic flow data from Austria are used as test bed for comparing the performance of the two learning procedures. The study illustrates the superiority of Alopex based global search, measured in terms of Kullback and Leibler's information criterion. Received: December 2001 / Accepted: August 2002 The author gratefully thanks Martin Reismann (Department of Economic Geography & Geoinformatics) for his valuable research assistance.     

Die Paleozän-Eozän-Transgression in Südlibyen

Zusammenfassung An der Peripherie der Serir Tibesti (Südlibyen und nördlicher Grenzbereich der Rebublik Tschad) wurden 10 lithostratigraphische Profile aufgenommen und miteinander korreliert, um das Vorgreifen der Paleozän-Eozän-Transgression auf den afrikanischen Kontinent zu verfolgen. Es zeigt sich, daß der Höhepunkt der in der Oberkreide begonnenen Transgression während der Ablagerung der oberpaleozänen Operculinoides-Schichten und des untereozänen Kheir Mergels erreicht war. Ein direkter Zusammenhang zwischen dem Sedimentationsraum nördlich des Tibesti und dem Niger-Becken mittels der Hoggar-Tibesti-Pforte (Furon 1951, p. 71) wird abgelehnt, da die größte Verbreitung des mer saharienne südlich des Hoggar-Tibesti in der Oberkreide und nicht im Paleozän stattfand.

---

10 lithostratigraphic sections were surveyed along the periphery of the Serir Tibesti (southern Libya and northern border area of Republic Chad). Their correlation resulted in analysing the Paleocene-Eocene transgression on the African continent. It was found out that this transgression starting with the Upper Cretaceous reached its climax during sedimentation of the upper Paleocene Operculinoides-Schichten and lower Eocene Kheir Marl. A direct connection between the sedimentation area north of the Tibesti mountains and the Niger basin via the Hoggar-Tibesti-Straits (Furon 1951, p. 71) is not encountered as the climax of the extension of the mer saharienne south of Hoggar-Tibesti coincides with the Upper Cretaceous and not with the Paleocene.

---

Résumé A la périphérie du Sérir Tibesti (Libye méridionale et zone frontière septentrionale de la République du Tchad) 10 coupes lithostratigraphiques ont été établies. Une corrélation effectuée entre eux démontre le caractère progressif de la transgression du Paléocène-Eocène vers le continent africain. Cette transgression dont le début a été daté au Crétacé supérieur a atteint son extension la plus large au temps de la sédimentation des séries à Operculinoides (Paléocène supérieur) et des marnes Kheir (Eocène inférieur). L'idée d'une connexion directe entre l'aréal de sédimentation au nord du Tibesti et celui du bassin de Niger par moyen du détroit Hoggar-Tibesti (Furon 1951, p. 71) n'a pas pu être acceptée étant donné que l'extension la plus large de la «mer saharienne» au sud du Hoggar-Tibesti s'est passée au Crétacé supérieur et non pas au Paléocène.

---

- , 10 ( ). , , Kheir. , Hogger-Tibesti (Turon 1951, p. 71) , . . mer saharienne Hogger-Tibesti , .

     

Untersuchung von Schwermineralien in Moselsedimenten

Zusammenfassung In den pleistozänen Sedimenten der Mosel und einiger ihrer Nebenflüsse konnte bei der Untersuchung der Schwerminerale ein sehr hoher Opakanteil festgestellt werden. Aufgrund dieses Tatbestandes und der nur geringen Differenzierung im transparenten Schwermineralspektrum wurden Erzanschliffe hergestellt und die opaken Mineralphasen untersucht. Das Ergebnis war: der Anteil der opaken Schwerminerale der pleistozänen Moselsedimente (HöT-NT) läßt sich eindeutig von denen der Nebenflüsse unterscheiden, da bis zu 50 Korn% des Opakanteils der Moselsedimente Brauneisenooide (Herkunft: Luxemburgische Minetteformation) führen.Damit ist der Schlüssel zur eindeutigen Charakterisierung von Moselsedimenten und zur Lösung der Frage nach der Herkunft der weitverbreiteten Schotterfluren im Bereich der Wittlicher Senke gegeben.Die Untersuchung der Vorkommen lieferte neben den geomorphologischen Anhaltspunkten den Beweis, daß die Mosel bis in die Zeit zwischen OMT und UMT auch durch die Wittlicher Senke (= Wittlicher Talweite) geflossen ist und somit eine Stromspaltung der Mosel zwischen Schweich und Mülheim/Lieser bis zu dieser Zeit bestanden hat. Dadurch wird eine alte Vermutung vonGrebe (1881) bestätigt.Zusätzlich konnte mit dieser Methode nachgewiesen werden, daß der in der Wittlicher und Trierer Talweite verbreitete Löß aus den Moselsedimenten ausgeweht wurde.

---

In the Pleistocene sediments of the Mosel and some of their tributaries the investigation of the heavy minerals showed a high content of opaques. This fact and the small differences in the distribution of the transparent heavy minerals initiated our ore-mineralogical studies of the opaque heavy minerals.The findings resulted in an exact distinction of the sediments of the Mosel from those of their tributaries by typical Limonite oölites (50 grain % of the opaques) originating from the Minette-formation of Luxemburg.By using this technique it is possible to characterize sediments of the Mosel and to elucidate the origin of the widespread Pleistocene sediments of the Wittlicher Senke area.Aside from geomorphological arguments, the investigations of these sediments produced exact evidence that the Mosel took its course through the Wittlicher Senke (= Wittlicher Mosel-Valley) during the Pleistocene period until the interval of the Upper to Lower MT-times (= Obere und Untere Mittelterrasse).Therefore, the Mosel ran through two valleys (bifurcation) between Schweich and Mülheim/Lieser until the Upper MT-time which was postulated byGrebe (1881). In addition this method enabled us to verify the loess being river loess, blown out of the sediments of the Mosel.

---

Résumé En analysant les minéraux lourds des sédiments du pléistocène de la Moselle et de quelques autres de ses affluents, nous avons pu constater un très haut pourcentage de minéraux lourds opaques. A cause de ce fait et à cause des différences insignifiantes dans le spectre des minéraux lourds transparents, nous avons étudié les différentes phases des minéraux opaques. Voici le résultat: les minéraux lourds opaques des sédiments du pléistocène de la Moselle se distinguent clairement de ceux des affluents parce qu'ils contiennent — jusqu'à 50 grain % de la part opaque — des oolites de ferrite. (Origine: formation minette du Luxembourg).Il est ainsi possible de caractériser exactement les sédiments de la Moselle et d'éclairer l'origine des cailloux du pléistocène largement répandus dans la région de la Wittlicher Senke.En dehors de ces arguments géomorphologiques l'analyse de ces sédiments a fourni la preuve que — jusqu'à la période entre le niveau haut et le niveau bas de la terrasse moyenne — la Moselle a traversé la Wittlicher Senke et que, par conséquent, il y a eu deux bras entre Schweich et Mülheim/Lieser, ce qui confirme les idées deGrebe (1881).A l'aide de cette méthode on a pu prouver de plus que le loess répandu dans la Trierer et Wittlicher Talweite provient des sédiments de la Moselle, et a été emporté par le vent.

---

. . . , , .. 50 % , Minnette, . , , . , UMT / . .. (1881 .) . , .

     

Plate tectonics and intracratonic mountain ranges in Morocco - The mesozoic-cenozoic development of the Central High Atlas and the Middle Atlas

Plate tectonic processes in the Atlantic and western Tethyan realm directed the post-Variscan sedimentary and structural evolution of the High Atlas and Middle Atlas intracontinental mountain ranges of Morocco. Plate movements caused a reactivation of an inherited pan-African or Hercynian fault pattern by the variation of stress regimes through time. This resulted in strike-slip as well as vertical tectonics. During times of relative tectonic quiescence eustatic sea-level changes governed the sedimentary development. The most important, often interacting, global tectonic determinants are: taphrogenesis of the NW-African continental margin lasting until the Early Cretaceous (Triassic rifting and subsequent mid-Atlantic spreading), strike-slip-faulting at the Newfoundland-Gibraltar fault zone (Liassic — earliest Eocene), and continental convergence between Europe (Iberia) and Africa which started in the Late Cretaceous and reached its acme in the Neogene. In the realm of the Central High Atlas and the Middle Atlas the interaction of these processes triggered continental rifting (Triassic) and subsequent marine flooding of the intergrown riftgrabens prograding from the Tethys realm (Early Jurassic — earliest Middle Jurassic). After its abortion, the former Atlas rift was filled up with marine sediments (Bajocian — Bathonian), followed by continental redbeds and final uplift (late Mid Jurassic — late Early Cretaceous). Eustatic sea-level changes mostly governed the sedimentary evolution from Aptian to latest Mid Eocene. After a first weak uplift of the central High Atlas during the Senonian major uplift of the intracontinental chains commenced at the Mid/ Late Eocene transition. Diastrophism of the Atlas ranges during the Miocene and Pliocene coincided with the main orogenic movements of the Betico-Rifean arc.

---

Zusammenfassung Die post-variskische sedimentäre und strukturelle Entwicklung der intrakontinentalen Gebirgsketten des Hohen und des Mittleren Atlas wurde durch plattentektonische Prozesse im atlantischen und westmediterranen Raum gesteuert, die auf ein ererbtes panafrikanisches bzw. variskisches Störungsmuster einwirkten. Zu Zeiten relativer tektonischer Ruhe bestimmten eustatische Meeresspiegelschwankungen die sedimentäre Entwicklung. Die von horizontaler und vertikaler Bruchtektonik beherrschte Atlas-Tektonik wurde durch in der Zeit wechselnde Streßfelder bestimmt. Wichtigste großtektonische, oft zusammenwirkende Steuerungsfaktoren waren: die bis in die Unterkreide wirkende Taphrogenese im Bereich des NW-afrikanischen Kontinentalrandes (triadisches Rifting und anschließendes Spreading im Atlantik), Transform-Bewegungen an der Neufundland-Gibraltar-Störungszone zwischen Lias und ältestem Eozän sowie die Konvergenz zwischen Europa (Iberia) und Afrika, welche nach ihrem Beginn in der späten Kreide ihren Höhepunkt im Neogen erreichte. Das Zusammenspiel dieser Faktoren bewirkte im zentralen Hohen Atlas und im Mittleren Atlas die Entstehung kontinentaler Riftgräben während der Trias, die von der Tethys ausgehende Flutung der zusammengewachsenen Riftgräben im Lias und untersten Dogger (Sinemurium — Aalenium) sowie die zuerst marine (Bajocium — Bathonium), dann kontinentale Füllung und Heraushebung der stillgelegten Riftzone zwischen spätem Mitteljura und später Unterkreide. Zwischen Apt und Ende des Mitteleozän kontrollierten vor allem eustatische Meeresspiegelschwankungen die sedimentäre Entwicklung. Nach ersten Hebungen im zentralen Hohen Atlas während des Senon begann die eigentliche Heraushebung der intrakontinentalen Ketten an der Wende Mittel-/Obereozän. Die mio-/pliozäne Hauptphase der Atlas-Tektogenese war zeitgleich mit der Orogenese im Betico-Rif-Bogen.

---

Résumé L'évolution sédimentaire et structurale postvarisque des chaînes intracontinentales du Haut et du Moyen Atlas a été régie par les processus de la tectonique des plaques qui se déroulaient dans l'Atlantique et dans la Méditerrannée occidentale. Ces mouvements de plaques ont réactivé un ensemble de failles héritées des phases pan-africaine et hercynienne, en y produisant des déplacements verticaux et horizontaux en réponse aux fluctuations temporelles des champs de contraintes. Au cours des périodes de calme tectonique relatif, les changements eustatiques du niveau de la mer ont déterminé le régime de la sédimentation. Les facteurs de la tectonique les plus importants, à action d'ailleurs souvent combinée, ont été: la traphrogenèse, active jusqu'au Crétacé inférieur, de la marge nord-ouest du continent africain (rifting triasique et ouverture subséquente de l'Atlantique), les mouvements de décrochement le long de la zone transformante de Terre Neuve-Gibraltar depuis le Lias jusqu'au début de l'Eocène, ainis que la convergence entre l'Europe (Ibérie) et l'Afrique qui débuta au Crétacé supérieur pour culminer au Néogène. Dans le Haut Atlas central et dans le Moyen Atlas, le jeu combiné de ces facteurs a provoqué la formation de grabens continentaux (Trias) puis leur envahissement par la mer à partir de la Téthys (Lias-Dogger inférieur). Après la fin de son activité, le rift atlasique a été comblé par des sédiments, d'abord marins (Bajocien-Bathonien), puis continentaux jusqu'au soulèvement final (fin du Jurassique moyen — fin du Crétacé inférieur). De l'Aptien à la fin de l'Eocène moyen, l'évolution sédimentaire a été régie par les changements eustatiques du niveau de la mer. Après un premier soulèvement modéré du Haut Atlas central au cours du Sénonien, le soulèvement majeur des chaînes intracontinentales débuta à la limite Eocène moyen-Eocène supérieur. Le diastrophisme de l'Atlas au cours du Mio-Pliocène a coïncidé avec les mouvements orogéniques de l'arc bético-rifain.

---

- , , . , , . , , , , , . , , : - , ( ), - , () , . , , (- ), , (- ), , ; . o . . / .

     

Visual Event Screening of Continuous Seismic Data by Supersonograms

We present a new visualization method for human inspection of seismic data called supersonograms, which maximizes the amount of time and stations visible on screen while retaining the possibility to detect short and low-signal to noise ratio (SNR) signals. This visualization approach is integrated into a seismological software suite used in the seismic aftershock monitoring system (SAMS) of Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO) on-site inspections (OSI) to detect suspicious events eventually representing aftershocks from an underground nuclear explosion (UNE). During an OSI, huge amounts of continuous waveform data accumulate from up to 50 six-channel mini-arrays covering an inspection area of 1,000 square kilometers. Sought-after events can have magnitude as low as $\hbox{M}_{\rm L}\,{-2.0}$ and duration of just a few seconds, which makes it particularly hard to discover them in large, noisy datasets. Therefore, the data visualization is based on nonlinearly scaled, noise-adapted spectrograms, i.e., sonograms, which help to distinguish weak signal energy from stationary background noise. Four single-trace sonograms per mini-array can be combined into supersonograms, since the array aperture is small and sonograms suppress differences of local site noise, allowing an analyst to check array-wide signal coherence quickly. In this paper, we present the supersonograms and the software on the basis of a dataset from a creeping, inhabited landslide in Austria where the same station layout is used as in an OSI. Detected signals are fracture processes in the sedimentary landslide, i.e., slidequakes, with $\hbox{M}_{\rm L}\,{-0.5} \,\hbox{\,to}\,{-2.5}$ between July 2009 and July 2011. These signals are comparable in magnitude and duration to expected weak UNE aftershocks.     

Assessing tectonic and climatic causal mechanisms in foreland‐basin stratal architecture: insights from the Alborz Mountains,northern Iran

The southern foreland basin of the Alborz Mountains of northern Iran is characterized by an approximately 7.3‐km‐thick sequence of Miocene sedimentary rocks, constituting three basin‐wde coarsening‐upward units spanning a period of 10⁶ years. We assess available magnetostratigraphy, paleoclimatic reconstructions, stratal architecture, records of depositional environments, and sediment‐provenance data to characterize the relationships between tectonically‐generated accommodation space (A) and sediment supply (S). Our analysis allows an inversion of the stratigraphy for particular forcing mechanisms, documenting causal relationships, and providing a basis to decipher the relative contributions of tectonics and climate (inferred changes in precipitation) in controlling sediment supply to the foreland basin. Specifically, A/S > 1, typical of each basal unit (17.5–16.0, 13.8–13.1 and 10.3–9.6 Ma), is associated with sharp facies retrogradation and reflects substantial tectonic subsidence. Within these time intervals, arid climatic conditions, changes in sediment provenance, and accelerated exhumation in the orogen suggest that sediment supply was most likely driven by high uplift rates. Conversely, A/S < 1 (13.8 and 13.8–11 Ma, units 1, and 2) reflects facies progradation during a sharp decline in tectonic subsidence caused by localized intra‐basinal uplift. During these time intervals, climate continued to be arid and exhumation active, suggesting that sediment supply was again controlled by tectonics. A/S < 1, at 11–10.3 Ma and 9‐6–7.6 Ma (and possibly 6.2; top of units 2 and 3), is also associated with two episodes of extensive progradation, but during wetter phases. The first episode appears to have been linked to a pulse in sediment supply driven by an increase in precipitation. The second episode reflects a balance between a climatically‐induced increase in sediment supply and a reduction of subsidence through the incorporation of the proximal foreland into the orogenic wedge. This in turn caused an expansion of the catchment and a consequent further increase in sediment supply. Copyright © 2013 John Wiley & Sons, Ltd.     

History of anthropogenic nitrogen input to the German Bight/SE North Sea as reflected by nitrogen isotopes in surface sediments,sediment cores and hindcast models

The German Bight/SE North Sea is considered a hot-spot of river-induced eutrophication, but the scarce observational data of river nitrate loads prior to the 1970s complicate the assessment of target conditions for environmental management and legislation. Stable nitrogen isotope ratios (δ¹⁵N) in sediment records can be used to decipher historical river nitrate contributions. To better constrain pre-1970s conditions, we determined δ¹⁵N in archive sediment samples (1950–1969) and dated cores from the Helgoland depositional area. We also modeled the δ¹⁵N in past situations (1960 and 1860) using an N-isotope-tracking ecosystem model. The modeled spatial distribution of δ¹⁵N in sediments for 1960 conditions and the observed spatial pattern of δ¹⁵N in archive sediment samples (1950–1969) represent a period of moderate eutrophication. The modeled spatial distribution of δ¹⁵N in sediments for 1860 conditions (pre-industrial) showed a moderate δ¹⁵N gradient from the Elbe river mouth (δ¹⁵N<4‰) to the open sea (δ¹⁵N∼5‰). This pattern contrasts with the δ¹⁵N pattern in modern surface sediments, which exhibits a steep and inverted δ¹⁵N gradient from the Elbe river mouth (δ¹⁵N>9‰) to the open sea (δ¹⁵N<7‰). Modeled δ¹⁵N for 1860 conditions are consistent with δ¹⁵N values observed in dated sediment cores that span the last 900 years. Value of δ¹⁵N in sediment cores increased from approximately 1860 to 2000 by 2.5‰. The increasing trend reflects changes in the abundance and isotopic composition of riverine nitrate loads caused by anthropogenic activities. Sensitivity tests suggest that loads and isotopic ratios of nitrogen forms other than nitrate (ammonium and organic nitrogen) have minor impact on the modeled surface sediments, despite their higher abundance in the riverborne loads in the past. Our results suggest that eutrophication of the German Bight predates the 1960 period of documented rapidly increasing river loads. Pre-industrial levels of δ¹⁵N modeled with 28% of the modern annual (1990–1999) atmospheric loads and 10% of the modern annual river loads agree best with levels of δ¹⁵N (∼6‰) observed in sediments of the cores dated to 1860.     

Tectonic controls on Cenozoic foreland basin development in the north‐eastern Andes,Colombia

In order to evaluate the relationship between thrust loading and sedimentary facies evolution, we analyse the progradation of fluvial coarse‐grained deposits in the retroarc foreland basin system of the northern Andes of Colombia. We compare the observed sedimentary facies distribution with the calculated one‐dimensional (1D) Eocene to Quaternary sediment‐accumulation rates in the Medina wedge‐top basin and with a three‐dimensional (3D) sedimentary budget based on the interpretation of ～1800 km of industry‐style seismic reflection profiles and borehole data. Age constraints are derived from a new chronostratigraphic framework based on extensive fossil palynological assemblages. The sedimentological data from the Medina Basin reveal rapid accumulation of fluvial and lacustrine sediments at rates of up to ～500 m my^?1 during the Miocene. Provenance data based on gravel petrography and paleocurrents reveal that these Miocene fluvial systems were sourced from Upper Cretaceous and Paleocene sedimentary units exposed to the west in the Eastern Cordillera. Peak sediment‐accumulation rates in the upper Carbonera Formation and the Guayabo Group occur during episodes of coarse‐grained facies progradation in the early and late Miocene proximal foredeep. We interpret this positive correlation between sediment accumulation and gravel deposition as the direct consequence of thrust activity along the Servitá–Lengupá faults. This contrasts with one class of models relating gravel progradation in more distal portions of foreland basin systems to episodes of tectonic quiescence.     

Soil frost effects on soil water and runoff dynamics along a boreal forest transect: 1. Field investigations

To determine how soil frost changes flowpaths of runoff water along a hillslope, a transect consisting of four soil profiles directed towards a small stream in a mature forest stand was investigated at Svartberget, near Vindeln in northern Sweden. Soil temperature, unfrozen water content, groundwater level and snow depth were investigated along the transect, which started at the riparian peat, and extended 30 m upslope into mineral soils. The two, more organic‐rich profiles closest to the stream had higher water retention and wetter autumn conditions than the sandy mineral soils further upslope. The organic content of the soil influenced the variation in frost along the transect. The first winter (1995–96) had abnormally low snow precipitation, which gave a deep frost down to 40–80 cm, whereas the two following winters had frost depths of 5–20 cm. During winter 1995–96, the two organic profiles close to the stream had a shallower frost depth than the mineral soil profile higher upslope, but a considerably larger amount of frozen water. The fraction of water that did not freeze despite several minus degrees in the soil was 5–7 vol.% in the mineral soil and 10–15 vol.% in the organic soil. From the measurements there were no signs of perched water tables during any of the three snowmelt periods, which would have been strong evidence for changed water flowpaths due to soil frost. When shallow soil layers became saturated during snowmelt, especially in 1997 and 1998, it was because of rising groundwater levels. Several rain on frozen ground events during spring 1996 resulted in little runoff, since most of the rain either froze in the soil or filled up the soil water storage. Copyright © 2001 John Wiley & Sons, Ltd.