Earth Observation Scientific Workflows in a Distributed Computing Environment

Geospatially Enabled Scientific Workflows offer a promising toolset to help researchers in the earth observation domain with many aspects of the scientific process. One such aspect is that of access to distributed earth observation data and computing resources. Earth observation research often utilizes large datasets requiring extensive CPU and memory resources in their processing. These resource intensive processes can be chained; the sequence of processes (and their provenance) makes up a scientific workflow. Despite the exponential growth in capacity of desktop computers, their resources are often insufficient for the scientific workflow processing tasks at hand. By integrating distributed computing capabilities into a geospatially enabled scientific workflow environment, it is possible to provide researchers with a mechanism to overcome the limitations of the desktop computer. Most of the effort on extending scientific workflows with distributed computing capabilities has focused on the web services approach, as exemplified by the OGC's Web Processing Service and by GRID computing. The approach to leveraging distributed computing resources described in this article uses instead remote objects via RPyC and the dynamic properties of the Python programming language. The Vistrails environment has been extended to allow for geospatial processing through the EO4Vistrails package ( http://code.google.com/p/eo4vistrails/ ). In order to allow these geospatial processes to be seamlessly executed on distributed resources such as cloud computing nodes, the Vistrails environment has been extended with both multi‐tasking capabilities and distributed processing capabilities. The multi‐tasking capabilities are required in order to allow Vistrails to run side‐by‐side processes, a capability it does not currently have. The distributed processing capabilities are achieved through the use of remote objects and mobile code through RPyC.     

Sedimentary processes and facies of the waterlaid Anvil Spring Canyon alluvial fan, Death Valley, California

The Anvil Spring Canyon fan of the Panamint Range piedmont in central Death Valley was built entirely by water-flow processes, as revealed by an analysis of widespread 2- to 12-m-high stratigraphic cuts spanning the 9·7 km radial length of this 2·5–5·0° sloping fan. Two facies deposited from fan sheetfloods dominate the fan from apex to toe. The main one (60–95% of cuts) consists of sandy, granular, fine to medium pebble gravel that regularly and sharply alternates with cobbly coarse to very coarse pebble gravel in planar couplets 5–25 cm thick oriented parallel to the fan surface. The other facies (0–25% of cuts) comprises 10- to 60-cm-thick, wedge-planar and wedge-trough beds of pebbly sand and sandy pebble gravel in backsets sloping 3–28°. Both facies are interpreted as resulting from rare, sediment-charged flash floods from the catchment, and were deposited by supercritical standing waves of expanding sheetfloods on the fan. Standing waves were repeatedly initiated, enlarged, migrated, and then terminated either by gradually rejoining the flood or by more violent breakage and washout. The frequent autocyclic growth and destruction of standing waves during an individual sheetflood resulted in the deposition of multiple coarse and fine couplet and backset sequences 50–250 cm thick across the active depositional lobe of the fan. Erosional intensity during washout of the standing wave determined whether early-phase backset-bed deposits or washout-phase sheetflood couplet deposits were selectively preserved in a given cycle. Two minor facies are also found in the Anvil fan. Pebble–cobble gravel lags (0–20% of cuts) are present above erosional scours into the sheetflood couplet and backset deposits. They consist of coarse gravel concentrated through fine-fraction winnowing of the host sheetflood facies by sediment-deficient water flows. This reworking occurred during recessional flood stage or from non-catastrophic discharge during the long intervals between major flash floods. This facies is common at the surface, giving rise to a ‘braided-stream’ appearance. However, it is stratigraphically limited, present as thin, continuous to discontinuous beds or lenses that bound 50- to 250-cm-thick sheetflood sequences. The other minor facies of the Anvil fan consists of clast-supported and imbricated, thickly stratified, pebbly, cobbly, boulder gravel present in narrow, radially aligned ribbons nested within sheetflood deposits. This facies is interpreted as representing deposition in the incised channel of the fan, a subenvironment characterized by greater flow competence resulting from maintained depth from channel-wall confinement, and by more frequent water flows and winnowing events caused by its direct connection with the catchment feeder channel.     

Middle to late Triassic mélange exhumation along a pre-Andean transpressional fault system: coastal Chile (26°–42° S)

Mélanges occur as discontinuous, mappable, units along an extensive N–S-trending, steeply dipping zone of distributed shear in metamorphic complexes along the coast of central Chile. Large mélange zones, from north to south, near Chañaral, Los Vilos, Pichilemu, and Chiloé Island, contain variations in lithologic and structural detail, but are consistent in exhibiting cross-cutting fabric features indicating a progressive transition from earlier ductile to more brittle deformation. In the Infiernillo mélange near Pichilemu, Permian to Early Triassic, sub-horizontal schistosity planes of the Western Series schist are disrupted, incorporated into, and uplifted along high-angle, N–S- to NNE–SSW-trending brittle–ductile shears. Mylonitic and cataclastic zones within the mélange matrix indicate active lateral shear during cumulative exhumation from depths exceeding 12 km in some areas. Exotic lithologies, such as Carboniferous mafic amphibolite and blueschist, formed during earlier Gondwanide subduction, match well with similar rocks in the Bahia Mansa to Los Pabilos region 750 km to the south, suggesting possible dextral offset. The development of the Middle to Late Triassic, N–S=trending, near-vertical shear zones formed weaknesses in the crust facilitating later fault localization, gravitational collapse, and subduction erosion along the continental margin. The length and linearity of this zone of lateral movement, coincident with a general hiatus of regional arc magmatism during the Middle to Late Triassic, is consistent with large-scale dextral transpression, or possible transform movement, during highly oblique NNE–SSW convergence along the pre-Andean (Gondwana) margin. The resultant margin parallel N–S-trending shear planes may be exploited by seismically active faults along the present coastal area of Chile. The palaeo-tectonic setting during the transitional period between earlier Gondwanide (Devonian to Permian) and later Andean (Late Jurassic to present) subduction may have had some similarity to the presently active San Andreas transform system of California.     

A comparison of mud- and sand-dominated meanders in a downstream coarsening reach of the mixed bedrock-alluvial Klip River, eastern Free State, South Africa

Along a 28 km reach of the Klip River, eastern Free State, South Africa, mud- and sand-dominated meanders have developed in close proximity within a floodplain wetland up to 1.5 km wide, providing an unusual opportunity to compare their characteristics under similar hydrological conditions. Throughout the reach, the channel bed is grounded on sandstone/shale bedrock although the banks are alluvial, and most river activity occurs during summer high flows. The reach can be divided into three geomorphological zones: Zone 1 (0–11 km), a muddy proximal part with a single meandering channel (w/d < 10) and near-permanent standing water in oxbows and backswamps; Zone 2 (11–17.5 km), a transitional mud-to-sand part with one main channel (w/d  20–30), a number of sinuous palaeochannels and oxbows, and only limited standing water; and Zone 3 (17.5–28 km), a sandy distal part with a single meandering channel (w/d  15–30), scroll bars and oxbows, and little standing water. Each zone also has a distinctive sedimentology: Zone 1 is characterised by an  3–4 m thick succession of basal sand and minor granules overlain by dominantly muddy sediment deposited primarily by oblique accretion in meander bends; Zone 2 is characterised by < 4 m of interbedded sand and mud deposited primarily by lateral point-bar accretion, although a history of avulsions also attests to the importance of abandoned-channel accretion; and Zone 3 is characterised by < 3 m of dominantly sand deposited primarily by lateral point-bar accretion. This unusual downstream sediment coarsening trend, and the associated changes in channel and floodplain character, are independent of sediment inputs from tributaries, and result from a downstream increase in bankfull unit stream power from < 3.5 W m^− 2 (Zone 1) to  4–10 W m^− 2 (Zone 3). Mud is deposited primarily in low-energy Zone 1 but is conveyed in suspension more effectively through higher energy Zones 2 and 3, only forming drapes over sandy lateral accretion deposits during waning flood stages. The downstream increase in unit stream power is controlled in part by a slight downstream increase in floodplain gradient that may be related to a subtle variation in the erosional resistance of the bedrock underlying the channel bed. These findings add to previous work on meandering rivers by demonstrating that mud-dominated meanders can occur in long-term erosional settings where the channel bed is grounded on bedrock, and that downstream fining trends may be reversed locally.     

Breaks and unit roots in global and hemispheric temperatures: an updated analysis

The unit root testing within a breaking trend framework for global and hemispheric temperatures of Gay-Garcia, Estrada and Sánchez Clim Chang 94:333–349, 2009 is extended in two directions: first, the extended HadCRUT3 temperature series from Brohan et al. J Geophys Res 111:D12106, 2006 are used and, second, new breaking trend estimators and unit root tests are employed, along with direct modelling of breaking trend and unit root processes for the series. Some differences to the results of Gay-Garcia et al. are found: break dates are shifted to 1976 for global and northern hemisphere temperatures and to 1964 for the southern hemisphere. Although the results are somewhat ambiguous, global and northern hemisphere temperatures are probably best modelled by unit root processes with a break in drift, while southern hemisphere temperatures follow a breaking trend process with stationary fluctuations about this trend. Irrespective of the models selected, there is little evidence of trend warming before the breaks, i.e., until the third quarter of the 20th century, and after the breaks northern hemisphere and global trend temperatures warm quicker than in the southern hemisphere, the range being between 0.01 and 0.02 °C per annum.     

Cause of dominance by sheetflood vs. debris-flow processes on two adjoining alluvial fans, Death Valley, California

Two large, adjoining alluvial fans of the Panamint Range piedmont, Death Valley, California, are composed of different facies assemblages deposited by contrasting sedimentary processes. The Anvil Spring fan was built solely by water-flow processes (incised-channel floods and sheetfloods), whereas the neighbouring Warm Spring fan has been constructed principally by debris flows. The boundary between these fans delineates a sharp provincial piedmont contact between sheetflood-dominated fans to the south and debris-flow-dominated fans to the north. Factors such as climate, catchment area, fan area, catchment relief, aspect, vegetation types and density, and neotectonic setting are essentially identical for these two fans. The key difference between them is that their catchments are underlain by dissimilar bedrock types, which weather to yield distinctive sediment suites. Weathering of the granite and andesite of the Anvil fan catchment produces significant volumes of medium to very coarse sand, granules, pebbles, cobbles and boulders, but minimal silt and clay. In contrast, the shale, quartzite and dolomite that dominate bedrock in the Warm Spring catchment weather to yield a wide suite of sedimentary particles spanning from clay to boulders. The abundance of mud, and the unsorted character of the yielded sediment, cause precipitation-induced slope failures in the Warm Spring catchment to transform readily into debris flows. This propensity is due to the low permeability of the colluvial sediment, which causes added water to become trapped quickly and pore pressure to rise rapidly, promoting transformations to debris flows. In contrast, the limited volume of sediment finer than medium sand yielded from the Anvil fan catchment causes the colluvium to have high permeability. This factor prevents the transformation of wet colluvium to a debris flow during hydrologically triggered slope failures, instead maintaining sediment transport as entrained bed load or suspended load in a water flow.     

A 24 h investigation of the hydrogeochemistry of baseflow and stormwater in an urban area impacted by mining: Butte,Montana

Changes in water quality during a storm event were continuously monitored over a 24 h period at a single location along an urban stormwater drain in Butte, Montana. The Butte Metro Storm Drain (MSD) collects groundwater baseflow and stormwater draining Butte Hill, a densely populated site that has been severely impacted by 130 years of mining, milling, and smelting of copper‐rich, polymetallic mineral deposits. On the afternoon of 26 June 2002, a heavy thunderstorm caused streamflow in the MSD to increase 100‐fold, from 0·2 ft³ s⁻¹ to more than 20 ft³ s⁻¹. Hourly discharge and water quality data were collected before, during, and following the storm. The most significant finding was that the calculated loads (grams per hour) of both dissolved and particulate copper passing down the MSD increased more than 100‐fold in the first hour following the storm, and remained elevated over baseline conditions for the remainder of the study period. Other metals, such as zinc, cadmium, and manganese, showed a decrease in load from pre‐storm to post‐storm conditions. In addition to the large flush of copper, loads of soluble phosphorus increased during the storm, whereas dissolved oxygen dropped to low levels (<2 mg l⁻¹). These results show that infrequent storm events in Butte have the potential to generate large volumes of runoff that exceed Montana water quality standards for acute exposure of aquatic life to copper, as well as depressed levels of dissolved oxygen. This study has important implications to ongoing reclamation activities in the upper Clark Fork Superfund site, particularly with respect to management of storm flow, and may be applicable to other watersheds impacted by mining activities. Copyright © 2005 John Wiley & Sons, Ltd.     

Petrogenesis of orthopyroxene- and amphibole-bearing andesites, Mustique, Grenadine Islands, Lesser Antilles Arc: Isotope, trace element and physical constraints

Abstract On the island of Mustique, fresh and propylitized olivine–plagioclase–clinopyroxene basalt, plagioclase–clinopyroxene–orthopyroxene and plagioclase–clinopyroxene–amphibole andesite lavas and minor intrusions are interbedded with Oligocene pyroclastic and epiclastic rocks. Chemical data show that two isotopically identical, but chemically different, suites of lava are present: (i) the OPXS (⁸⁷Sr/⁸⁶Sr 0.70403–0.70454; ¹⁴³Nd/¹⁴⁴Nd 0.512952–0.512986; δ¹⁸O_cpx 5.49 and 5.61), comprising basalts and orthopyroxene‐bearing andesites; and (ii) the AMPHS (⁸⁷Sr/⁸⁶Sr 0.70401–0.70457; ¹⁴³Nd/¹⁴⁴Nd 0.512981–0.513037; δ¹⁸O_cpx 5.54), made up of basalts and amphibole‐bearing andesites. The OPXS has higher contents of TiO₂, P₂O₅, light rare earth elements, Sm, Pb, Th, U, Zr, Y and Nb, and higher La/Yb ratios than the AMPHS. The isotopic data suggest that both suites formed from melts derived from the same subduction‐modified depleted mantle source as the volcanic rocks of nearby St Vincent and Bequia, and the northern islands of the Lesser Antilles Arc. The immobile trace element contents, and La/Yb ratios, of the OPXS are indicative of ～10% partial melting of the source, whereas those of the AMPHS are indicative of ～25% partial melting. The within‐suite chemical variation of the OPXS is consistent with ～45% fractional crystallization of its intratelluric mineral assemblages, and that of the AMPHS is consistent with the removal of ～65% of its intratelluric assemblages. Experimental evidence suggests that both suites of basalt crystallized at pressures <8 kbar from melts containing 1–2 wt% water. After extensive fractional crystallization, the andesites crystallized at pressures between approximately 5 and 2 kbar. The OPXS magmas appear to have lost more of their water content than the AMPHS magmas. Thus, the OPXS andesites formed from melts with an estimated water content of 2–3 wt%, whereas the AMPHS andesites formed from melts containing at least 4.5 wt% water.     

Assessing and correcting the effects of the chemical weathering of potsherds: A case study using soft-paste porcelain wasters from the Longton Hall (Staffordshire) factory site

Unglazed soft-paste porcelain wasters from the Longton Hall factory site are variably depleted (75–80 rel %) in CaO relative to comparatively insoluble components (e.g., Al₂O₃, TiO₂) due to the dissolution of wollastonite (CaSiO₃, a pyroxenoid) by subsurface water. The degree of desilicification is variable (0–45 rel % SiO₂). Petrographic data and element-abundance plots suggest that these were the principal effects of the chemical weathering process in most samples. The preferential dissolution of a single phase in the unglazed Longton Hall sherds permits the semiquantitative “reversal” of weathering phenomena. Alteration effects can be corrected using porosity–volume data to constrain the amount of wollastonite originally present in the weathered sherds. The original compositions of the unglazed wasters are bracketed by arithmetically “adding back” the missing pyroxenoid components according to two endmember assumptions concerning element mobility: (1) the total leaching of wollastonite components and (2) the preferential leaching of wollastonite-derived CaO. These calculations—particularly the latter—yield results that compare favourably with the compositions of relatively unaltered (wollastonite-bearing), glazed samples from the Longton Hall site. Given the potential susceptibility of archaeological ceramics to chemical weathering, it would seem prudent that these phenomena be carefully assessed, and corrected where possible, so that analytical data for these artifacts can be judiciously interpreted. © 1998 John Wiley & Sons, Inc.