Three-Dimensional Facies Analysis using 2 Object-Based Geobody Modeling : A Case Study for 3 the Farewell Formation , Maui Gas Field , Taranaki 4 Basin , New Zealand 5

The early-mid Paleocene Farewell Formation is stratigraphically distributed across the 10 southern Taranaki Basin (STB) which is also encountered within the Maui Gas Field. Using 11 available 3D seismic and well log data, a challenging task to delineate the spatial distribution and 12 geobody patterns of the potential reservoir sands of the formation was performed. Object based 13 modeling coupled with sequential indicator simulation were used to analyze the spatial 14 distribution of facies configuration and a conceptual model was developed based on the outputs 15 from the structurallymodeled grids. The facies modeling followed a hierarchical object-based 16 mechanism which was set to perform with constraints like channel geometry and heterogeneity 17 within the formation. The resultant 3D geobody model showed that the distributary channels, 18 mainly braided geobodies flowed from northeast cutting through several regional normal-fault 19 systems to the southwest. Overbank facies was adhered to the fringe of the channels whereas the 20 floodplain facies was at the periphery of the model. Meandering channel-sand facies were mostly 21 observed at the center of the model flowing in a more random manner, occupying major flow 22 directions of northwest to southwest and southeast to northwest within the model. 23


Introduction
The Maui Field produces gas from three main reservoirs of Paleocene to Eocene in age.Farewell Formation is the deepest and the thinnest of the reservoirs of this gas field having measured drilling depths ranging from 3100-3200m TVDSS [1][2][3][4].The Farewell Formation in the Maui Gas Field includes gas and condensate-saturated interval of approximately 100m thickness.The reservoir interval consists of highly discontinuous fluvial sandstones and associated siltstones and shales deposits as identified in the well log responses.The main reservoir sandstones are braided and meandering stacked channel sand deposits that are highly lenticular, with lateral dimensions ranging from few hundred to several kilometres in length [5][6][7][8].Crevasse-splay deposits comprise a relatively lower proportion of the overall sandstone volume and are generally of lower reservoir quality [5].
Therefore, an understanding of the structure as well as the thickness, distribution and connectivity of the sandstone deposits as separate geobody units are important for the well planning during future development of the field.This is significant because during well planning phase, 500-1500m spacing between wells is often necessary to effectively deplete the reservoir given the limited lateral extent of the fluvial sandstones within prospective formations like that of Farewell [9][10].Even with intensive development of the Farewell reservoir with spacing of 1-2 km for future drillings, bottom-hole pressure tests revealed that almost all the major sand packages tested in all four wells were in pressure communication showing partial pressure depletion [11][12][13].
The primary objectives of this study were (1) to interpret the primary stratigraphic framework of the Farewell interval within the study area of the Maui Gas Field, (2) to characterize and model lithology distribution according to the environment of deposition, (3) to explain the predicted fluvial sand-bodies within the reservoir and paleo-depositional modeling of the Farewell reservoir for deeper reservoir predictions.The study area is within the northern portion of the offshore Taranaki Basin (Fig. 1) and includes well information and 3D seismic data.The seismic survey covers an area of ~150 sq.km with average grid size of the cube being 50m X 50m (Fig. 2).The interval of interest for this study approximately coincides with the mainly condensate/gas-productive interval at the Maui Gas Field and includes an average of ~100m thick interval of the Farewell Formation.sediment thicknesses, and incorporate paleo-bathymetry; no allowance was made for sediment de-compaction, and the sections have not been structurally balanced.The original thickness of Miocene sediment is uncertain.

Geological Setup and Stratigraphy
The Taranaki Basin has a composite morphology resulting from diverse episodes of tectonic activity [2,[14][15].It consists of several superimposed sub-basins, depocentres, and areas of uplift, that range in age from mid-Cretaceous to Recent (Fig. 3).The basin's structural development was  Stratigraphically Farewell Formation falls within Kapuni Group of Paleocene age (Fig. 4).
Paleocene and Eocene deposits belong to the laterally equivalent Kapuni and Moa groups, which constitute a late-rift and post-rift transgressive sequence [6,18].The Apart from Farewell Formation, Kapuni Group also consists of Kaimiro, Mangahewa, and McKee formations.It is predominantly terrestrial, including some marginal marine deposits [6][7][8].A paraconformity at the top of the Moa-Kapuni succession reflects waning subsidence and sediment supply, and marks the culmination of this depositional phase in the latest Eocene-Early Oligocene.Formation for all the available wells were also utilized.3D seismic data on the Maui Gas Field was the basis for the study on which the model was built and studied.Velocity model was later generated for the conversion of the seismic cube from time to depth domain.Farewell horizon was identified based on the formation top, seismic signature and regional geological understanding.For the Farewell Formation, there were no available core data and therefore the 3D depositional facies reconstruction was largely dependent on well log interpretation and seismic attribute analysis of the Farewell Formation.

Three-Dimensional Structural Reconstruction
Structural reconstruction was the key component for this study because it acted as the primary skeletal for the geobodies modeling.A grid resolution of 50X50m was used for our Initially the farewell zone was identified within the seismic cube (in depth) along with possible faults that had cut through the zone and the horizon-fault system was incorporated within horizon modeling phase (Fig. 5).This incorporation ultimately placed the zone boundary of the Farewell and ensured the connectivity of the faults and their seals for the reservoir [21].
During the next phase, structural gridding was performed using stair-stepping of the fault frameworks within the model and then subsequent pillar gridding along the Fault Framework Model (FFM).It was observed that according to the structural model, the regional dip of the Farewell Formation was around 12° in the north and 17° in the south.Most of the faults that cut through Farewell Formation were of extensional in nature [2,21] and strikes NE-SW with an average dip angle of 60-70° [21].

Workflow
Object-Based Facies Modeling (OBFM) allowed us to populate discrete logical facies model with specific objects (geobodies such as paleo-channels etc.) which were generated and distributed stochastically [22][23][24][25][26].This study was completely based on geometrical inputs that control particular body shapes (thickness/width etc.).It was important to note that OBFM built for the study was constrained by the well log which honours the well log as well as insertion of the facies as "bodies" within the model.We also used proper erosional/replacement rules for different "geobodies" based on their spatial distribution in space.Vertical and areal trends were used for defining the spatial distribution [22,[27][28][29][30][31].
Facies modeling workflow for the Farewell Formation followed three robust steps based on the above mentioned strategies.During the initial stage, (1) depositional-bodies were generated using stochastic algorithm.(2) defining the boundaries of the interpreted geobodies, (3) finally the internal geometry and heterogeneity of the facies were developed to generate the final OBFM of the Farewell Formation (Fig. 6).
In this study, we used 3D objects like paleo-channels and simulated the channels with levees in the model.For defining the shape of the objects, 3D-pipe and ellipsoid shapes were selected to create the geobodies within the OBFM process [27,30].

Results
Three (3) depositional facies were identified and interpreted within the Farewell Formation.It is to be noted that, there were no physical samples or core cuttings available for this study, therefore the conventional approaches of core-litholog-texture analysis were not followed for this study.
Therefore, well log response along with object-based simulation were utilized to interpret all the possible lithologies and subsequent geobodies.

Meandering Channel Sand Association (Single-Stacked Association)
This facies association was made up of several, 15 to 20 m thick sandstone bodies, that showed sheet geometry in cross-section (width/thickness ratio>30).The basal bounding surfaces of sandstone bodies were flat to concave-up as seen in the well log response of the Maui wells (Fig. 7).
These basal lags were overlain by different facies, making up distinct facies successions that could be arranged into two main types.The first type was comprised of fine-to medium-grained sandstones, massive or with horizontal cross-bedding [32][33][34][35].The second type was composed of very-fine to medium-grained sandstones containing possible planar and trough cross bedding (~0.5 m thick sets) and could probably be commonly formed ripple cross-lamination [33][34].Both types showed abrupt facies shifts and could either form fining-upward cycles or be consistent in grain-size as shown in the log response of Maui-3, Maui-2 and Maui-1 wells.

Interpretation
This facies association was interpreted as fluvial channel deposits.The basal erosion surfaces, overlain by intraformational conglomerates, was associated to 5th order surfaces [36][37][38][39][40][41][42][43][44].They represent basal channel boundaries.The sandstone bodies composed of massive and parallel to low-angle cross-bedded sandstones, were interpreted to represent poorly confined sheet flood deposits [41][42].The common occurrence of both parallel-bedding (upper flow regime) and massive sandstone (hyper-concentrated flows) indicated fast and intermittent, high capacity streams.These channel deposits probably represented fields or trains of individual bedforms that accumulated predominantly by vertical aggradation.This type of architecture was similar to sand-carrying rivers developed on distal braidplains, mostly in arid regions where ephemeral streams normally form a network of shallow channels (distal, sheet flood, sand-bed rivers as explained before [41].Ancient examples of this fluvial style were studied earlier as well [45][46][47][48].Intraformational conglomerates at the base of the fluvial sandstone packages represented reworking of overbank deposits as interpreted in the log response (Fig. 7).

Braided Fluvial Channel Belt Facies Association
This facies association was composed of sheet-like sandstone bodies, 20 to 40 m thick.The sandstone bodies were bounded by a flat to concave-up erosional surface and showed internally an upward-fining in grain-size (Fig. 8).Locally, fine-grained sandstone, possible massive cross-lamination, occured at the top of units.These relatively large-scale cycles comprised several smaller-scale (5-10 m thick, mean thickness of 7.5 m), fining-upwards sub-cycles (Fig. 8).

Interpretation
The fining-upward sandstone bodies bounded by upward concave erosional surfaces (5th order bounding surfaces) [40][41] was interpreted as fluvial channel deposits.Low-angle, down-current dipping surfaces suggested the presence of large-scale, down-current accreting macroforms [41].The sheet geometry of the sand bodies (width / thickness>30), the prevailing coarse-grained nature of the deposits and the dominance of mid channel bar deposits suggested that this facies association consist of braided fluvial channel belt deposits (Fig. 9).
However, it is worth discussing the meaning of the word "channel" within the context of braided rivers for this study.At low discharge, braided rivers form a network of interconnected channels separated by sandy or gravelly bars [44].To avoid confusion, most authors adopted the term channel-belt to designate the entire river, which was internally subdivided into bars and enclosing channels [44,[49][50][51], therefore we also used channel-belt for defining the entire braided system.Herein, sandstone bodies bounded by erosional 5th order bounding surfaces (Miall 1988(Miall , 1996) ) were interpreted as braided channel belts, and were filled with bar and channel floor deposits packages of which were bounded by 4th order surfaces [39][40][41].Smaller fining-upward cycles represented the lateral and vertical juxtaposition of bars and channels within the channel-belt [51][52][53][54][55].

Floodplain Facies Association
This facies association consisted of 2 to 6 m thick, fine-to medium-grained, well-sorted (according to GR response) sandstone units, which was possibly massive in nature.Some sandstone bodies were overlain by massive or laminated mudstone, forming fining-upwards successions (Fig. 10).These sandstones were commonly associated with aeolian sand sheet deposits, and their identification was restricted to the subsurface [40-41, 49-51,54].

Interpretation
The aforementioned sedimentary features suggested a link between this facies association and ephemeral shallow streams [23,41,56].Fining-upwards cycles represented complete waning of

Sequence Stratigraphic Framework of the Farewell Formation
One of the main stratigraphic concerns of the last decades entailed the identification and mapping of genetically related units bounded at their top and base by unconformities [41,[58][59][60].
Following this trend of analyzing the geological record, a large step forward was reached through the development of the sequence stratigraphic method, which has been most widely applied to coastal and shelf deposits [58][59][60][61].Within these depositional settings, sequence accumulation and preservation were controlled by relative sea-level changes.
The identification of systems tracts within fluvial systems was based on several criteria, including geometry and stacking pattern of fluvial channels, channel to overbank deposit ratios.
Often, intervals of multi-storey and multi-lateral, amalgamated, sheet sandstone bodies, with rare over-bank deposits were interpreted to record low rates of accommodation space creation [43,58,60,[62][63].On the other hand, intervals characterized by single-storey, ribbon or sheet fluvial channels sand bodies encased within fine-grained over-bank deposits were often interpreted to record periods of higher rates of accommodation space creation [43,[63][64][65][66].The studied interval of the Farewell Formation was herein analyzed using a sequence stratigraphic approach to define depositional sequences, which were defined as "stratigraphic units composed of a relatively conformable succession of genetically related strata bounded by unconformities or their correlative conformities" [59,67].

Sequence I
Sequence I was recognized in all studied wells and its minimum thickness ranges from 30-60 m.
The lower boundary could not be precisely delineated as it interfingers with the underlying North Cape Formation.Its upper boundary consisted of a regional-scale erosional surface.Major facies associations of the Sequence I was Ephemeral Fluvial Channels mainly meandering deposits.

3D-Modeling Approach
The Facies Composite algorithm in Petrel was adopted to construct facies model of the Farewell Formation, which is an object-based facies modeling technique that can be used for a wide range of heterogeneities and depositional environments.The basic concept of the Facies Composite algorithm was that the facies within a grid model can be subdivided into a background facies and one or more object facies, which can have different shapes, sizes and orientations [22,[69][70].The Facies Composite algorithm models the geometries and distribution of the object facies [23,69,71].Facies sizes and shapes were drawn from a user-specified distribution that might differ locally within the reservoir [23,[72][73][74].The entire process was based on a 3D geocellular modeling grid for the study and the discrete well data in a suitable format.

Modeling Grids and Constraints
For the 3D lithology models, seismic-derived structure maps were used to define the base and the top of the stratigraphic interval of the formation.The top of the model was then assigned within the layering algorithm denoting arbitrary layer number 110, and the base of the model being 120, which was around 100m in thickness on an average along the studied wells (Fig. 12).Between the reservoir, 9 zones of equal thickness were constructed for better enhancement (along the vertical Z-axis) of the 3D facies grid.It was essential to define these zones within the reservoir interval to accurately model the lithologic variability within the reservoir.For the construction of the modeled grid, following process was used to create the vertical proportion curve.
A vertical proportion curve (VPC) (Fig. 13) is a cumulative probability plot that shows the percentage of lithology versus depth.In this study we characterized VPC analysis considering all the studied wells within the gridded model.To create the vertical proportion curve, the interval between Farewell Formation was subdivided into an equal number of approximately 10m thick layers.
For this reason, proportional layers were defined within this stratigraphic interval in the well logs.The proportional layers had an approximate thickness of 10m in general but the thickness varied laterally, from well to well, given the same number of layers per reservoir zone on each well log.For each layer, the percentage of each lithology (sandstone, shale, coal) in that interval was determined from the well logs.The original well-log data were used in this analysis.By combining the well log data for each layer in this way, a vertical proportion curve was created that provides a 1D vertical trend to show how the lithology percentages vary with respect to depth in this study area.
From the VPC, it was evident that 6 different geobodies were three-dimensionally distributed along the Farewell Formation.From figure 13(a) meandering channel sand proportion curve was interpreted.We interpreted that, probability of occurrence for meandering channel sands were varying in different layers of the model.Layer 111-115 had higher probability of occurrence whereas layers from 116-118 have 20-29% probability of occurrence within the model.Layer 119-120 had only 10-15% probability of having meandering channel sand deposits.For floodplain mudstones, highest percentage of occurrence concentrated within the deeper layers (30-60%) (Fig. 13b).For crevasse splay sands, probability of occurrence was very low; around 10% for all the interpreted layers of the Farewell Formation (Fig. 13c).Figure 13d represented VPC for levee sands.Probability of occurrence for levee sands within the model was very low in the deeper layers (2-8%) compared to that of the shallower layers (5-9%).Mouth bar sands had a distribution of lower VPC all along the modeled layers (Fig. 13e).Braided channel sands had higher probability of occurrence in the shallow layers, compared to the meadering ones.
To create the 3D model grid based on the VPC, facies thickness maps were generated for each of the reservoir zones.The isopach maps were added to a single reference surface in the model to create the additional structure maps (reservoir model surfaces) needed to define the 3D model grid or framework.This process ensured all structure maps to be consistent with the structural trends.
A detailed 3D facies model of 300,000 geo-gridded cells was created for 3D facies modeling.
Fine-scale layers were distributed proportionally between the key stratigraphic surfaces.Each fine layer was approximately 10m thick to capture the stratigraphic detail of both meandering and braided deposits.Lateral cell dimensions of 50m X 50m were used to capture the details of the lateral distribution of the fluvial deposits and in order to have several cells between wells.
To honour the lithologic variability in wells, lithology logs were created for each well and subsequently upscaled to the resolution of the model grid and used in the modeling process.To create the lithology logs for all 4 wells, various cutoffs for the well-log data were used for each lithology for generating point attributes for each well, that later were incorporated into the 3D model.It was to be noted that in the study area, values less than 70 API units on the gamma-ray log was defined as sandstones (North and Boering 1999;Rider 1990).Values greater than 70-90 API units on the gamma-ray log was defined as silty sandstone to clayey siltstone and values above that were all termed as shales.
During the modeling process, the scaled sand-shale ratio volume was used as a conditioning parameter to control the spatial distribution of sandstone bodies or sandstone lithology within the 3D models.The calculations were performed using the Data Analysis plug-in of Petrel (Fig. 14).As a result, if a modeled layer within the Farewell reservoir exhibits a high percentage of sandstone, the corresponding reservoir layer in the 3D model will have a high percentage of sandstone.
For the object-based lithology (sandstone, shale) modeling, the dimensions of sand bodies were derived from previous depositional and paleo-depositional studies performed on the basin [23,  around 12-15000 m perpendicular to paleoflow in the minor direction and vertically around 100 m on an average within the studied wells (Fig. 15).The few zones with longer vertical correlation lengths appeared to correspond to stacked mouth-bar deposits (multistory channel-fill deposits) rather than individual mouth bars.It was important to note that the correlation lengths do not directly provide information on the dimensions, geometry and sinuosity of the fluvial deposits.
We used sequential-indicator simulation and conditioned it to fit with the object based geobody modeling [30,75] to generate a 3D facies model for the Farewell Formation.A value for lithology was simulated in each cell of the model based on probabilities calculated from well data and user-defined inputs.The percentages of each lithology were defined for each zone based on well data.
For the sequential-indicator simulation model (indicator-based model), the global volume fraction for sandstone was 45% sandstone and 55% shale.Because indicator-based simulation used variograms (a two-point statistics) to incorporate the spatial correlation of the reservoir lithologies, the geometries of channel sand bars were, as expected, not reproduced using this method.However, the general distribution of sandstone honored all the constraints.The lower portion of the facies model (layer 115-120) of the indicator-based model was composed primarily of discontinuous sandstones and shales which whereas stacked and more continuous sandstone and interbedded shale intervals were more common in the upper part of the facies model.The upper part was therefore denominated as possible braided stacked channel sands and the lower being multistoried meandering channel sand deposits.Relatively small and discontinuous sand bodies were observed throughout the modeled interval.

Object geometry
The geometrical properties for each selected facies association in all zones were set by defining different and reasonable geometries (Fig. 15a-f).Distributary channel, the main H/C bearing facies, was defined as a backbone which was a variant of the axial shape where the object was deformed locally to follow a direction parameter.Its thickness and width remained constant along axis and the cross sectional formed as U shapes (Fig. 15f).The overbank was also defined as an axial shape with constant thickness and slight width variation.The mouth bar (Fig. 15b) was defined as an axial lobe body whose width increased then decreased along axis with a lentoid cross sectional form.It was worth noting that these definitions do not determine the size or orientation of the bodies, they only specify a prototype shape for the object and this shape was then rescaled and orientated using the size and orientation attribute values.
Evidence of possible multi-storey stacking within the sand-body profile was usually in the form of a minor gamma peaks that might result from thin units of fine-grained sediment or, more likely, horizons of intra-formational mud clasts resting on erosion surfaces at the bases of stacked storeys.
Such sandbodies range in thickness from around 10 m up to over 20 m (Fig. 15a/f).In the object-based facies modeling, it was necessary to define facies body log to split a multi-storey channel interval into different facies objects and get a better statistics of properties and thus a better simulation result [79][80].The body division results were showed along with the prospective major and minor directions in figure 15 (a-f).

Object-based simulation of the geobodies
Object-based simulation involves defining lithology (or facies, architectural element) objects with a range of dimensions and characteristic shapes that were used to populate the 3D model [80][81].This method honored geologic rules for stacking patterns and erosion and user-defined shapes that, in some cases, can produce more geologically realistic models than other modeling methods [30,75].
For this study, the modeled elements were crescent-shaped objects that represent the main reservoir elements (mouth bars) within the studied formation.The thickness of individual channel sands within the grid was modeled using a triangular distribution with values ranging from 2-15m with an average of 8.5m.Individual bar widths range from 200-400m with an average of 300m.The mouth-bar shape was modeled as a scalable crescent-shaped body based on the study of paleo-channel reconstructions of mouth bars [23,[77][78].Therefore this method used objects with distinct shapes to control the spatial continuity and distribution of lithologies (i.e., variograms were  To simplify the modeling process, separate stochastic object-based models were generated for each lithology and then merged to create the final lithology model.The two sandstone and shale models were constructed using the same ranges of object dimensions for the sandstone bodies but used different object orientations.These models were then merged to create a final model with point orientations that were consistent with outcrop measurements.The orientations of the geobodies were consistent with the approximate paleoflow direction of the Farewell Formation of the Maui Gas Field.The range in orientations produced crescent-shaped and lenticular objects that represent bar deposits on opposite sides of a sinuous meandering stream system.Using this approach, crescent-shaped sand bar objects were modeled, along with the fluvial channels; therefore, the sinuosity, amplitude and wavelength of the inferred channels that deposited the sandstones were not used as input into the modeling process.The sinuosity of Farewell Formation fluvial channels was estimated to be approximately 1.7-1.9based on sandstone body width-to-thickness ratios and paleo channel direction measurements from the 3D seismic cube [31,82].Although channel amplitude and wavelength were not modeled explicitly, the modeled object dimensions and orientations were consistent with the previous studies [23,31,77,82].With the combination of indicator and object based modeling, the final outputs were constrained by well data and seismic picks and resulted in a 3D geobody grid for the studied formation.
Using horizontal seismic patterns within a small travel-time window that followed the sedimentary facies, a stratal slice could reveal more detailed geobody distributions and sedimentary facies in much higher resolution with less ambiguity than that was possible by using classic seismic stratigraphy [89][90].

Spatial distribution and temporal evolution of incised valleys
The incised valley fill was used to identify the sediment transport direction, deduce the sedimentary dispersal system, and served as a potential high-quality reservoir [91][92][93][94].As a typical sand-prone fill and the main drainage system in the study area, the spatial distribution and temporal evolution of incised valleys were investigated in great detail.and crossline cross sections, respectively, which were used for interpreting the spatial distribution of incised valleys.
Moving the Z slices on the seismic sections from figure 16a to 16b, the numbers of incised valleys changes and their width-to-thickness ratio was evident from upstream to downstream region of the stratal slice(Fig.16).In figure 16a, at least 9 prominent incised valley fills were interpreted on the seismic section with multiple, very strong converging and branching belt-shaped amplitude anomalies in the slice from upstream to downstream (NE-SW).The anomalies had a close corresponding relationship with incised valleys on the seismic section, indicating the planar distribution of multiple incised valleys.
11 incised valleys were interpreted in figure 16b.There were also combined channel forms, branching out from main channel seen in the seismic slice.This corresponds to high amplitude anomalies (coherence attribute) in the stratal slice.It was also evident that the width-to-thickness ratio of incised valleys gradually decreases from upstream to downstream.

Temporal evolution of incised valley fills
Representative seismic profiles shown in 3-D visualization from the Farewell Formation sequences revealed the temporal evolution of the incised valley fills from the bottom to top boundary (Fig. 17).In figure 17, the bottom of the illustration was the vertical slice showing the reflection amplitude of different stages of the two major sequences.The higher amplitudes were characterized by channel sands distributed spatially.
The belt-shaped high amplitudes reflected the difference of incised valley fills in the stratal slices, indicating the differences in width, sinuosity, continuity, and extended length (Fig.

Conclusions
The following conclusions can be made from this study:-1) We identified three major depositional facies distributed over the paleo-environmental setup of the Farewell Formation.It was also observed that the current workflow of facies occurrence and probability according to the geological knowledge-base and grid sequencing showed considerable improvements within the modeled region of the Maui Gas Field.
(2) The Farewell reservoir was interpreted to be of fluvial paleo-environment in origin.
Interpretations lead us to believe that about 45% of the depositional facies are of channel sandstone origin and the rest 55% are floodplain/overbank shales.
(3) The Farewell model was sequence stratigraphically interpreted to be of Type-I origin having Lowstand Systems Tract (LST) of fluvial origin.There were combined channel-forms, branching out from main channel corresponding to high amplitude anomalies predicted to be higher quality reservoir sands.
(6) The geobody modeling process had effective advantage of quantitative propagation of uncertainty via the object-based algorithm of the identified geobodies distributed across of the formation.Thus the facies model was interpreted to be used as a significant new tool for future studies on deeper reservoirs and their development strategies for better production.

Figure. 1 .
Figure.1.Maui Gas Field and the surrounding regions of the Taranaki Basin, New Zealand (after Haque et al. 2016).The yellow colored compartments are two producing blocks of the Maui Gas Field.N-S elongated lines are the major faults cutting through the gas field.The gas field sits on the Central Graben which is located at the Southern Taranaki Basin (STB).

Figure. 2 .
Figure.2.Seismic cube of the Maui Gas Field showing well locations and 3D geomodel of the Farewell Formation.

Figure. 3 .
Figure.3.Regional Structural setup of the Taranaki Basin.Grey shading represents basement (after King and Thrasher 1996).The pre-Miocene configuration of the basin's eastern margin is uncertain (Reilly et al. 2015).Depth scale in metres below sea level.Vertical exaggeration = 5.The restorations are based on present-day influenced by contrasting Australian-Pacific plate boundary kinematics in the Late Cretaceous-Paleocene and in the Oligocene-Neogene, separated by a relatively quiescent period throughout most of the Eocene [6, 16-18].Taranaki Basin is broadly divided into Western Stable Platform and the Eastern Mobile Belt [6-7, 18], and Maui Gas Field sits on the EMB.EMB in the previous literatures reflects the composite architecture and evolution of the basin, involving several tectonic phases of differ in age, origin and style [8, 19-20].

Figure. 4 .
Figure.4.Generalized Stratigraphy of the Taranaki Basin.Farewell Formation is at the bottom of the Paleocene period of the Taranaki stratigraphy (after King and Thrasher 1996).Parts of the yellow colored (dotted) sands are Farewell Formation on the stratigraphic chart.

Figure. 5 .
Figure.5.The Farewell model is identified in accordance with the 3D seismic interpretation from the Maui cube.The mesh grill top and bottom (blue and yellow) are model boundaries whereas the different coloured vertical pillars are the faults interpreted along the entire 3D structural grid of the Maui Gas Field.The seismic surface (pink coloured) denotes the interpreted Farewell reservoir zone.

Figure. 6 .
Figure.6.Standardized workflow for the 3D facies modeling of the Farewell Formation, Maui Gas Field, Taranaki Basin.The entire workflow is divided into two segments; first being the construction of the 3D geo-grid of the gas field (left) followed by the objected-based process for building the facies model of the Farewell Formation.

Figure. 7 .
Figure.7.Meandering channel sand deposits in Maui-3 and Maui-2 wells.Log response clearly shows multiple fining-upwards cycles as interpreted on the GR logs of both the wells.Stacked sand bodies are evident from the log interpretation.Aggrading channel-fills are also seen in Maui-3 well.

Figure. 8 .
Figure.8.Braided channel sand deposits in Maui-3, Maui-2 and Maui-1 wells.Log response clearly shows fining-upwards cycle at the top of the formation in each well as interpreted on the GR logs.Braided channel sand is thickest in Maui-2 well and thins Northeast towards Maui-3 well.

Figure. 9 .
Figure.9.Braided sand bodies as interpreted in the well log responses of the Maui wells.Channel sands are developed from Northeast and have flowed towards Southwest within the Maui Gas Field (top).Note the extensions of the braided channels (Maui-3 and Maui-2) and a distributary from the main braided channel towards Southwest as seen in Maui-1 well (bottom).Red color indicates braided channel sands whereas light blue indicating crevasse splay sands.

Figure. 10 .
Figure.10.Floodplain deposits are spatially distributed in the facies model.(a) to (d) are suggestive floodplain deposits in different layers within the Farewell Formation; (a) being the shallowest and (d) being the deepest layer of the Farewell Formation.Please note that the floodplain mudstones also define the extent of the sand deposition within the Maui Gas Field as seen in the figure 10(a) to (d).Figure 10(a) represents layer 110 (3200 TVDSS) whereas figure 10(d) represents layer 120 (3300TVDSS) in the model.

Figure. 11 .
Figure.11.Sequence Stratigraphy of the Farewell Formation.This formation can be subdivided into two major unconformity based, regional scale depositional sequences as interpreted from the well logs.Sedimentation for the meandering deposits are associated with Sequence I whereas braided channel sands are associated with Sequence II.

Sequence I was interpreted
to be a Type-I Falling Stage Systems Tract FSST) boundary marked by an unconformity.This was characterized by stream rejuvenation and fluvial incision, sedimentary bypass of the shelf, and abrupt basin-ward shift of facies and coastal onlap.This might be formed when the rate of eustatic fall exceeds the rate of basin subsidence at the depositional shoreline break, producing a relative fall in sea level below the position of the paleoshoreline [61].3.4.1.1.Depositional architecture Sequence I displayed a progradational facies succession.The base was dominated by fine-grained shallow marine deposits of the North Cape Formation that were overlain by fluvial strata.While fluvial flow was dominantly from NE to SW, indicating sediment transport towards the southern border of the basin.Fluvial deposits were laterally and vertically adjacent, producing a complex depositional architecture.Both single storied and stacked meandering sand packages occur within the sequence consisting of 2-15m thick and >1 km in horizontal directions towards the X and Y axis of the model, the upper boundaries were fluvial scour surfaces with <5 m of erosional relief into the Aeolian deposits.Wherever single stack fluvial erosion was substantial, multiple channel deposits were severely to completely truncated (Fig.11).

Figure. 12 .
Figure.12.3D geocellular facies distribution of the Farewell Formation.The image below represents the facies model in SIMBOX view, vividly displaying interpreted facies in X-Y-Z axis within the facies model of the Farewell Formation.

Following
the structural restoration, the digital reconstruction of the boundaries of the sedimentary bodies involved the following steps: (a) visual construction of structured geobodies along with defined geometries of the interpreted lithology (b) variogram analysis of the geobody dimensions (c) incorporation of the paleo-depositional interpretation within each geobody and (d) construction of a consistent 3D subsurface-based model.The process of reconstructing the sedimentary bodies was used as a quality control on the existing field correlation.This reconstruction supported the interpreted rectilinear geometry of the channeled sandbodies.

Figure. 14 .
Figure.14.Facies proportions thickness Vs probability of occurrence for the Farewell Formation.Colors represent different geobodies interpreted in the model.

75- 78 ]
. For the indicator based modeling, the spatial correlation of the lithologies of the Farewell Formation was quantitatively analyzed through variogram analysis.Experimental variograms were calculated separately for each reservoir zone and lithology and were fitted with exponential and spherical variogram models.Correlation lengths of the model variograms ranged from approximately 25-35000 m parallel to paleoflow (Northeast to Southwest) in the major direction and

Figure. 15 .
Figure.15.Object geometries (a-f) of the interpreted geobodies of the Farewell Formation.Each geobody represents its spatial distribution within the formation along the X-Y-Z axis of the modeled grid.Braided channel sands to meandering channel deposits have been modeled for the analysis.Figure 15g represent the entire facies model combined with all possible geobodies of the Farewell Formation, Maui Gas Field.

Figure. 16 .
Figure.16.Seismic attribute (coherence) showing the paleochannels within the Farewell Formation.Solid black lines are possible fluvial channels as interpreted on the Z slice of the seismic cube.

Figure. 17 .
Figure.17.Vertical seismic profiles (in TWT) for the Farewell Formation.AA' and BB' profiles represent major channel forms along the facies model of the formation.Note that the interpreted paleochannels (BB') have more width-thickness ratio compared that of the westward profile (AA').

3. 9 . 1 . 1 .
Figure 16 explained the spatial distribution of incised valleys using coherence attribute.Coherence attribute was used in the study because of its effectiveness in identifying edge detection along with possible channel boundaries.In figure 16a to figure 16b, the bottom of each figure was the stratal slice showing the reflection amplitude of the bottom boundary of sequence II; in inline figure17 were almost faded in the bottom vertical slice; whereas incised valleys developed at the top slice were characterized by branching and converging.From the interpretation of channel distribution and occurrence, the variations in number, width, and sinuosity of incised valleys indicated the evolution of incised valleys from initiation to extinction[91,[94][95].

( 4 )
It was also observed that the lower portion of the geobody facies model (layer 115-120) was composed primarily of discontinuous sandstones and shales.Sandstone and interbedded shale intervals were more common in the upper part of the facies model.The upper part was therefore denominated as possible braided stacked channel sands and the lower being multistoried meandering channel sand deposits.(5)We modeled at least eleven prominent incised valley fills on the seismic section with multiple, very strong converging and branching belt-shaped amplitude anomalies in the slice from upstream to downstream (NE-SW).The anomalies had close corresponding relationship with incised valleys on the seismic section, indicating the planar distribution of multiple incised valleys.Preprints (www.preprints.org)| NOT PEER-REVIEWED | Posted: 15 January 2018 doi:10.20944/preprints201801.0123.v1