Reflectivity and spectral attribute-based reservoir characterization: a case study from "MUN" onshore Niger delta field, Nigeria

Abstract

Reservoir characterization plays a significant role in the exploration, development, and production of hydrocarbon reservoirs. The use of integrated approach in characterization improves on the accuracy, certainty, and robust interpretation of reflectivity data. The study aimed to integrate reflectivity and spectral attributes to adequately characterize hydrocarbon reservoirs in MUN onshore Niger delta field. Well log and rock physics analyses identified and delineated reservoirs, discriminated lithology, characterized fluid, and established relations between elastic and reservoir properties for field-wide interpretation of the reflectivity data. The wells were tied to the reflectivity data, and H4 seismic horizon was mapped. Subsequently, sweetness, reflectivity and spectral attributes were extracted along H4 horizon after inversion and spectral decomposition of the reflectivity data and independently interpreted. Channel-like structure with high-amplitudes, low-to-moderate acoustic impedance (Ip), lambda–rho (λρ) and mu–rho (µρ) attribute values, and high-amplitude low-frequency spectral attributes (15–35 Hz), respectively, characteristics of hydrocarbon saturated channel sands were delineated. The channel reservoir sand is thick, porous with low volume of shale and low to moderate water saturation. Analysis of data further shows that reflectivity attributes provided a better description of fluid characteristics than the sweetness and spectral attributes, but are less sensitive to structure and exaggerated the shape and limits of the channel sands. However, the spectral attributes seem to be more robust than the reflectivity attributes in providing subtle structural and stratigraphic details of the reservoir as well as delineating by-passed hydrocarbons in the field.

Introduction

Reflectivity and spectral attributes derived from well and seismic dataset have become potent and robust tools in reservoir characterization, relating seismic responses to reservoir properties. Predrill reservoir characterization is key in estimating the potential for hydrocarbon accumulation and production (Mohamed et al. 2016; Ehirim and Akpan 2017). Characterization entails adequate description of the reservoir properties and structure for improved exploitation and production of hydrocarbons.

Reservoirs are characterized by their properties such as fluid types, water saturation, volume of shale, porosity, permeability, pressure, temperature, thickness, and depth. Therefore, hydrocarbon reservoir characterization is the mapping of geometry, structure/stratigraphy and estimation of relevant properties of the potential reservoir (Eshimokhai and Akhirevbulu 2012; Ajisafe and Ako 2013; Edigbue et al. 2014; Mahmoud et al. 2016; Sanuade et al. 2017).

Seismic reflectivity is an interface related property of the seismic signal, which occurs due to the contrast in elastic properties across a geologic interface. The nature of the interface, source wavelet, and the contrasting properties of the geologic layers determine the magnitude, phase and polarity of the reflectivity signal. This signal carries information about the lithology, pore fluids, structure/stratigraphy and geometry of the reservoir (Hilterman 2001).

Through acoustic impedance inversion, these reservoir properties can be estimated from reflectivity attributes extracted from the resulting acoustic impedance data. Reflectivity attributes such as acoustic impedance, Poisson impedance, VpVs ratio, Lambda rho, and Mu rho have been employed routinely to distinguish lithology and characterize pore fluids (Hamada 2004; Ekwe et al. 2012; Bello et al. 2015).

Spectral attributes derived from decomposition of seismic reflectivity data in time to their component frequency bands is effective for qualitative and quantitative interpretation of bed thickness, lithology, discontinuities, channels, pore fill and structure (Marfurt and Kirlin 2001; Ahmad and Rowell 2012; Wei et al. 2019), as well as for direct hydrocarbon indicator for gas charged reservoirs (Sinha et al. 2005). Amplitude–frequency anomalies have always been interpreted as hydrocarbon saturated porous thin/thick reservoir beds (Ahmad and Rowell 2012; Ehirim and Akpan 2017).

Integrating reflectivity and spectral attributes in reservoir studies enhances hydrocarbon detection, delineation, and reservoir characterization (Tian et al. 2010). These therefore ensures that fails hydrocarbon shows and drilling of dry holes are avoided and consequently, informed decisions regarding hydrocarbon prospectivity, producibility and drilling locations for wells could be made.

The present study is aimed at integrating reflectivity and spectral attributes derived from well and 3D Post-Stack time migrated (PSTM) reflectivity data to improve reservoir detection, delineation and characterization in MUN Niger Delta Onshore field. This study was carried out in the coastal swamp depobelt of the Niger Delta (Fig. 1). The Niger Delta Basin, situated at the apex of the Gulf of Guinea on the West Coast of Africa, is one of the most prolific deltaic hydrocarbon provinces in the World (Ajisafe and Ako 2013). It covers an area within longitude 4° E—9° E and latitudes 4° N—9° N. The swamp depobelt is characterized by rain forest and mangrove vegetations, with landscape incised by numerous brackish rivulets and creeks, high precipitation and relative humidity.

Fig. 1
figure1

Location map of the study area (Reijers 2011)

Geology of the study area

The Niger Delta is composed of an overall regressive clastic sequence of Tertiary age (Evamy et al. 1978). The Delta has prograded southwards overtime forming depobelts, namely Northern, Ughelli, Central swamp, Coastal swamp and Offshore depobelts, each with its own sedimentation, deformation and petroleum history (Fig. 2), that represents the most active portion of the delta at each stage of its development. These depobelts coalesced in time to form one of the largest and massive regressive deltas in the world (Doust and Omatsola 1990).

Fig. 2
figure2

Depobelts and direction of sediment deposition in Niger delta (Tuttle et al. 1999)

The Tertiary section of the Niger Delta is divided into three formations: Benin, Agbada (the reservoir rock) and Akata (the source rock and reservoir in deep offshore) Formations (Fig. 3). The Niger delta province has been identified to have only one petroleum system referred to as the Akata-Agbada Tertiary Niger Delta petroleum system (Tuttle et al. 1999). Early global plate tectonic movements and local gravity tectonics caused deformation of the Agbada Formation, resulting into anticlinal and growth fault structures in the formation. These structures controlled the subsequent sedimentation, migration and accumulation hydrocarbons in the Agbada Formation.

Fig. 3
figure3

Geologic map of Niger delta (Tuttle et al. 1999)

Methodology

Data used in this study are well logs with corresponding check shot information from wells 14, 15 and 16 and 3D post-stack time migrated (PSTM) reflectivity volume of “MUN” onshore Niger delta field. These datasets were loaded into the software (Petrel and HR) and quality checked for spurious and missing data points. Well log analysis, petrophysical modelling, reflectivity inversion and spectral decomposition are the methods adopted to achieve the aim of this study (Fig. 4).

Fig. 4
figure4

Schematic workflow of the study

The well logs: density (RHOB), Gamma ray (GR), resistivity (RT) and sonic (Vp), were used to discriminate lithologies into sand and shale formations and characterize pore fluids at the wellbore (Fig. 5).

Fig. 5
figure5

Well logs, mapped reservoirs and correlation across the wells

Pseudo-logs of well-based seismic elastic properties were generated, cross-plotted and color coded with reservoir properties in 3-D cross-plot space with the aim of discriminating lithology, characterize pore fluids and establish relations between seismic elastic properties and reservoir properties. Well correlation, time-depth conversions and seismic to well tie were done to facilitate horizon mapping on the seismic data. Subsequently, three horizons were mapped, namely HA, H4 and H7, guided by respective reservoir sand markers in the wells (A, 4 and 7). For the purpose of this study, sand 4 and H4 seismic horizon were used for analysis.

Amplitude-related attribute (sweetness) was extracted along H4 horizon to identify and delineate hydrocarbon zones. Subsequently, the reflectivity data was inverted into acoustic impedance volume using model-based inversion scheme which involves generating low-frequency model, guided by well-logs and interpreted horizons. The acoustic impedance attribute was extracted along H4 horizon from the impedance volume. Other attributes such as lambda–rho and mu–rho were generated to fieldwide characterize the reservoir.

Furthermore, spectral frequency volume was generated using a continuous wavelet transform (CWT) decomposition method. The method convolves the time-dependent reflectivity data with a suitable mother wavelet to produce a frequency volume. Spectral attribute slices were then extracted from the volume along H4 seismic horizon but within the seismic band width of the seismic data and analyzed to delineate possible structures, evaluate indication of hydrocarbons and determine hydrocarbon presence and distribution in the field.

Presentation of results

The results of well log evaluation, petrophysical modelling (cross-plots), reflectivity inversion and spectral decomposition are presented and analyzed. Three reservoirs were delineated by low gamma ray (< 75 API) and high resistivity (> 500 Ωm) values with average thickness of 67 m (Dagogo et al. 2016).

Well-based seismic elastic properties were cross-plotted and color coded with reservoir properties in 3D cross-plot domain. Results show that Ip versus λρ and µρ versus λρ property pairs were more robust in lithology identification and fluid prediction than other cross-plotted property pairs. Analysis of the results show that hydrocarbon saturated sands (black oval) depict low λρ and low to moderate Ip and µρ values corresponding to low Sw and Vsh and high ϕ (Figs. 6, 7). Brine sands (blue oval) and shale (purple oval) have high Ip, λρ and µρ corresponding to high Sw and Vsh, and low ϕ values (Ehirim and Dagogo 2016).

Fig. 6
figure6

Cross-plots of acoustic impedance (Ip) versus lambda–rho (λρ), color coded with a water saturation, b porosity and c shale volume for SAND 4

Fig. 7
figure7

Cross-plots of mu–rho (µρ) versus lambda–rho (λρ), color coded with a water saturation, b porosity and c shale volume for SAND 4

Time structure map extracted along H4 seismic horizon show gradual topographical gradient NE to SW (Fig. 8). The wells are in moderate topography between low time (SW) and high time (NE), bounded by two major NW–SE growth faults in the NE and SW of the map, respectively.

Fig. 8
figure8

Time structure map along H4 horizon

Sweetness attribute analysis along H4 seismic horizon show elongated high-amplitude anomalous zone (sweet zone) especially around the well locations bounded by growth faults to the NE and SW (Fig. 9), interpreted as a channel sand. Sweet zone is a direct indicator of thick and porous hydrocarbon saturated sand (Hart 2008). Amplitude brightening to the NE of the existing wells in the channel could be indications of by-passed hydrocarbons. The reduction in amplitude to NW and SE of the channel sand shows the limit of hydrocarbon saturated channel sands in the field.

Fig. 9
figure9

Extracted sweetness attribute map along H4 horizon

The reflectivity attributes [acoustic impedance (Ip), lambda–rho (λρ) and mu–rho (µρ)] extracted along H4 seismic horizon are characterized by elongated structures of low attribute values interpreted as channel sands with probable hydrocarbon saturation (Fig. 10a–c).

Fig. 10
figure10

Rock attributes horizon maps showing a acoustic impedance, b lambda–rho and c mu–rho along H4 horizon

The Ip and µρ sections exhibit similar amplitude patterns. They delineated two parallel NE–SW channel structures and with the wells lying on low attribute values. These structures are elongated and well resolved in the µρ than Ip section. However, the λρ section show complete disappearance of the NW channel sand structure while the SE channel structure was well imaged. Wells 14 and 15 lies on low λρ values while well 16 lie on high λρ in contrast to the Ip and µρ sections.

Ip is sensitive to rock matrix and fluid in the rock formation and thus its low value is a reflection of both sand formation and hydrocarbon fluid constituent. On the other hand, low λρ and µρ distinctly reflect hydrocarbon charged sands and rock matrix, respectively. Therefore, low to moderate Ip, λρ and µρ attribute values suggest porous hydrocarbon saturated reservoir channel sands. However, the λρ section show that well16 and the surrounding vicinity lie on high λρ value indicating that this well and their surroundings have watered out, while wells 14 and 15 lie on low and moderate λρ values, respectively, indicating hydrocarbon saturation. Moderate to high values of Ip, λρ and µρ are associated with brine saturated sands or shale (Goodway et al. 1997; Ehirim and Dagogo 2016; Dagogo et al. 2016).

The channel sand structure SE of the existing wells exhibit low attribute values of Ip, λρ and µρ and is interpreted as a by-passed play. The NW–SE growth faults bounding the channels as seen in the structural and sweetness maps were poorly imaged in the Ip, λρ and µρ sections. This suggests that these attributes are fairly sensitive to structure and could hardly be used for structural analysis.

Spectral attribute slices of 15–35 Hz frequency band extracted along H4 seismic horizon delineated a high-amplitude low-frequency anomalous channel structure trending NE–SW and bounded by growth faults to the NE and SW, respectively, indicating porous and thick hydrocarbon reservoir saturated sands (Fig. 11). The most resolute spectral attribute is the 25 Hz attribute, which coincidentally is the dominant frequency of the seismic data. Wells 14 and 15 lies on this high-amplitude low-frequency anomaly while well 16 lie on low-amplitude low-frequency zone characteristic of hydrocarbon depleted zone (Othman et al. 2016; Ehirim and Akpan 2017). East of the producing wells on the channel structure is characterized by low-frequency high-amplitude anomalies, which could likely be by-passed hydrocarbon charged sands.

Fig. 11
figure11

Spectral frequency attribute slices, 15–35 Hz along H4 horizon

Discussion of results

Reflectivity and spectral attributes analyses was carried out in MUN onshore Niger delta field using well and 3D (PSTM) reflectivity data sets to identify hydrocarbon charged reservoir channel sands, characterize the reservoir channel sand and delineate by-passed hydrocarbons. Three reservoirs (SAND A, 4 and 7) were delineated by low GR and high resistivity log values across the three wells with an average thickness of 67 m. SAND 4 reservoir was used for the present analyses because of its excellent reservoir properties. Petrophysical analyses reveal that SAND 4 reservoir is dominantly sandy with low volume of shale and high porosity, characteristic of the Agbada Formation.

Seismic elastic properties were cross-plotted at SAND 4 reservoir interval (Ip vs λρ and µρ vs λρ), color coded with the reservoir properties (ϕ, Sw and Vsh) in 3D cross-plot domain. The crossplots delineated lithologies, characterize pore fill and established relations between seismic elastic properties and reservoir properties at the well scale. The cross-plot of Ip versus λρ delineated hydrocarbon charged sands with low λρ and low to moderate Ip property values, while µρ versus λρ cross-plot delineated the reservoirs’ sand matrix by low μρ and low λρ for hydrocarbon charged sands (Goodway et al. 1997; Ehirim and Dagogo 2016).

Sand matrix is usually rigid and exhibit high μρ values (Chopra and Marfurt 2007). However, the low µρ sands delineated in this study could be attributed to the reservoir sands being unconsolidated (Goodway et al. 1997; Dagogo et al. 2016). The low λρ and low to moderate Ip associated with hydrocarbon saturated sands, corresponds to low water saturation (Sw) and volume of shale (Vsh) and high reservoir porosity (ϕ). These well-based established relations between the seismic elastic and reservoir properties were then used to constrain the interpretation of reflectivity attributes in order to characterize the reservoir field wide and delineate by-pass hydrocarbons.

Sweetness reflectivity attribute extracted along H4 horizon delineated channel-like structure of hydrocarbon saturated sands as high-amplitude zones (sweet spots). The high-amplitude anomalous zones show non-uniformity in amplitude intensity distribution, suggesting probable pore fill changes, especially around the producing wells (Hart 2008; Ahmad and Rowell 2012).

Inversion of reflectivity data and extraction of Ip, λρ and µρ reflectivity attributes along H4 seismic horizon delineated channels-like structures of probable hydrocarbon saturation. These structures are elongated and more resolved in the µρ than Ip section. This could be explained by the heightened sensitivity of µρ to lithology than Ip. Wells 14, 15 and 16 lies on low Ip and µρ attribute values. However, the λρ section show complete disappearance of the NW channel sand structure because of hydrocarbon depletion, while the SE channel structure saturated with hydrocarbons was well imaged. Wells 14 and 15 lies on low λρ values while well 16 lie on high λρ.

Low to moderate Ip, λρ and µρ attribute values suggests porous hydrocarbon saturated reservoir channel sands with low Sw and Vsh and high ϕ, in line with the results of the well-based cross plot analyses (Goodway et al. 1997; Ekwe et al. 2012; Dagogo et al. 2016). However, the λρ section show that well 16 and the surrounding vicinity lie on high λρ value indicating that this well and their surroundings have watered out, while wells 14 and 15 lie on low and moderate λρ values, respectively, suggesting hydrocarbon saturation. This explains the disappearance of the NW channel sand structure on the λρ section as was initially observed on the Ip and µρ sections. The channel sand structure SE of the existing wells exhibit low attribute values of Ip, λρ and µρ and is therefore, interpreted as a by-passed play. The NW–SE growth faults bounding the channels as seen in the structural and sweetness maps were poorly imaged in the Ip, λρ and µρ sections. This suggests that these attributes are fairly sensitive to structure and could hardly be used for structural analysis.

The spectral attribute slices (15–35 Hz) extracted along H4 horizon via spectral decomposition of the reflectivity data, delineated thick and porous channel-like structure of hydrocarbon saturated sands as high-amplitude low-frequency anomalies (Partyka et al. 1999; Marfurt and Kirlin 2001; Ehirim and Akpan 2017). The most resolute and robust spectral attribute is the 25 Hz attribute which coincidentally is the dominant frequency of the seismic data. Wells 14 and 15 lies on this high-amplitude low-frequency anomaly while well 16 lie on low-amplitude low-frequency zone characteristic of hydrocarbon depleted zone in line with the result of the λρ attribute.

However, the non-uniformity in the intensity of the spectral anomalies suggests thickness and pore fill variations along the channel structure in the field (Partyka et al. 1999; Othman et al. 2016). By-passed hydrocarbon resources were delineated east of the existing wells in agreement with the result of the reflectivity attributes.

Conclusion

Integrating reflectivity and spectral frequency attributes have greatly improved reservoir detection, delineation and characterization in MUN onshore Niger delta field. Hydrocarbon saturated channel sands were independently mapped by high amplitudes (sweetness), low reflectivity attribute values and high-amplitude low-frequency spectral attributes.

Analysis of data further show that Reflectivity attributes provided a better description of hydrocarbon content than the sweetness and spectral attributes, but are less sensitive to structure and exaggerated the shape and limits of the channel sands. However, spectral attributes seem to be more robust than the reflectivity attributes in providing subtle structural and stratigraphic details of the reservoir as well as delineating by-passed hydrocarbons in the field.

References

  1. Ahmad MN, Rowell P (2012) Application of spectral decomposition and seismic attributes to understand the structure and distribution of sand reservoirs within Tertiary rift basins of Gulf of Thailand. Search and Discovery. Article #40992

  2. Ajisafe YC, Ako BD (2013) 3D seismic attributes for reservoir characterization of “Y” field Niger delta. IS OR J Appl Geol Geophys 1(2):23–31

    Google Scholar 

  3. Bello R, Igwenagu CL, Onifade YS (2015) Cross plotting of rock properties for fluid and lithology discrimination using well data in a Niger delta oil field. J Appl Sci Environ Manag 19(3):536–546

    Google Scholar 

  4. Chopra S, Marfurt KJ (2007) Seismic attributes for prospect identification and reservoir characterization. Geophysical Development Series, No. 11. Houston, SEG Book, p 382

    Google Scholar 

  5. Dagogo T, Ehirim CN, Ebeniro JO (2016) Enhanced prospect definition using well and 4D seismic data in a Niger delta field. Int J Geosci 7:977–990

    Article  Google Scholar 

  6. Doust H, Omatsola M (1990) Divergent and passive margin basins. Am Assoc Pet Geol Bull 48(1):201–238

    Google Scholar 

  7. Edigbue PI, Komolafe AA, Adesida AA, Itamuko OJ (2014) Hydrocarbon reservoir characterization of “Keke” field, Niger delta using 3 seismic and petrophysical data. Am J Sci Ind Res 5(2):73–80

    Google Scholar 

  8. Ehirim CN, Akpan AS (2017) Continuous wavelet transform based spectral decomposition of 3D seismic data for reservoir characterization in Oyi Field, se Niger Delta. Am J Appl Sci 14(8):766–771

    Article  Google Scholar 

  9. Ehirim CN, Dagogo T (2016) Subsidence inferred from a time lapse reservoir study in a Niger delta field, Nigeria. Curr Res Geosci 6(2):98–106. https://doi.org/10.3844/ajgsp.2016.98.106

    Article  Google Scholar 

  10. Ekwe AC, Onuoha KM, Osayande N (2012) Fluid and lithology discrimination using rock physics modelling and lambdamurho inversion: an example from onshore Niger Delta, Nigeria. Search and Discovery. Article #40865. In: AAPG international conference and exhibition, Milan, Italy

  11. Eshimokhai S, Akhirevbulu OE (2012) Reservoir characterization using seismic and well logs data (a case study of Niger Delta). Ethiop J Environ Stud Manag EJESM 5(4):597–603

    Google Scholar 

  12. Evamy BO, Herembourne J, Kameline P, Knap WA, Molloy FA, Rowlands PH (1978) Hydrocarbon habitat of Tertiary Niger delta. Am Asso Petrol Geol Bull 62:1–39

    Google Scholar 

  13. Goodway W, Chen T, Downton J (1997) Improved AVO fluid detection and lithology discrimination using Lamé petrophysical parameters; λρ, μρ & λ/μ fluid stack from P and S inversions. In: 67th annual international meeting, SEG, Expanded Abstracts, pp 183–186

  14. Hamada GM (2004) Reservoir fluids identification using Vp/Vs ratio. Oil Gas Sci Technol 59(6):649–654

    Article  Google Scholar 

  15. Hart BS (2008) Channel detection in 3-D seismic data using sweetness. AAPG Bull 92(6):733–742

    Article  Google Scholar 

  16. Hilterman FJ (2001) Seismic amplitude interpretation. SEG/EAGE Distinguished Instructor Course No. 4

  17. Mahmoud M, Ghorab M, Shasly T, Shibl A, Abuhagaza AA (2016) Reservoir characterization utilizing the well logging analysis of Abu Madi Formation, Nile Delta, Egypte. Egypt J Pet. https://doi.org/10.1016/j.ejpe.2016.11.003

    Article  Google Scholar 

  18. Marfurt KJ, Kirlin RL (2001) Narrow-band spectral analysis and thin-bed tuning. Geophysics 66:1274–1283

    Article  Google Scholar 

  19. Mohamed AK, Ghazala HH, Mohamed L (2016) Integration between well logging and seismic reflection techniques for structural analysis and reservoir characterizations, Abu El Gharadig basin, Egypt. NRIAG J Astron Geophys. https://doi.org/10.1016/j.nrjag.2016.07.003

    Article  Google Scholar 

  20. Othman AAA, Fathy M, Maher A (2016) Use of spectral decomposition technique for delineation of channels for solar gas discovery, offshore West Nile Delta, Egypt. Egypt J Pet 25:45–51

    Article  Google Scholar 

  21. Partyka G, Gridley J, Lopez J (1999) Interpretational applications of spectral decomposition in reservoir characterization. Lead Edge 18:353–360

    Article  Google Scholar 

  22. Reijers TJA (2011) Stratigraphy and sedimentology of the Niger Delta. Geologos 17(3):133–162. https://doi.org/10.2478/v10118-011-0008-3

    Article  Google Scholar 

  23. Sanuade OA, Akanji AO, Oladunjoye IM, Olaojo AA, Fatoba JO (2017) Hydrocarbon reservoir characterization of “AY” field, deep-water Niger Delta using 3D seismic and well logs. Arab J Geosci 10:151

    Article  Google Scholar 

  24. Sinha S, Routh PS, Anno PD, Castagna JP (2005) Spectral decomposition of seismic data with continuous wavelet transform. Geophysics 70(6):P19–P25. https://doi.org/10.1190/1.2127113

    Article  Google Scholar 

  25. Tian L, Zhou D, Lin G, Jiang L (2010) Reservoir prediction using Poisson impedance in Qinhuangdao, Bohai Sea. In: SEG Denver 2010 annual meeting, pp 2261–2264

  26. Tuttle MLW, Brownfield ME, Charpentier RR (1999) The Niger Delta Petroleum System: Niger Delta Province, Nigeria, Cameroon, and Equatorial Guinea, Africa. Tertiary Niger Delta (Akata-Agbada) Petroleum System (No. 701901). Open-File Report 99-50-H

  27. Wei L, Dali Y, Wenfeng W, Wurong W, Shenghe W, Jian L, Depo C (2019) Fusing multiple frequency-decomposed seismic attributes with machine learning for thickness prediction and sedimentary facies interpretation in fluvial reservoirs. J Pet Sci Eng 177:1087–1102

    Article  Google Scholar 

Download references

Acknowledgements

We are grateful to Shell Petroleum Development Company of Nigeria for providing the 3D PSTM data for the study. Our thanks also go to World Bank, Africa Centre of Excellence in Oilfield Chemicals Research (ACE-CEFOR), UNIPORT; Jomo Kenyatta University of Agriculture and Technology (JKUAT) and RUFORUM for their support.

Author information

Affiliations

Authors

Corresponding author

Correspondence to J. M. Munyithya.

Additional information

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Munyithya, J.M., Ehirim, C.N. & Dagogo, T. Reflectivity and spectral attribute-based reservoir characterization: a case study from "MUN" onshore Niger delta field, Nigeria. J Petrol Explor Prod Technol 10, 2195–2205 (2020). https://doi.org/10.1007/s13202-020-00922-4

Download citation

Keywords

  • Spectral attributes
  • Reflectivity attributes
  • Reservoir characterization
  • Cross-plots