Land surface dynamics and environmental challenges of the Niger Delta, Africa: Remote sensing-based analyses spanning three decades (1986–2013)

Applied Geography 53 (2014) 354e368

Contents lists available at ScienceDirect

Applied Geography journal homepage: www.elsevier.com/locate/apgeog

Land surface dynamics and environmental challenges of the Niger Delta, Africa: Remote sensing-based analyses spanning three decades (1986e2013) Claudia Kuenzer a, *, Sybrand van Beijma b, Ursula Gessner a, Stefan Dech a, c a b c

Earth Observation Center, EOC, of the German Aerospace Center, DLR, Oberpfaffenhofen, 82234 Wessling, Germany Airbus Defence and Space, Europa House, Southwood Crescent, Farnborough, GU14 0NL, United Kingdom Department of Geography and Geology, University of Wuerzburg, Wuerzburg, Germany

a b s t r a c t Keywords: Niger Delta Remote sensing Oil industry Coastal change Gas flaring

The Niger Delta, the largest river delta on the African continent, is one of the most densely populated river deltas globally and hosts the world's third largest mangrove forest. It is a major biodiversity hot spot of our planet. At the same time the delta is home to Africa's largest oil reserves and responsible for a skyrocketing GDP development of Nigeria since the 1970s. Nigeria ranks 13th among all oil producing countries, but oil exploitation also brought with it severe environmental degradation, leading to the delta's nomination for a place on the top 10 list of the “World's Worst Polluted Places Report” in 2013. Despite the outstanding importance of the region for Nigeria, Africa, and the international community most studies published focus mainly on topics of geology, geochemistry, and environmental toxicology. Studies employing earth observation satellite data to assess Niger Delta dynamics are rare. This paper aims at contributing to an overview of Niger Delta geography and environmental threats and challenges, as well as to an understanding of Niger Delta land surface dynamics from 1986 to 2013. Covering the complete delta, we present results of land cover change analyses, results of an assessment of coastline dynamics, as well as the manifestation of oil exploitation activity as expressed via oil access canal dredging and gas flaring, monitored within the 27 year time span investigated. © 2014 Elsevier Ltd. All rights reserved.

Introduction: geography of the Niger Delta and socioecological threats Geography of the Niger Delta The Niger Delta in Nigeria is Africa's largest river delta and covers an area exceeding 29,900 km2 (Goudie, 2005) (see Fig. 1). The Niger River discharges on average over 30,000 m3 of water per second into the Gulf of Guinea. Currently, 20% of Nigeria's population e over 30 million people e live in the Niger Delta (National Bureau of Statistics, 2013). The largest city in the Delta is Port Harcourt, with a population exceeding one million inhabitants. Delta genesis started in the Cretaceous (Short & Staeuble, 2004) and progressed according to marine transgressive and regressive cycles of differing durations in response to eustatic sea level changes. The horizontal sediment structure of the delta is

* Corresponding author. Tel.: þ49 (0)8153 28 3280; fax: þ49 (0)8153 28 1458. E-mail address: [email protected] (C. Kuenzer). http://dx.doi.org/10.1016/j.apgeog.2014.07.002 0143-6228/© 2014 Elsevier Ltd. All rights reserved.

characterized by different marine and fluvially deposited layers composed of sand, silt, and clay (Abam, 1997). A most comprehensive and novel stratigraphy and sedimentology of the Niger Delta has recently been published by Reijers (2011). Geomorphologically, the recent Niger Delta can be categorized into three different regions: the continental part of the delta, the transitional area dominated by land and ocean interactions in the coastal zone, and the delta's marine territories (Ugbe, 2011). This humid region receives between 2400 and 4200 mm of precipitation per year, mainly during the rainy season. Flora and fauna in the delta are very diverse. The largest mangrove forest of Africa e and the third largest globally e can be found in the delta and comprises an area between 5000 and 8600 km2 depending on the literature source (Fatoyinbo & Simard, 2012; Isebor & Awosika, 1993; Ohimain, 2003). The Niger Delta region is characterized by an extraordinary aquatic and terrestrial biodiversity, above all related to the aforementioned mangrove forest areas, which also provide a substantial number of ecosystem services (Kuenzer & Quoc 2013; Prince & Arokoyu, 2010; Quoc, Oppelt, & Kuenzer, 2012). But not only the near-coastal areas are

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

355

Fig. 1. Location of the Niger Delta in Nigeria. The Niger Delta consists of seven states: Anambra, Delta, Imo, Bayelsa, Rivers, Abia, and Akwa Ibom. The Niger Delta also hosts the large cities of Warri, Yenagoa, Aba, and Port Harcourt.

relevant biodiversity hot spots. Important wetlands and swamp forests, such as the Apoi Creek Forest Reserve, Lake Oguta, and the Upper Orashi Forest Reserve were all designated as significant wetland areas under the Ramsar Convention in 2008. The seemingly most valuable natural resource in the Niger Delta is its enormous oil and gas deposit (Ajibade & Awomuti, 2009; Ikelegbe, 2005; Mbano, 2008; Osuoka, 2007; Oyegun, 1993). In 2012, over 853 million barrels of oil were exploited from over 5000 oil wells (NNP Annual Statistical Bulletin 2012, 72pp.). Further natural resources include, but are not limited to, timber and nontimber forest products, agriculture (raffia palm, banana, oranges, yam, pumpkin, to name just a few), aquaculture, and silica sand (Mbano, 2008; Okpara, 2004, 102pp.). In the Niger Delta, agricultural activities are characterized by traditional peasant farms. The largest portion of the GDP is generated by the oil industry (Abayomi, 1992; Mbano, 2008). Within the last decade alone, GDP grew from about 83 billion USD in 2000 to 178 billion USD in 2012. However, the profits generated from the oil industry could so far not help to alleviate the chronic poverty of this most densely populated country of Africa. Nigeria ranks 153rd of 186 countries listed in the human development index (CIA 2013). Environmental and socio-ecological threats in the Niger Delta Usually, river deltas globally belong to the most densely settled areas on Earth, as they are characterized by multiple benefits

offering a good livelihood for humans: a flat topography, fertile soils, access to marine and sweet water resources, access to harbors, numerous underground oil, gas, and salt deposits, as well as wetlands with extensive biodiversity (Kuenzer & Renaud, 2012). At the same time, numerous authors observe river delta resources exploitation in the Anthropocene with concern, discussing the concept of tipping points for socio-ecological systems (Renaud et al., 2013). Several socio-ecological threats are prevalent in the Niger Delta. The greatest threat is environmental pollution with hydrocarbons. Most multinational oil companies, such as Shell, Agip, Total, or Chevron, are active in the region, and the growth of the oil and petroleum industry during recent decades led to several serious environmental problems as well as social conflicts. Oil spills caused by pipeline leakages, technical accidents, or illegal discharge, as well as the release of further toxic substances, have negatively impacted surface water, aquifers, soils, and vegetation and led to the destruction of large parts of the mangrove forests. According to Baird (2010) between 9 and 13 million barrels of oil were spilled in the delta since 1958. Next to these resources, agricultural land, all fauna, as well as human health (food chain effects), are also affected (Bayode, Adewunmi, & Odunwole, 2010; Emoyan, Akpoborie, & Akporhonor, 2008; Twumasi & Merem, 2006; Ugoschukwu & Ertel 2008). Frequent flooding in the rainy season aggravates the oil-exploitation-related threats. Floodwater disperses large amounts of sediment in the delta as well as further inland. These

356

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

sediments are often enriched with heavy metals such as Zn, Cu, Fe, Hg, and Pb, contaminating the farmland (Achudume, 2007). The dredging of oil access canals to potential new oil well locations also leads to an increased release of sediment-bound toxins into the aquatic system (in addition to destroying the surface vegetative cover, which usually consists of mangroves or swamp forest). Next to leakage and spills, gas flaring accompanies oil production. Around 80% of all gas production is flared in Nigeria, and gas flaring is the greatest contributor to greenhouse gas and other toxic emissions in the Niger Delta (Emoyan et al., 2008; Kadafa, 2012). Flares release CO2 and methane, but also nitrogen and sulfur dioxides, as well as harmful volatile organic compounds. Benzene, toluene, xylene, and carcinogens like benzapyrene and dioxin are commonly measured. Gas flare exhaust coats the surrounding land with soot and damages adjacent vegetation. All this severe environmental pollution led to the delta's nomination for a place in the top 10 list of the “World's Worst Polluted Places Report” released by the Blacksmith Institute in 2013. While the oil companies are growing continuously, the resistance against them is prevalent as well (in the 1980s led by the later executed author Ken Saro Wiwa). Up to today, the consequences are riots, hostages, or attacks on pipelines, which, for example, led to Shell giving up operations in the western part of the delta in 2006 (Fig. 2). Another challenge for the delta region is climate-changeinduced sea level rise. Sea level rise leads to a strengthened progression of ocean water further inland, and thus accelerates salt water intrusion into soils and aquifers (Kuenzer & Renaud, 2012; Williams & Benson, 2010). According to the Intergovernmental Panel on Climate Change a sea level rise of 1 m in this region will put about 3 million people at risk (Amosu, Bashorun, Babalola, Olowu, & Togunde, 2012; IPCC, 1997; IPCC, 2007). Furthermore, it

will amplify ongoing coastal erosion processes in the delta (Amosu et al., 2012). Last but not least, no river delta is independent of activities occurring upstream in the respective river basin. Already in 1999 Abam (1999) investigated the impact of upstream hydropower dam construction on the fragile delta ecosystem. It was noticed that the closure of upstream dams leads to a sudden and prolonged drop in water levels (until reservoirs are filled), which aggravates the problems of diminished sediment delivery into the deltaic floodplains. Thus, upstream dams aggravate ongoing coastal erosion processes and threaten the ecological equilibrium of the delta. Especially sand bars and sand bank systems in front of the river outlets, which act as protection against storms, have shrunk after the closure of big dams, such as the Kainji dam. The questions we address in this present paper are: How has land cover for the whole Niger Delta changed over the past three decades? With respect to coastal dynamics: How has the coastline in the Niger delta changed between 1986 and 2013? In which areas do coastal erosion or coastal accretion prevail? Furthermore, we wanted to understand the manifestations of oil exploitation in remote sensing data. Questions we wanted to answer included: How have the oil access canals dredged by the oil industry e especially within the valuable mangrove regions e developed over the course of nearly three decades? And last but not least: How do gas flares manifest in the selected earth observation data, and what are the patterns of gas flare dynamics between 1986 and 2013? These questions will be addressed in the following sections. This study therefore aims at a comprehensive elucidation of the potential of the freely available Landsat archives for a remote sensing based analysis contributing to the understanding of Niger Delta dynamics.

Fig. 2. Impressions from the Niger Delta. All photographs were taken around the Escravos oil terminal near the town of Warri in the western part of the Delta. a and b: widely visible gas flares on the horizon. c and d: mangroves, coastal forest, and the invasive Nypa palm (Photographs: S. Van Beijma, C. Kuenzer, September 2007).

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Past earth observation-based studies of the Niger Delta Several authors used remote sensing data to assess the environmental impact of development in the region. Twumasi and Merem (2006) derived land cover change for a small eastern subset in the Niger Delta employing Landsat data from 1985 to 2000 which showed a decline in water bodies and mangrove forest, and revealed the consequences of several severe oil spills. In a similar study, land use/cover change was analyzed for a selected area of the delta between 1986 and 2008 using Landsat and Nigeriasat-1 imagery (Abbas, 2012). The same data were also used to identify different types of land degradation and to quantify the overall degraded area in the western part of the Niger Delta, which amounted to 10.6% in 1986 and 32.16% in 2008 (Abbas & Fasona, 2012). These findings are in line with a recent local study employing one Landsat TM scene for 1987 and the same frame covered in 2002 by Landsat ETMþ, where decreasing NDVI values between 1987 and 2002 indicate land degradation due to hydrocarbon exploration and human impact (Uchegbulam & Ayolabi, 2013). Next to these more general land cover change studies, which usually cover only small parts of the Delta (Salami, Akinyede, & de Gier, 2010), several remote sensing studies focused on forest assessments in the complete Niger Delta. Onojeghuo and Blackburn (2011) analyzed the spatial extent and rates of forest transition between 1986 and 2007, mapping “non-forest” and “forest” based on two time steps, considering the patterns, causes, and implications of landscape dynamics detected in Landsat images. Their results elucidated the spatial extent of deforestation, unchanged forest cover, and afforestation of 1.38, 2.39, and 1.15 million hectares, respectively, while annual deforestation and afforestation rates were 0.95 and 0.75% and forest fragmentation in the time span observed increased substantially. Landsat data were also utilized to assess the alteration of mangrove forest ecosystems between the mid-1980s and 2003 (James, Adegoke, Saba, Nwilo, & Akinyede, 2007). The results indicate that the spatial extent of mangrove loss summed up to 21,340 ha and was primarily caused by deforestation due to urbanization, dredging activities, oil exploration, and the spread of the invasive Nypa palm. One approach to derive canopy height and biomass for the mangrove forests in Africa (including the Niger delta) was recently presented by Fatoyinbo and Simard (2012). The authors used Landsat data for three time periods between 1973 and 2001 to determine the extent of mangroves, and in a further step SRTM DEM and LIDAR data from the ICESat/GLAS mission were used to estimate canopy height.

357

According to their findings, mangrove height in the Niger Delta can reach up to 45 m, and while in limited areas 10e20 m high mangroves prevail, the vast majority does not exceed 10 m of height. Coastal changes in Nigeria were monitored by the Nigerian Institute of Oceanography and Marine Research using a combination of aerial photographs and beach profiles between the 1960s and 1980s. Numerous erosion and accretion changes were mapped in this 20-year period, which revealed a predominance of erosion (Ibe, 1988). Adegoke, Fageja, Godstime, Agbaje, and Ologunorisa (2010) updated this coastal change assessment and analyzed optical and thermal bands of Landsat TM/ETM þ images for 1986 and 2003. This study showed that the total area of observed changes along the coastline summed up to 46,535 km2, with coastline erosion (27.65 km2, 59.43%) exceeding sediment deposition (18.88 km2, 40.57%). Another type of remote sensing analysis focused on the detection of gas flares in the Niger Delta. Anejionu, Blackburn, & Whyatt (2012) used a hot spot detection algorithm on Landsat data for gas flare detection. A second gas flare detection approach by the same authors was based on MODIS-acquired night-time thermal imagery of the Niger Delta region and successfully detected 150 active onshore and offshore flares from 2004 to 2006 (Anejionu, Blackburn, & Whyatt, 2013). Materials and methods A first challenge for remotely sensing the Niger Delta is frequent cloud coverage in the region and the fact that e in the case of Landsat efive image frames (see Fig. 3) are needed to cover the entire Niger Delta. The delta is located directly within the intersection area of four Landsat frames. Investigating the free Landsat archives for cloud-free data (0% cloud coverage) covering this complete area in the dry season (NovembereApril) left us with only a few possible dates for an analysis covering the late 1980s, the early 2000s, as well as the year 2013. The downloaded, preprocessed, mosaicked, and analyzed data is listed in Table 1. All satellite data was downloaded via the USGS Earth Explorer tool and georectified with sub-pixel (better than 30 m) accuracy towards the master geometry of the most recent scenes from 2013. Furthermore, all data were atmospherically corrected using the ATCOR-2 atmospheric correction code, which enables sensor calibration (the relationship between radiance received at the satellite's sensor and the gray value or digital number, DN) and a correction of the atmospheric influence of gasses, water vapor, and aerosols based on MODTRAN radiative transfer code (Richter, 1998,

Fig. 3. Landsat data coverage of the Niger Delta. Five frames need to be processed to cover the entire area of Africa's largest delta. 15 frames were analyzed for this study.

358

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Table 1 Landsat data analyzed in this study. Dates in format yyyy/mm/dd.

Row 56

Path 190

Path 189

Path 188

1986/01/15a 2003/11/22c 2013/01/01c

1987/12/21a 2002/12/30b 2013/01/10c 1987/12/21a 2002/12/30b 2013/01/10c

1986/12/19a 2003/01/08b 2013/01/03c 1986/12/19a 2003/01/08b 2013/01/03c

Row 57

a b c

Landsat 4/5 TM. Landsat 7 ETM SLC-on. Landsat 7 ETM SLC-off.

2003) to transfer DN values on the individual bands to reflectance and temperature. The data then underwent different types of analysis. We focused on an analysis of delta dynamics in three different fields. 1. Firstly, we generated land cover maps based on image mosaics from each of the three time steps, and subsequently investigated land cover change over the time span from 1986/87 to 2013. 2. Secondly, we analyzed coastal changes within the same time frame focusing on regions of outstanding coastline change and delta-wide coastline change rates. 3. Thirdly, the impact of the oil exploiting industry, manifested in the expansion of access canal networks within mangrove areas as well as gas flare activity, was analyzed. Satellite image processing is only described briefly, as the methods have been previously published in literature and as this manuscript does not have an in depth image processing but a geographic application focus. For land cover mapping the image data underwent pixel-based image classification using an extensive set of training samples (each class with between 5000 and 10,000 pixels distributed evenly over the image mosaic) collected in high resolution (Quickbird, Ikonos, Worldview) Google Earth imagery (Congalton & Green, 2007). Half of the samples were used as training samples for automatic generation of a decision tree classifier within the software package Twopac (Huth, Kuenzer, Wehrmann, Gebhardt, & Dech, 2012), which is based on the wellknown C5.0 code and enables decision trees to be built based on the concept of information entropy (Quinlan, 1993). The other half of the sample set was used for classification validation. The main classes that were distinguished were clouds, water, bare sand, artificial (urban), burned area, forest (non-mangrove), mangroves (differentiated into low, tall, and degraded mangroves) (Kuenzer, Bluemel, Gebhardt, Quoc, & Dech, 2011), and areas dominated by agriculture. For coastline analysis, the coastline positions of the three time steps were extracted by manual digitization. Coastal sand areas (e.g., beaches and pure sand bars) were not digitized as they are highly dynamic at short time scales. Including them in an analysis that spans several decades could lead to the detection of spurious, short-term coastal changes that are not relevant at longer timescales. The coastline digitization was therefore also insensitive to differences in tide level that usually strongly influence coastline comparisons (Adegoke et al., 2010). However, in general the Niger Delta is not characterized by huge tidal flats that are exposed at low tide, thereby not placing the boundary between land and water on a considerable different location than at high tide. The digitized coastlines were used as input in the Digital Shoreline Analysis System (DSAS) developed by the United States Geological Survey, USGS (Thieler, Himmelstoss, Zichichi, & Miller, 2005). DSAS enables the calculation of coastline change rates based on intersections of

the digitized coastline with transects perpendicular to the coast. Coastline erosion or accretion over the 27 year observation time span was then transferred into average annual coastline change rates. Oil well access canals were also digitized for all three time steps. The artificially dredged canals can be easily distinguished from natural canals due to their artificially straight or angular/perpendicular shapes. Segment length was then calculated, and the extent of the canal network for the three time steps was compared. Gas flare detection has among others been performed by Croft (1978), Muirhead and Cracknell (1984), and Elvidge et al. (2009). Gas flares usually do not have the extent of a 60 m Landsat thermal band pixel, but rather cover an area of a few square meters. Since the flares have differing temperatures the sub-pixel hotspots elevate the overall temperature of a pixel to different degrees. To detect gas flares a tool for the regional extraction of thermal anomalies presented in Kuenzer et al. (2007) was applied for all 15 Landsat images to the band located in the thermal infrared domain (band 6). The algorithm presented by Zhang and elucidated further by Kuenzer et al. (2007) uses raw (DN) or corrected (Kelvin) satellite data as input for sub-image statistical analyses. In the margins of a moving window of varying size, image subset histograms are analyzed with respect to the occurrence of thermally anomalous pixels. Curve features within the subset histograms are then defined as thresholds, enabling the separation of thermally anomalous versus background pixels. Due to the moving window approach each pixel of a thermal band is sampled over 1000 times. If a pixel is declared as thermally anomalous in >70% of the cases it is declared to be a thermal anomaly. The advantage of this approach compared to simple temperature thresholding is that regional thermal anomalies of varying temperature can be extracted. This approach has been proven suitable for thermal anomaly extraction in many case studies (e.g., Kuenzer, Hecker, Zhang, Wessling, & Wagner, 2008). Further details on thermal image data analyses can be found in Kuenzer and Dech (2013a) as well as in Kuenzer and Dech (2013b). Results of earth-observation-based analyses Results of land cover change analyses for the Niger Delta Fig. 4 depicts the results of land cover mapping for the years 1986/87 and 2013. Overall classification accuracy for the 2013 land cover classification as validated based on high resolution GoogleEarth-derived training samples is 81.6%. Table 2 depicts the classification accuracy of the individual classes. What is clearly evident is a strong expansion of urban areas in the delta. Especially the growth of the cities of Port Harcourt in the southeast delta (1), the growth of Warri in the western delta (2), as well as Aba and Owerri in the eastern center of the delta (3,4) is striking. Along with urban expansion and an expansion of the agricultural area goes a loss of swamp forest along the river branches. It can clearly be observed that swamp wetlands and swamp forest along the Niger's main stem and the river branches have thinned out. This phenomenon can also be observed south of Owerri and south of Onitsha e the northernmost urban cluster in the delta. From 1986/87 to 2013 urban areas expanded from 1516 km2 to 1730 km2, and the agricultural area has increased from 31,700 km2 to 33,895 km2. The forest and swamp forest area (green) has decreased from 18,325 km2 to 15,408 km2, and mangrove forest (overall, combining all mangrove classes) has remained more or less the same with 10,311 km2 in 1986/87 and 10,072 km2 in 2013 with a decrease period until 2003 (James et al. 2007) and a stabilization partially also due to protection and rehabilitation activities up to today. However, the once rain-

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Fig. 4. Land cover classification of the Niger Delta for 1986/87 (upper map) and for 2013 (lower map). Ten land cover classes were derived for the depicted time steps.

359

360

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Table 2 Land cover classification accuracy for the 2013 map. Overall accuracy: 81.6%.

Agriculture Artificial Sand bar Forest Low mangrove Tall mangrove Water Degraded mangrove Total

Agriculture

Artificial

Sand bar

Forest

Low mangrove

Tall mangrove

Water

Degraded mangrove

83.8 14.0 0.0 2.2 0.0 0.0 0.0 0.0 100.0

14.6 81.8 1.8 0.2 1.3 0.0 0.0 0.3 100.0

14.5 8.4 65.0 0.7 3.5 1.7 5.2 1.0 100.0

5.3 0.0 0.0 94.3 0.0 0.4 0.0 0.0 100.0

0.2 0.0 0.0 0.9 63.8 27.3 0.1 7.8 100.0

0.9 0.0 0.0 31.0 26.0 41.0 0.0 1.1 100.0

0.2 0.2 0.0 0.5 3.6 0.4 95.0 0.1 100.0

0.9 0.4 0.0 0.7 45.1 0.9 16.2 35.7 100.0

Bold represent overall individual class accuracies.

forested agricultural area nowadays only contains remnants of secondary growth forest, and also the mangrove ecosystem of the Niger Delta is facing stresses resulting from severe air and water pollution as well as the impact of ship traffic induced river bank erosion. It should be mentioned here that any change detection can of course only be as good as the individual classifications. For the 1986/87 map well as for the 2013 map classification accuracy exceeded 80%. This means that absolute numbers have to be interpreted with care, although we are confident that the general patterns we depicted represent the currently ongoing processes. Results of coastline change analyses for the Niger Delta Analysis of coastline dynamics in the Niger Delta revealed both erosion and accretion processes. Between 1986/87 and 2002/03, annual accretion rates were higher than erosion rates (Figs. 5 and 6). During this period, along the entire Niger Delta coastline, an average annual accretion of 2.6 km2 (>40 km2 over the 16 year time span) was observed, while an average of 0.8 km2 of coastal land surface was eroded each year (about 13 km2 for the 16 year time span). After 2002/03 in contrast, annual accretion rates decreased and erosion rates increased in all coastal states except for Akwa Ibom. This led to a reverse effect with erosion prevailing over accretion in the later period between 2002/03 and 2013. Particularly high erosion has been observed in Bayelsa, where the net loss within these 11 years accounts for 3.8 km2. In general, hot spots of coastal dynamics are found in the vicinity of the larger river mouths where the sediment load from the delta and stronger currents can lead to erosion and accumulation processes. Strong coastal dynamics are also found in areas with currents parallel to the coast,

where sandspits several kilometers long accumulate and erode (e.g., Fig. 6, detail maps A and B). The results of our analyses contradict the coastal change detection results of Adegoke et al. (2010), who found erosion prevailing over accretion in their observation period between 1986 and 2003. These differences might be due to the different approaches used, i.e., image differencing (Adegoke et al., 2010) versus our detailed manual coastline digitization. Furthermore, Adegoke et al. (2010) did not elaborate on the data acquisition times of their data, which are greatly influenced by tidal differences from image to image. Pure image differencing will lead to results, but they will be influenced by comparing imagery at different tidal levels. We circumvented this problem by digitizing the coastline based on the vegetation/no vegetation line. At the same time our results are in line with the recent findings of Fashae and Onafeso (2011), which depicted severe coastline losses in the area around Lagos between 1999 and 2009. Additionally, they are in harmony with the findings of Abam (1999), who depicted a severe decline in sediment load after dams were erected upstream of the Niger Delta. Given the fact that during recent decades numerous new dams have been closed, it is to be expected that the sediment load is further decreasing (Abam, 2001). Fig. 7 depicts coastal change rates as derived based on the USGS DSAS. Reddish, orange and yellow tones indicate areas where coastal erosion prevails, while green tones indicate areas of overall coastal accretion. As depicted in the figure, annual coastal change rates at discrete locations vary between 64.8 m of coastal erosion per year (amounting to 1.75 km over the 27 year time span) to accretion rates of 59.8 m per year (or 1.6 km of accretion over the 27 year time span). In the past decade in most of the Niger Delta coastal erosion exceeded new accretion. Accretion occurs rather in

Fig. 5. Annual coastal area accreted and eroded during the periods 1986/87 to 2002/03 and 2002/03 to 2013 for the four coastal states and for the complete Niger Delta. In all states, an increase in erosion and a decrease in accretion rates were observed. While for the period 1986/87 to 2002/03 accretion prevailed, net erosion is observed from 2002/03 to 2013.

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

361

Fig. 6. Accretion and erosion of coastal areas in the Niger Delta during the periods 1986/87 to 2002/03 and 2002/03 to 2013 (light and dark green and red tones). The detailed zoom maps A, B and C show areas of high dynamics; over the observed time span of over 27 years in some regions coastlines shifted by several kilometers. Subset C depicts a region where a protective sandpit over 2 km long was completely lost. In other areas (subset A) sandspits are growing due to currents and sediment accumulation parallel to the coast.

the western delta. Erosion is strongest in the center, where the split up main stem enters the ocean. In the eastern delta regions accretion and erosion alternate. These findings reflect the fact that the Niger Delta is a delta equally influenced by river, wave, and tide

action. The Niger River transports vast amounts of sediment into the ocean, which are then also distributed and partially accreted via coast-parallel currents along the Niger Delta shores. As mentioned, most coastline changes along the delta occur mainly via the erosion

362

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Fig. 7. Niger Delta coastline change rates (meters/year) for the past three decades. The detailed zoom maps A, B and C depict representative locations where coastal erosion dominates over the 27 year time span.

or expansion of spits and sandbars and thus the opening or closure of backwaters changing with coast-parallel currents (see also Fig. 6). Results: oil well access canal development for the Niger Delta Oil wells in the Niger Delta are often located in inaccessible mangrove swamps. To access these locations, access canals are

dredged from the nearest creek. We analyzed the development of these access canal networks based on digitized maps derived from Landsat imagery for the three different time steps. The most recent time step (2013) includes a smoke covered area in the Port Harcourt region (Rivers State), covering a number of canals that had been detected in 2002/03. It is assumed that the access canals located under this smoke plume did not disappear during the period from 2002/03 to 2013, so they are included in the analyses of the total

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

length of access canals. Access canals occur in the states of Bayelsa, Delta, and Rivers. Fig. 8 as well as Fig. 9 visualize the expansion of the access canal network between 1986 and 2013. In 1986/87, 91.5 km of access canals had already been dredged in Bayelsa. This number increased to 103.4 km in 2002/03, and expanded further to 150.2 km in 2013. Therefore, the length of access canals grew by over 30%. In Delta State access canals expanded from 112.7 km to 132.7 kme150.3 km respectively. In Rivers State only 26 km of dredged canals existed in 1986/87, while 40.3 km were mapped for 2013. In total the access canal network for all three states expanded from a total of 230.4 km in 1986/87 to 269.5 km in 2002/03 to 349.3 km in 2013. Results: gas flare dynamics for the Niger Delta Figs. 10 and 11 present the results of gas flare detection. The thermal anomalies extracted from the 15 Landsat scenes were all visually checked for reasonability (location, context) in high resolution Google Earth and aerial imagery showing flares. Most flares could be associated with gas flares related to the oil industry, although some of the thermal anomalies detected could not directly be associated. These false alarm detections were omitted in Fig. 11. Fortunately, only few other sources for thermal anomalies exist in the delta. Forest fires, which can occur especially between October and March, are too large in extent to mimic a gas flare, which creates a thermal anomaly of smaller extent (Kuenzer et al., 2007). The overall detection accuracy of gas flares was estimated to be 88% for the 2002/2003 time step and 86% for the 2013 time step. As in April 2003 problems occurred with the ETM þ scanner, the Landsat-7 imagery for 2013 was acquired with the Scan Line Corrector (SLC) turned off, creating data gaps. The gas flare locations that were detected in these gap areas in the data sets prior 1986/87 or 2002/03 were added to the 2013 data as potentially still

363

active gas flare locations, as gas flare rigs are usually long-term installations. However, the flares are presented separately in Figs. 10 and 11 to ensure transparency of data processing. At the same time new gas flare rigs that might have been installed in regions located within the ETM þ data gaps between 2003 and 2013 could not be identified in this way. Flare numbers for 2013 therefore have to be interpreted with care. They are probably much higher than in the column presented for the year 2013, as we can assume that new gas flare rig installations have occurred also within data gap areas. At the same time, several government and NGO-driven programs in the delta aim at reducing flaring by 2017, as flares not only release CO2, CO, and CH4, but also toxins such as dioxins, benzenes, and toluenes, and are directly related to the degradation of natural resource and biodiversity in their greater vicinity. Overall 167 flares were detected in the three time steps investigated. A clear increase in flares from 1986/87 to the time step of 2002/2003 can be observed. While flare numbers in Bayelsa and Delta States are more or less constant, a tremendous increase in Rivers State can be observed. Here, flare numbers double. The increase in the Rivers State in the mid-2000s can most likely be attributed to the fact that especially the western Niger Delta became more and more instable and unsafe due to militant groups opposing oil exploitation. Shell e for example e moved out of the western area in 2006. In Fig. 11 it can be observed that most permanent gas flares (active during all three time steps, in dark orange) occur east of the Niger main stem, and here especially in the coastal area south of Port Harcourt. Furthermore, many new flare locations developed on the east side of the Niger River, parallel to it, between the towns of Ahoada and Ihiala. At the same time some recent new flare activities (2013 only) occurred in the very western part of the delta. A large proportion of flares is located within the ecologically valuable mangrove forests of the Niger Delta. Our

Fig. 8. Niger Delta oil well access canal development statistics for the past three decades. Access canal development (waterways) occurs mainly in the states “Delta”, “Bayelsa”, and “Rivers”. Below: examples of the appearance of dredged canals in satellite imagery. Artificial versus natural canals can clearly be distinguished.

364

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Fig. 9. Niger Delta oil well access canal development within the mangrove zones over the past three decades. The detailed zoom maps A, B and C depict some exemplary areas exhibiting strong activity and dynamics. In total, the access canal network for all three states expanded from a total of 230.4 km in 1986/87 to 269.5 km in 2002/03 to 349.3 km in 2013.

results are in good accordance with the results of Anejionu et al. (2012, 2013), who detected 64 flares in 2002/2003, whereas we detected 67 flares for this time step. It should be mentioned that for gas flare detection the proportion of flares is more relevant than the overall number. Including more time steps in the analyses would most likely lead to the detection of further flares. The new thermal band of Landsat 8DCM

now in orbit is a promising source of information for future flare analyses. Conclusion Covering the complete Niger delta, we presented the results of land cover change analyses, results of the assessment of coastline

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

365

Fig. 10. Niger Delta gas flare development statistics over the past three decades (top). Gas flares that were detected in 2003 but are obscured by data gaps in 2013 are hatched. Examples of the appearance of gas flares as found in high resolution satellite imagery (bottom).

dynamics, as well as the manifestation of oil exploitation activity expressed by access canal dredging and gas flaring, monitored within the 27 year time span investigated (1986e2013). Land cover change analyses reveal a strong expansion of urban areas. From 1986/87 to 2013 urban areas expanded from 1516 km2 to 1730 km2. Likewise, the agricultural area increased from 31,700 km2 to 33,895 km2 e both to the cost of regions covered with forest and swamp forest, which decreased from 18,325 km2 to 15,408 km2. The mangrove area of the delta decreased slightly (in some areas losses, in others expansion), but this does not reveal information on mangrove vigor, health or biodiversity status. Analyses of coastline dynamics revealed that between 1986/87 and 2002/03, annual accretion rates were slightly higher than erosion rates. After 2002/03 erosion rates increased in all coastal states except for Akwa Ibom. Annual coastal change rates at discrete locations vary between 64.8 m of coastal erosion per year (amounting to 1.75 km over the 27 year time span) to accretion rates of 59.8 m per year (or 1.6 km of accretion over the 27 year time span). The impacts of oil industry activities in the Niger Delta are clearly observable. Here, especially the effects of access canal dredging and gas flaring could be extracted. Access canals occur in the states of Bayelsa, Delta, and Rivers. In Bayelsa 91 km could be detected in 1986/87, and the network expanded to 150 km in 2013. Similar trends could be detected in Delta State (from 112 km to 150 km), and in Rivers (from 26 km to 40 km). For the time span analyzed, the access canal network within the valuable mangrove ecosystems of all three affected delta states expanded from a total of 230.4 km in 1986/87 to 269.5 km in 2002/03 to 349.3 km in 2013.

Based on the detection of hot spots in Landsat thermal band data overall 167 flare were detected and a clear increase in flares from 1986/87 to 2002/2003 could be observed. While flare numbers in Bayelsa and Delta remain more or less constant, a tremendous increase in the delta state of Rivers can be observed. Here, flare numbers more than doubled. A large proportion of gas flare sites is located within the valuable mangrove ecosystems of the Niger Delta. Overall, the Niger Delta is undergoing rapid change, caused on the one hand by the local drivers of population growth, increasing urbanization, socio-economic transformation and related pressures on natural resources, and on the other hand by the activities of the international oil industry, which seriously impact the ecological system. However, especially challenging is the fact that e next to these local drivers of change e the delta's dynamics are also affected by sea level rise from the coastal side, as well as by distant e and often trans-boundary e upstream developments. With predicted sea level rise and without mitigation efforts coastal forefront areas of the delta will be lost to the sea over time, and salinization of soils and aquifers will protrude further inland. The Niger Basin is shared by of nine countries, and interests of the individual countries often conflict. With numerous hydropower plans in operation and many more planned it can be expected that the sediment load reaching the delta will decrease further over time. Additional flood pulse changes will disturb natural rhythms, and this will aggravate delta erosion on the one hand, and agricultural practices and livelihoods on the other. Despite the valuable oil reserves present in the delta: the other

366

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Fig. 11. Niger Delta gas flare distribution over the past three decades. The detailed zoom maps A, B and C depict selected areas in the western, central, and eastern delta that are especially in the focus of large oil companies.

value of the delta e its ecosystems, its immense biodiversity and its provision of a base for millions of livelihoods e needs to be better understood. To preserve the Niger Delta and its livelihoods as a valuable and complex ecosystem, a large variety of technological, ecological, educational, and political measures has to be orchestrated. These include safety and precautionary measures of the oil industry to minimize its negative impacts, environmentally sound utilization of natural resources, protection of

coastal and inland forest, wetland restoration, education in the fields of ecology, climate, and adaptation to change, basin-wide information sharing, as well as the initiation of regulations and norms accompanied by strict law enforcement. This is a process of years and decades. The supply of remote-sensing derived information products e also to local stakeholders e can be one small contribution in support of informed and intelligent decision making.

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Acknowledgments The authors thank K. Genetzke, and R. Sonnenschein for support with literature collection and review. Further thanks go to the anonymous reviewers for their constructive comments on the manuscript.

References Abam, T. K. S. (1997). Genesis of canal bank overhangs in the Niger Delta and analysis of mechanisms of failure. Geomorphology, 18, 151e164. Abam, T. K. S. (1999). Impact of dams on the hydrology of the Niger Delta. Bulletin of Engineering Geology and the Environment, 57, 239e251. Abam, T. K. S. (2001). Regional hydrological research perspectives in the Niger Delta. Hydrological Sciences Journal, 46(1), 13e15. Abayomi, Y. O. (1992). The agricultural sector in Nigeria. The way forward. Central Bank of Nigeria Bulletin, 21(3), 11e25. Abbas, I. I. (2012). An assessment of land use/land cover changes in a section of Niger Delta, Nigeria. Frontiers in Science, 2(6), 137e143. Abbas, I. I., & Fasona, M. J. (2012). Remote sensing and geographic information techniques: veritable tools for land degradation assessment. American Journal of Geographic Information System, 1(1), 1e6. Adegoke, J. O., Fageja, M., Godstime, J., Agbaje, G., & Ologunorisa, T. E. (2010). An assessment of recent changes in the Niger Delta coastline using satellite imagery. Journal of Sustainable Development, 3(4), 277e296. Ajibade, L. T., & Awomuti, A. A. (2009). Petrolium exploitation or human exploitation? an overview of Niger Delta oil producing communities in Nigeria. African Research Review, 3(1), 111e124. Amosu, A. O., Bashorun, O. W., Babalola, O. O., Olowu, R. A., & Togunde, K. A. (2012). Impact of climate change and anthropogenic activities on renewable coastal resources and biodiversity in Nigeria. Journal of Ecology and the Natural Environment, 4(8), 201e211. Anejionu, O., Blackburn, A., & Whyatt, D. (2012). Improved remote survey of Gas flaring in the Niger Delta region of Nigeria with landsat imagery. In D. Whyatt, & B. Rowlingson (Eds.), Proceedings of the GIS Research UK 20th Annual Conference. Lancaster: Lancaster University. Anejionu, O., Blackburn, A., & Whyatt, D. (2013). Remote mapping of gas flares in the Niger Delta with MODIS imagery. Earsel Special Research Forum dedicated to Remote Sensing for Developing Countries. Matera, Italy, 5 June 2013. Achudume, A. C. (2007). Assessment of farmland sediments after flooding in Ubeji land in Niger Delta of Nigeria. Environmental Monitoring and Assessment, 135, 335e338. Baird, J. (July 26, 2010). Oil's shame in Africa. Newsweek 27. Bayode, O. J. A., Adewunmi, E. A., & Odunwole, S. (2010). Environmental implications of oil exploration and exploitation in the coastal region of Ondo State, Nigeria: a regional planning appraisal. Journal of Geography and Regional Planning, 4(3), 110e121. Congalton, R. G., & Green, K. (2007). Assessing the accuracy of remotely sensed data. Principles and practices (2nd ed.) (p. 183). Boca Raton, FL, USA: CRC Press. Croft, T. A. (1978). Nighttime images of the earth from space. Scientific American, 239, 68e79. Elvidge, C. D., Ziskin, D., Baugh, K. E., Tuttle, B. T., Ghosh, T., Pack, D. W., et al. (2009). A fifteen year record of global natural gas flaring derived from satellite data. Energies, 2, 595e622. Emoyan, O. O., Akpoborie, I. A., & Akporhonor, F. E. (2008). The oil and gas industry and the Niger Delta: implications for the environment. Journal of Applied Sciences and Environment Management, 12(3), 29e37. Fashae, O. A., & Onafeso, O. D. (2011). Impact of climate change on sea level rise in Lagos, Nigeria. International Journal of Remote Sensing, 32(24), 9811e9819. Fatoyinbo, T. E., & Simard, M. (2012). Height and biomass of mangroves in Africa from ICESat/GLAS and SRTM. International Journal of Remote Sensing, 34(2), 4859e4871. Goudie, A. S. (2005). The drainage of Africa since the Cretaceous. Geomorphology, 67, 437e456. Huth, J., Kuenzer, C., Wehrmann, T., Gebhardt, S., & Dech, S. (2012). Land cover and land use classification with TWOPAC: towards automated processing for pixeland object-based image classification. Remote Sensing, 4, 2530e2553. Ibe, A. C. (1988). Coastline erosion in Nigeria. The Nigerian Institute for Oceanography and Marine Research and A.C. Ibe. Ibadan. University Press, Ibadan, Nigeria. ISBN: I978-2345-041. Ikelegbe, A. (2005). The economy conflict in the oil rich Niger Delta Region of Nigeria. Nordic Journal of African Studies, 14(2), 208e234. IPCC. (1997). Special report on the regional impacts of climate change: An assessment of vulnerability. UK: Cambridge University Press. IPCC. (2007). Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment. Report of the Intergovernmental Panel on Climate Change. 27th session of the IPCC in Valencia, Spain. Isebor, C. E., & Awosika, L. F. (1993). Nigerian mangrove resources: status and management. In E. D. Diop (Ed.), Conservation and Sustainable utilization of mangrove forests in Latin America and Africa regions (pp. 169e185). Part II e

367

Africa. International Society for Mangrove Ecosystems. Mangrove Ecosystems Technical Reports No. 3, PD 114/90. James, G. K., Adegoke, J. O., Saba, E., Nwilo, P., & Akinyede, J. (2007). Satellite-based assessment of the extent and changes in the mangrove ecosystem of the Niger Delta. Marine Geodesy, 30(3), 249e267. Kadafa, A. A. (2012). Environmental impacts of oil exploration and exploitation in the Niger Delta of Nigeria. Global Journal of Science Frontier Research Environment & Earth Sciences, 12(3). Version 1.0. Kuenzer, C., & Quoc, T. V. (2013). Assessing the ecosystem services value of Can Gio Mangrove Biosphere Reserve: combining earth-observation- and householdsurvey-based analyses. Applied Geography, 45, 167e184. Kuenzer, C., & Dech, S. (2013b). Thermal infrared remote sensing e Sensors, methods, applications. In Remote sensing and digital image processing series (Vol. 17) (p. 572), ISBN 978-94-007-6638-9. Kuenzer, C., & Dech, S. (2013a). Theoretical background of Thermal infrared remote sensing. In C. Kuenzer, & S. Dech (Eds.), Remote sensing and digital image processing series: Vol. 17. Thermal infrared remote sensing e Sensors, methods, applications (pp. 1e26), ISBN 978-94-007-6638-9, 572. Kuenzer, C., & Renaud, F. (2012). Climate change and environmental change in river deltas globally. In F. Renaud, & C. Kuenzer (Eds.), The Mekong Delta system e interdisciplinary analyses of a river delta (pp. 7e48), ISBN 978-94-007-3961-1. Springer Environmental Science and Engineering. Kuenzer, C., Bluemel, A., Gebhardt, S., Quoc, T. V., & Dech, S. (2011). Remote sensing of mangrove ecosystems: a review. Remote Sensing, 3, 878e928. Kuenzer, C., Hecker, C., Zhang, J., Wessling, S., & Wagner, W. (2008). The potential of multi-diurnal MODIS thermal bands data for coal fire detection. International Journal of Remote Sensing, 29, 923e944. Kuenzer, C., Zhang, J., Li, J., Voigt, S., Mehl, H., & Wagner, W. (2007). Detection of unknown coal fires: synergy of coal fire risk area delineation and improved thermal anomaly extraction. International Journal of Remote Sensing, 28, 4561e4585. Mbano, E. P. (2008). Alternative natural resources development to petroleum in the rural areas of the Niger Delta region of Nigeria. Journal of Environment, 2(1), 4e14. Muirhead, K., & Cracknell, A. P. (1984). Identification of gas flares in the North Sea using satellite data. International Journal of Remote Sensing, 5(1), 199e212. NNPC, Nigerian National Petroleum Corporation. (2012). 2012 Annual statistical bulletin e NNPC. Record Number: 476, Abuja, Nigeria. Ohimain, B. L. (2003). Environmental impacts of oil mining activities in the Niger Delta Mangrove Ecosystem. In Proceedings of the 8th International Congress on Mine Water & the Environment, Johannesburg, South Africa (pp. 503e517). Okpara, E. E. (2004). Post-rio realities of sustainable development in the Niger Delta region of Nigeria. N-COGEP-D. Owerri: Ihem Davis Press. Onojeghuo, A. O., & Blackburn, G. A. (2011). Forest transition in an ecologically important region: patterns and causes for landscape dynamics in the Niger Delta. Ecological Indicators, 11, 1437e1440. Osuoka, A. I. (2007). Oil and gas revenues and development challenges for the Niger Delta and Nigeria. Expert group meeting on: The use of non-renewable resource revenues for sustainable local development. Organized by the UN department of economic and social affairs, 21 September 2007. Oyegun, C. U. (1993). Land degradation and the coastal environment of Nigeria. Catena, 20, 215e225. Prince, C. M., & Arokoyu, S. B. (2010). Mangrove forest depletion, biodiversity loss and traditional resources management practices in the Niger Delta, Nigeria. Research Journal of Applied Sciences, Engineering and Technology, 2(1), 28e34. Quinlan, J. R. (1993). C4.5: Programs for machine learning. San Mateo, CA, USA: Morgan Kaufmann. Quoc, V. T., Oppelt, N., & Kuenzer, C. (2012). Remote sensing in mapping mangrove ecosystems e an object-based approach. Remote Sensing, 5(1), 183e201. Reijers, T. A. J. (2011). Stratigraphy and sedimentology of the Niger Delta. Geologos, 17(3), 133e162. Renaud, F., Syvitski, J. P. M., Sebesvari, S., Werners, S., Kremer, H., Kuenzer, C., et al. (2013). Tipping from the Holocene to the Anthropocene: how threatened are major world deltas? Current Opinion in Environmental Sustainability, 5(6), 644e654. Richter, R. (1998). Correction of satellite imagery over mountainous terrain. Applied Optics, 37, 4004e4015. Richter, R. (2003). Atmospheric and topographic correction for satellite imagery. ATCOR-2/3 user guide, version 5.5. DLR-I3 564e02/03, Wessling, Germany, 57 pp. Salami, A. T., Akinyede, J., & de Gier, A. (2010). A preliminary assessment of NigeriaSat-1 for sustainable mangrove forest monitoring. International Journal of Applied Earth Observation and Geoinformation, 12, 18eS22. Short, K. C., & Staeuble, A. J. (2004). Outline of geology of Niger delta. AAPG Bulletin, 51(5), 761e799. Thieler, E. R., Himmelstoss, E. A., Zichichi, J. L., & Miller, T. L. (2005). Digital Shoreline Analysis System (DSAS) version 3.0: An ArcGIS extension for calculating shoreline change. U.S. Geological Survey Open-File Report, 2005-1304. Twumasi, Y. A., & Merem, E. C. (2006). GIS and remote sensing applications in the assessment of change within a coastal environment in the Niger Delta region of Nigeria. International Journal of Environmental Research and Public Health, 3(1), 98e106. Uchegbulam, O., & Ayolabi, E. A. (2013). Satellite image analysis using remote sensing data in parts of Western Niger Delta, Nigeria. Journal of Emerging Trends in Engineering and Applied Sciences (JETEAS), 4(4), 612e617.

368

C. Kuenzer et al. / Applied Geography 53 (2014) 354e368

Ugbe, F. C. (2011). Basic engineering geological properties of lateritic soils from Western Niger Delta. Research Journal of Environmental and Earth Sciences, 3(5), 571e577. Ugoschukwu, C., & Ertel, J. (2008). Negative impacts of oil exploration on biodiversity management in the Niger Delta of Nigeria. Impact Assessment and Project Appraisal, 26(2), 97e125. Williams, A. B., & Benson, N. U. (2010). Interseasonal hydrological characteristics and variabilities in surface water of tropical estuarine ecosystems within Niger Delta, Nigeria. Environmental Monitoring and Assessment, 165, 399e406.

Internet resources CIA (Central Intelligence Agency) (2013). https://www.cia.gov/library/publications/ the-world-factbook/geos/ni.html. Last access 14.10.13. Dudley, P.M. (2006). Cross-Niger transition forests. http://www.eoearth.org/view/ article/51cbed597896bb431f691a36/. Last access 07.10.13. McGinley, M. (2008). Niger Delta swamp forests. http://www.eoearth.org/view/ article/154853/. Last access 7.10.2013.