Remote sensing appraisal of Lake Chad shrinkage connotes severe impacts on green economics and socio-economics of the catchment area

Lake Chad commonly serves as a major hub of fertile economic activities for the border communities and contributes immensely to the national growth of all the countries that form its boundaries. However, incessant and multi-decadal drying via climate change pose greater threats to this transnational water resource, and adverse effects on ecological sustainability and socio-economic status of the catchment area. Therefore, this study assessed the extent of shrinkage of Lake Chad using remote sensing. Landsat imageries of the lake and its surroundings between 1987 and 2005 were retrieved from Global Land Cover Facility website and analysed using Integrated Land and Water Information System version 3.3 (ILWIS 3.3). Supervised classification of area around the lake was performed into various land use/land cover classes, and the shrunk part of its environs was assessed based on the land cover changes. The shrinkage trend within the study period was also analysed. The lake water size reduced from 1339.018 to 130.686 km2 (4.08–3.39%) in 1987–2005. The supervised classification of the Landsat imageries revealed an increase in portion of the lake covered by bare ground and sandy soil within the reference years (13 490.8–17 503.10 km2) with 4.98% total range of increase. The lake portion intersected with vegetated ground and soil also reduced within the period (11 046.44–10 078.82 km2) with 5.40% (967.62 km2) total decrease. The shrunk part of the lake covered singly with vegetation increased by 2.74% from 1987 to 2005. The shrunk part of the lake reduced to sand and turbid water showed 5.62% total decrease from 1987 to 2005 and a total decrease of 1805.942 km2 in area. The study disclosed an appalling rate of shrinkage and damaging influences on the hydrologic potential, eco-sustainability and socio-economics of the drainage area as revealed using ILWIS 3.3.


Introduction
In recent years, there are alarming reports of phenomenal contraction and total disappearance of many freshwaters and wetlands in the world. Retractions and extinctions of waterbodies are primarily connected to climate change, hydro-climatological pressures and anthropological workouts [1][2][3][4][5][6]. Of concern are lakes and their catchment ecosystems, due to the various importance they serve. Shrinkage of lakes is not principally of local concern, but also a matter of regional and global attentions [7]. The rapid damaging impacts on sustainability of environmental green economics and deterioration of socio-economic status in the retracting lake basins call for priority conservational efforts. For instance, diminution of many African lakes [8,9] and accompanying paralysed livelihoods, food shortage, disease outbreaks, high penury rates, violence outbreaks, conflicts, terror attacks and social insecurity etc. in lakeshores/catchments have been reported [10][11][12][13][14][15]. Hence, various research enterprises have been channelled at monitoring and assessing the extent of changes in lakes and their eco-environmental status to alert policymakers towards precautionary activities and plans for future water resources. In addition, informational needs for future land use, wetland managements, political decisions and activities, and water management are essentials for which many investigations were conducted [5,15,16].
Previous studies on Lake Chad and vicinity employed instrumentation of direct site survey including interviews, questionnaires and photographs [15]. Besides that these methods are backbreaking, time depleting and cost-ineffective, they involve low coverage because information about inaccessible and remote areas of the lake is usually excluded. Investigations using remote sensed dataset or satellite imageries have also been reported on the lake [24][25][26]. Remote sensing provides an excellent environment for cost-effective mapping, cheap imageries, geographical information system, that allow efficient monitoring, areal change detection and modelling of inaccessible natural resources, environmental variables and phenomena [26][27][28]. Remote sensed imageries and approaches not only serve as powerful tools for information collection, processing and management in inaccessible and hostile environment, hydro-meteorological events such as hurricanes, floods, land cover/land use and environmental scars, but also they are effective communication tools [26,29,30].
Nevertheless, the effective, efficient and accurate application of remote sensing in assessment of areal extent of lakes and other freshwater resources is largely not independent of the dataset, data source and instrumentation. For illustration, Adewuyi [23] employed Modis, Landsat and Argon imageries between 1963 and 2001 (dataset for 1963, 1973, 1987, 1997 and 2001); and GIS tools including Arc view 3.1 and Arc map 8.1 in his investigation of Lake Chad. Similarly, Alfa et al. [25]   suitability of the different approaches in assessment of Lake Chad extent contraction. Therefore, the present study appraised Lake Chad shrinkage using remote sensing and instrumentation of Integrated Land and Water Information System v. 3.3 (ILWIS 3.3). This was the first investigation that analysed remote sensed dataset (Landsat imageries) of the lake and its surroundings using ILWIS 3.3. The study monitored shrinkage and reduction in Lake Chad using satellite images. Firstly, supervised classification of the satellite images of the Lake Chad basin was done into various land use/land cover classes for the reference years (1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005), secondly land cover extent of the various land cover classes was determined, and finally the shrinkage pattern within the study period was computed.

The study area
The 'standard' Lake Chad is situated between latitudes 12°10 N and 14°30 N, and longitudes 13°E and 15°30 E in the hot semi-arid region. The lake is located along the international boundary of four countries which are Nigeria, Niger, Chad and Cameroon, and it is the fourth largest lake in Africa. Its hydrological basin (2.4 million km 2 extent) constitutes freshwater source shared by eight African countries which are Niger, Cameroon, Nigeria, Chad Republic, Central African Republic, Sudan, Libya and Algeria. It is fed by Chari-Logone river systems from the south and Komadugu/Yobe-Ngadda river systems from the western part of the lake. The catchment area occupies approximately 2 434 000 km 2 (approx. 8%) of African total continental land surface [17]. It is situated on an altitudal plateau of estimated 283 m above average sea level [31] (figure 1).  that each specific wavelength range was stored as a separate image (band). The band combination of satellite images was carried out in order of red, green and blue.

Image classification, interpretation and embellishment
The pictorial image or general overview (features) of the location obtained from band combination was further subjected to supervised classification using ILWIS 3.3. This was done to categorize physical (Earth) characteristics feature of the study location. Unlike unsupervised classification whereby the software picks the characteristics features (points) of the images and places them in categories without specific name, the supervised classification involves a human operator deciding the points or area of the combined image to be categorized with specific name tag. The land cover classes considered in the supervised classification include (i) current water in the lake, (ii) sandy soil and bare ground, (iii) the shrunk part of the lake covered with vegetation, (iv) the shrunk part of the lake covered with soil and vegetation, and (v) the shrunk part of the lake reduced to sand and turbid water. The knowledge of correct band combination helps in the interpretation of the already classified image. The classified image was then embellished by adding appropriate layout (north arrow, title bar, legend, scale, text and other features). Also, comparison plots of the years were performed. In all cases, the number of pixels corresponding to a specific land cover class was determined with raster calculator. Then, corresponding coverage area was estimated through cross-multiplication of its number of pixels in the attribute table and the image resolution.

Land use features of Lake Chad from 1987 supervised classified image
The supervised classification of Lake Chad and its environs for year 1987 is presented in figure 2, while the corresponding area coverage for the various land use/land cover identified is presented in table 2. The area coverage of the current water in the lake as at 1987 was 1339.018 km 2 . The portion covered by sandy soil and bare ground was 13 490.8 km 2 . While the shrunk part covered with soil and vegetation was 11 046.44 km 2 , the shrunk area vegetated only was 4610.831 km 2 . A 2306.929 km 2 area of the shrunk part was reduced to sand and turbid water at 1987.

Land use features of Lake Chad from 2001 supervised classified image
The 2001 supervised classification image of Lake Chad basin is presented in figure 4 and the associated coverage area in table 4. The classification image showed only a part of the lake and not the entire catchment area due to unavailability of the complete imagery and high percentage of cloud cover. The 2001 supervised classified image of the lake was excluded from further analysis in this study.

Discussion
The application of remote sensing via ILWIS 3.3 instrumentation to hydro-climatic phenomena associated with Lake Chad catchment area was carried out in this study. Owing to the merits that remote sensing allows vivid assessment of the extent of changes and estimate of accompanying earth physical characteristics, hydro-ecologic to socio-economic manifestations, the various changes in the lake extent and land cover classes of its environs were assessed. Remote sensing provides synoptically repeated observations, frequent wetland maps, surrounding land uses attributes and time-serial changes, which overcome limitations posed by spatial and temporal coverage in conventional measurements (field surveys/gauge stations) in wetland monitoring [32][33][34].

Current water in the lake
This is the portion of the supervised classification of the Landsat imageries of Lake Chad for the individual year that depicts the real extent of water present in the lake (i.e. 1987, 1999 and 2005). In other words, it is the Earth surface that the lake presently covers due to its shrinkage in each reference year. The reduction in the lake water coverage area and extent from 1339.018 km 2 in 1987 to 876.297 km 2 in 1999, and a total size decrease of 130.686 km 2 from 1987 to 2005 (table 6), could be attributed to both climatic and anthropogenic influences. Anthropogenic demands and climatic variability have been associated with the induced and accelerated eco-environmental changes in the lake catchment area [21]. Climatic phenomena such as high evaporation and low precipitation are reported contributors to severe lake shrinkage [35][36][37]. The mean annual evaporation rate (1600 mm) of the lake basin previously reported was twofold higher in magnitude compared to its average annual rainfall rate (approx. 625 mm) [38]. Furthermore, annual maximum temperatures of the lake ranged between 35 and 40°C, especially in the northern catchment area [39] with annual average temperature of 21.4°C [8].
The lake extent or current water size obtained in this study showed discordant values to estimates in other studies [22,23,25]. However, the results fell within the range and followed similar patterns to those reported. This could probably be due to differences in dataset used, study periods and instrumentation. For instance, Adewuyi [23] employed Modis, Landsat and Argon imageries between 1963 and 2001 (dataset for 1963, 1973, 1987, 1997 [40] and GIS techniques were adopted by Ebenki [24]. Ebenki [24] in addition used the K-means unsupervised classification in contrast with the supervised classification employed in this study. Thus, the supervised classification used in this study overcame possible classification of homogeneous spectra classes within the dataset which necessarily do not equal the same information family, unlike misclassification that commonly accompanies unsupervised classification [24,41]. Certain estimated shrinkage in Lake Chad reported in the literature includes 13 000-26 000 km 2 [22], 40 000-4837 km 2 from 1963 to 1997, an approximately 88% areal size [23] and 20 900-304 km 2 , an approximately 95% extent loss in 1963-2000 [24]. This present study observed fluctuation in the area extent of the lake, depicted as 4.08% decrease in 1987 to 2.6% in 1999 and then to 3.39% in 2005. This agrees with Adewuyi [23], who estimated 56% increase in the lake extent between 1997 and 2001. Also, Ebenki [24] estimated that the lake fluctuated in size in 1975-1990 between 8065 and approximately 12 813 km 2 (a 15% extent increase) and a decline in 2000-2007 from 10 011 to approximately 8251 km 2 (11% extent reduction). Generally, Ebenki [24] noted 15% increase in the lake area extent between 1975 and 1990, 9% extent decline between 1990 and 2000 and approximately 11% area decline in 2000-2007.

Bare ground and sandy soil
This is the portion of the supervised classified Landsat images of Lake Chad for individual year (1987, 1999 and 2005) that depicts its surrounding covered by sandy soil and the ground surface only (table 6). The increase in this land cover class from 13 490.8 km 2 in 1987 to 17 503.10 km 2 in 2005, and with a range of 6016.774 km 2 , could be due to incessant abstraction of the lake for various purposes, desertification and frequent drought experienced in the basin. Drought conditions have been linked to increase in bare ground around lakes and diminished surface extent [21,26,42]. Conventionally, prolonged droughts are frequently connected with dune or bare ground formation in lake and other wetland areas [21]. Oftentimes, there is a general inclination towards dune establishment and major changes in lake hydrology and vegetation, provided there is sufficient wind energy [21]. Fluctuations in the total area coverage by bare ground and sandy soil from 41.14% in 1987 to 34.14% in 1999 (decline trend), and from 34.14 to 49.12% in 1999 and 2005 (increase trend), could be connected to alternating encroachment, afforestation/agricultural activities and deforestation of the lake environment. Increasing farmland and livestock density, and overgrazing beyond the carrying capacity of the lake's grassland could additionally contribute to bare land formation [43]. A total range of 4.98% increase in the lake extent covered by bare ground and sandy soil obtained in this study is not without its attending negative consequences on the lake ecology and socio-economics of the drainage area. This could be manifested as reduction in farmable land area, dwindling food and animal feed resources, loss of fishing ground, water inaccessibility, forced migrations and resettlement, competition, conflict among settlers, recession farming, transboundary activities and national tension [21,43].

Shrunk part of the lake covered with soil and vegetation
This is the portion of the supervised classification of the Landsat images of Lake Chad intersected with a combination of vegetated ground and soil for individual year (1987, 1999 and 2005) (table 6). This land cover is primarily dominated by shrubs and fuel. A reduction in this areal extent from 11 046.44 km 2 (33.68%) in 1987 to 10 078.82 km 2 (28.28%) in 2005, totalling a 967.62 km 2 (5.40%) decrease in the area, could generally be adduced to over-exploitation of wood and hydro-climatic forces. Ecopastoral behaviours and grazing activities are potential linkages connected to the areal extent reduction in vegetated soil and ground coverage around the lake. Extensive bush burning, and deforestation, are other contributors to reduction in vegetated soil around the lake basin [44]. The disappearance of vegetated ground has ecologic effects on animal habitants. Game usually show migratory tendencies towards dense vegetated areas of the lake to seek shelter, and thus induce ecological imbalances in the lake ecosystem. Aggressive reactions and violence against intruders in game population could increase in species defence of their territories. Also, there is accompanying loss of both animal and plant biodiversity due to decrease in the lake vegetated soil and ground areal extent. Population shifts among human settlers towards farmable land area of the lake due to loss of fertile vegetated ground could be noticed. Southward migration among human settlers in search of grazing land, fertile farmland and irrigatable land has been documented along Lake Chad catchment for several years [21]. This has also resulted in degradation of the southern drainage area due to increased pressure on the available water resources, tree felling and livestock treading down of plants. Regional conflict and tension are other socio-economic factors. (14.06%) in 1987 to 9362.57 km 2 (27.83%) in 1999, and then a decline to 6342.15 km 2 (17.8%), could depict possible precipitation pattern and interaction of various influences. Generally, a total of 2.74% areal extent increase in the shrunk part of the lake covered by vegetation from 1987 to 2005 could be attributed to hydro-climatological variables including water level and precipitation that might modulate the lake biomass and areal vegetation. Ecological succession, natural disturbances, impoundment and damming of inflow tributaries are other factors that could contribute to areal variability of lake vegetation [34].

Shrunk part of the lake reduced to sand and turbid water
This is area extent of the supervised classification of Lake Chad neighbourhood in 1987, 1999 and 2005 reduced to combination of eroded sand dunes, unclean or unclear pools of water. This part is in no way regarded as part of the present water in Lake Chad (table 6). The 5.62% (1805.942 km 2 ) total decrease in the areal extent of the shrunk lake reduced to sand dune and turbid water from 1987 to 2005 could be ascribed to increasing successional colonization of the area by wetland plant species. Exposure of the zone due to contraction in lake water could lead to active germination of dormant lakeshore plant seeds and consequent regeneration of the region [45]. Increased lake water turbidity has been linked to direct modulation of submersed light, suppression of subaqueous vegetation and productivity, zonal stock reduction and senescence of fish/shrimp and threats to migratory bird habitats [45][46][47][48].
Other contributors to Lake Chad shrinkage include lake water discharge to groundwater, and dam construction on the tributaries. For instance, 45% water inflow reduction to the lake has been attributed to Tiga and Challawa dams constructed on Komadugu Yobe river in 1974 and 1992, respectively [21]. Also, 42%, approximately 60% and approximately 83% inflow reduction into the lake have been linked to Chari River, Kumadugu Yobe and El Beid, respectively, over the last 50 years [21]. Socio-economic outcome of the lake contraction could be evidenced as reduced crop yield and production, food and feed insecurity, biodiversity decline and loss, loss of fishing ground, increasing unemployment, increasing relocation and migration, high crime rate, and hunger and diseases [15].

Conclusion
The study revealed an alarming rate of Lake Chad contraction and damaging effects on the hydrologic potential, eco-sustainability and socio-economics of the drainage area. Also, ILWIS 3.3 showed a comparable utility in assessing hydro-meteorological events associated with freshwaters and wetlands. Its outputs concordantly fall within the same range and display similar patterns to other remote sensing instrumentations.
Therefore, the present study appraised Lake Chad shrinkage using remote sensing and instrumentation of ILWIS 3.3. This was the first investigation that analysed remote sensed dataset (Landsat imageries) of the lake and its surroundings using ILWIS 3.3.
Data accessibility. Landsat imageries and ILWIS 3.3 of the lake and its environs for this study were acquired from Global Land Cover Facility website (http://glcfapp.glcf.umd.edu/) and the National Space Research and Development Agency, Nigeria, respectively. The data attributes are presented in table 1. ILWIS 3.3 was acquired from NASRDA, Nigeria.