- Open Access
Evaluation and prediction of land use/land cover changes in the Andassa watershed, Blue Nile Basin, Ethiopia
Environmental Systems Researchvolume 6, Article number: 17 (2017)
Land use/land cover (LU/LC) change is the challenging and continuous drivers of environment change. Understanding the rate and process of change is, therefore, basic for managing the environment. This study was intended to analyze the LU/LC changes from 1985 to 2015 periods, and predict the situation to 2030 and 2045 in the Andassa watershed of Blue Nile basin, Ethiopia. The hybrid classification technique for extracting thematic information from satellite images and CA-Markov model for prediction of LU/LC were employed.
Cultivated land was expanding from 62.7% in 1985 to 73.1% in 2000 and to 76.8% in 2015. The area of built-up also slightly increased (0.1–1.1%) between 1985 and 2015 periods. In contrast, forest, shrubland and grassland were reduced from 3.5 to 1.9%, 26.2 to 15.3% and 7.6 to 4.9% in 1985 and 2015 periods, respectively. The increase of cultivated land and built-up area, and the withdrawing of forest, shrubland and grassland were further continued in 2030 and 2045 periods.
Significant amount of LU/LC conversions had occurred in the watershed from 1985 to 2015 periods, and expected to continue in 2030 and 2045 periods. Thus, appropriate interventions to revert the trends are very much critical.
Our planet earth is endowed with plenty of natural resources which sustain life for millennia. However, land use/land cover (LU/LC) changes are major environmental challenges in various parts of the world. Globally, there had been an increase of cropland and pastureland during 1970–1990 and 1700–1990 periods, respectively. Within these periods, Lambin et al. (2003) indicated that cropland and pastureland globally increased approximately five and six fold, respectively. The increase was at the expense of forest, natural grassland and savannas. Nevertheless, the direction of LU/LC change was not uniform in all parts of the world. In temperate forest, the increment was by almost 3*106 ha/year, while the tropical forest decreased by 12*106 ha/year (MEA 2005).
LU/LC changes are also common phenomena in Ethiopia. There was a rapid expansion of cultivated land at the expense of vegetative land cover types in various parts of the country. For example, Gete and Hurni (2001) study in Dembecha area of northwestern Ethiopia showed an increase of cultivated land from 39% in 1957 to 77% in 1995 while natural forest declined from 27 to 0.3%. The decline of forest cover from 50.9 to 16.7% was also observed in Upper Gilgel Abbay catchment of Blue Nile basin between 1973 and 2001 periods, basically due to the expansion of agricultural land (Rientjes et al. 2011). Gessesse and Kleman (2007) study in the South Central Rift Valley Region of Ethiopia also showed the reduction of natural forest cover from 16% in 1972 to 2.8% in 2000, which amounts to a total natural forest loss of 40,324 ha. In the entire area of Blue Nile basin also there had been a shrinking of wooded grassland, wood land, shrubs and bushes, natural forest and afro-alpine vegetation between 1973 and 2000 periods while rain fed cropland, grassland, water body and barren land had increased. The increase of water body is due to the construction of different dams in the basin (Gebremicael et al. 2013). Belay (2002) study in Derekolli catchment of the South Wello Zone also reported the decline of scrubland at the rate of 1.6 and 0.31% per year between 1957–1986 and 1986–2000, respectively. In contrast, cropland had increased from 65.1% in 1957 to 70.6% in 2000. In Northern Afar rangelands also a rapid reduction in woodland cover (97%) and grassland cover (88%) from 1972 to 2007 were reported, and in contrary bush land cover and cultivated land increased more than three and eightfold, respectively (Diress et al. 2010). A decrease in coverage of scrublands, riverine vegetation and forests, and an increase in open areas, settlements, floodplains, and water body were also observed in Kalu District of Southern Wello between 1958 and 1986 periods (Kebrom and Hedlund 2000). While, some studies conducted in the previously degraded parts of northern Ethiopia, revealed improvement of vegetation cover due to community afforestation and land rehabilitation activity (Amare (2007) and Amare et al. (2011) in Eastern Escarpment of Wello, Ethiopia; Muluneh (2003) in west Gurage land and Munro et al. (2008) in Tigray highlands). The increase of forest cover was also reported in Chemoga watershed from 1957–1998 periods (Woldeamlak 2002).
LU/LC change is increasingly recognized as an important driver of environmental change on all spatial and temporal scales. LU/LC change contributes significantly to earth atmosphere interactions, forest fragmentation, and biodiversity loss (Fu et al. 2000). In addition, it is also one of the factors for local environment disturbance by influencing runoff, soil loss and stream flow (Woldeamlak 2002). Due to these, modeling the dynamics of LU/LC is crucially important for managing the environment. The study area, Andassa watershed, is known to be the productive area in the country, and the head stream of Blue Nile River. Hence, identifying the rate and process of LU/LC changes is fundamental, which have national and international significance. However, the watershed LU/LC change is not well investigated. Thus, this study was aimed to analyze the LU/LC changes from 1985 to 2015 periods and predict the situation to 2030 and 2045 periods.
Andassa watershed is within the Blue Nile basin of Ethiopia (Fig. 1). It is situated approximately 560 km northwest of Addis Ababa (the capital city of Ethiopia) and in a close proximity to the capital city of the Amhara regional state (Bahir Dar). Geographically, the watershed extends between 11°08′N–11°32′N latitude and 37°16′E–37°32′E longitudes. The watershed covers a surface area of 58,760 ha belonging to three Administrative Districts (Woredas), which are Bahir Dar Zuria, Mecha and Yilmana Densa. The topography is hilly and elevation ranges from 1701 m to 3216 m a.s.l. Its agro-climate is remarkably dominated by sub-tropical climate (85.2%) with a small segment of temperate climate (14.8%). According to the data obtained from the GIS department of Ministry of Water and Energy, the major soil types in the studied watershed include Haplic Alisols, Eutric Leptosols, Chromic Luvisols, Haplic Nitisols and Eutric Vertisols. Its geology is also characterized by Alluvium, Ashangi basalts, basalts related to volcanic center and Termaber basalts. Andassa River is the major river of the studied watershed, which is also among the tributaries of Blue Nile River. Agriculture is the foremost economic activity and the main sources of livelihood for the population and rainfall is bimodal which include spring and summer rainfall.
Data types and sources
Three satellite images (Landsat-5 TM 1985, Landsat-7 ETM+ 2000 and Landsat-8 OLI–TIRS 2015) with 30 m spatial resolution were used for the LU/LC change analysis of the studied watershed. Details of the images characteristics are tabulated in Table 1. The data required for the study were collected from various sources. Landsat data were downloaded free of charge from U.S Geological Survey (USGS) Center for Earth Resources Observation and Science (EROS) (https://earthexplorer.usgs.gov/). ASTER GDEM with 30 m cell size was obtained from Aster Global Digital Elevation Map (http://gdex.cr.usgs.gov/gdex/), and river and road data were collected from the GIS department of Ministry of Water and Energy (Ethiopia). Population data of 1994 and 2007 at the smallest administrative unit (Kebele) were also obtained from the Ethiopian Central Statistics Agency. In addition to these secondary data, primary data were also obtained through extensive field works and in-depth focus group discussions with agricultural development agents and local elders.
Image classifications were carried out to extract useful thematic information (Boakye et al. 2008; Al-sharif and Pradhan 2013) from the three Landsat images (Table 1). Preprocessing tasks such as geometric and radiometric corrections (Giriraj et al. 2008; Schulz et al. 2010; Teferi et al. 2010; Mosammam et al. 2016; Temesgen et al. 2017) were applied before classifying the images. Image classifications were undertaken using the hybrid classification technique, which combines both unsupervised and supervised classification techniques (Teferi et al. 2010; Solomon et al. 2014). The hybrid classification technique improves the classification accuracy better than using either unsupervised or supervised classification techniques alone (Lillesand and Kiefer 2000). Primarily, unsupervised classification using Iterative Self-Organizing Data Analysis (ISODATA) clustering (Boakye et al. 2008; Teferi et al. 2010) method was undertaken as a baseline for collecting ground truth points. Using signature editor of unsupervised classes, a pixel based supervised classification with Maximum Likelihood Classification (MLC) algorism (Solomon et al. 2014; Temesgen et al. 2014a) was undertaken using the ground truth points collected from each LU/LC category. A total of 450 GPS points (75 GPS points in each LU/LC), which were collected between 9:00 a.m. and 5:00 p.m., were undertaken for supervised classification. The LU/LC classes together with their descriptions are presented in Table 2. In classifying the 1985 and 2000 images, reference data from Google earth images from the corresponding time periods were collected. Furthermore, geo-linking techniques and in-depth focus group discussions with local elders were also undertaken. ERDAS Imagine 2014 and ArcGIS 10.3 software were used for image classification and mapping purposes, respectively.
Accuracy assessment was done to understand the representation of the classified images on the ground (Congalton 1991; Congalton and Green 1999; Congalton 2005; Temesgen et al. 2014a; Mosammam et al. 2016). Any classified image without accuracy assessment limits the confidence of the result (Congalton 1991; Congalton and Green 1999). Accuracy assessment is commonly done with reference to other images (Congalton 1991; Congalton and Green 1999; Foody 2002; Congalton 2005; Mekuria 2005; Gessesse and Kleman 2007; Schulz et al. 2010; Teferi et al. 2010). To do accuracy assessment for the classified images, 480 random sample points in ArcGIS 10.3 was created. Reference points were collected for the 1985, 2000 and 2015 classified images from the corresponding Google Earth images (i.e. 05 February, 1985; 21 February, 2000 and 28 February, 2015, respectively). A similar procedure was followed by Abineh and Bogale (2015) and Temesgen et al. (2017). Then, the classified images were compared with the reference images by means of error matrix (Foody 2002; Schulz et al. 2010; Rientjes et al. 2011; Ariti et al. 2015). Various measures of accuracy assessment such as producer accuracy, user accuracy (Congalton 1991), overall accuracy and Kappa coefficient were done. Overall accuracy (Congalton 1991; Foody 2002; Congalton 2005) was calculated as Eq. 1 while Kappa coefficient (Congalton 1991) was calculated using Eq. 2.
where, OA is overall accuracy, X is number of correct values in the diagonals of the matrix, and Y is total number of values taken as a reference point.
where, K is Kappa coefficient, r is the number of rows in the matrix, xii is the number of observations in row i and column i, xi+ are the marginal totals of row i, x + i are the marginal totals column i, and N is the total number of observations.
The accuracy report of the 1985, 2000 and 2015 classified images are presented in Table 3. An overall accuracy of 86.9, 85.8 and 88.8% were attained for the 1985, 2000 and 2015 classified images, respectively. A kappa coefficient of 0.83, 0.81 and 0.85, respectively, were also obtained for 1985, 2000 and 2015 LU/LC maps. According to Monserud (1990), a Kappa values between 0.70 and 0.85 are generally rated as very good indicators of the classified image in representing the ground truths. Hence, the validation data set indicated a very good agreement of the classified image with the ground truths. The details of the accuracy assessment reports of the classified images are found in the Additional file 1: Appendix.
Land use/land cover prediction
Prediction of LU/LC conditions for the 2030 and 2045 periods were undertaken with Cellular Automata-Markov (CA-Markov) model. CA-Markov predicts not only the trend but also the spatial structure of different LU/LC categories (Arsanjani et al. 2011; Wang et al. 2012; Li et al. 2015). The model is widely applied in LU/LC change modeling elsewhere (Kamusoko et al. (2009) in Zimbabwe; Arsanjani et al. (2011) in Iran; Sang et al. (2011) in China; Al-sharif and Pradhan (2013) in Libya; Adhikari and Southworth (2012) and Singh et al. (2015) in India among others). Predictions were undertaken in IDRISI; Geospatial software for monitoring and modeling the Earth system; version 17.0 using the 2015 classified image as a basis LU/LC image and by considering factors and constraints (Clark Labs 2012; Eastman 2012; Omar et al. 2014; Singh et al. 2015). At first, however, transitions between 2000 and 2015 periods (with the proportion error of 0.15) were undertaken using Markov transition estimator in the IDRISI module. Factors are criterions that indicate the relative suitability of areas under consideration (Clark Labs 2012; Eastman 2012), and they are mapped in a continuous scale. Constraints are criterions which limits the alternatives under consideration (Clark Labs 2012; Eastman 2012), and they are mapped as Boolean image. In the Boolean images of each class, a value of either 1 or 0 was assigned for non constraint and constraint criterions, respectively. The factors and constraints considered were distance to river, distance to town, distance to road, proximate to developed area, suitable areas for conversion to each class, elevation and slope. Suitable areas for conversion to each class were assigned by giving the value for the five classes from 0 (no probability for conversion) to 1 (high probability for conversion). It was done through in-depth focus group discussions with agricultural development agents. Slope was considered as a common constraint for cultivated and built-up area categories since increasing slope gradient inhabits both cultivation and built-up purposes. However, slope was not considered as a constraint for forest, shrubland and grassland uses. Table 4 presents the details of factors and constraint considered for each LU/LC class with their weights. In-depth focus group discussions with agricultural development agents and local elders were held to assign a set of relative weights for a group of factors, and a very good acceptable consistence ratio (Clark Labs 2012) for the group of factors considered were obtained for each class as shown in Table 4. The factors and constraint were integrated using a Multi-Criteria Evaluation (MCE) decision support system with Weighted Liner Combination (WLE) fuzzy membership function to produce a single suitability map for each class. To do so, factors were first changed to binary format from 0 to 255; in which 255 is high suitable and 0 is none suitable. The factors and constraint maps considered for each LU/LC class can be found from the Additional file 1: Appendix where the suitability maps for each LU/LC class are shown in Fig. 2.
Model validation is a fundamental component in any modeling activity (Pontius and Schneider 2001; Al-sharif and Pradhan 2013; Singh et al. 2015; Mosammam et al. 2016). To check the quality of CA-Markov model in simulating future LU/LC conditions, the model was validated (Giriraj et al. 2008; Adhikari and Southworth 2012; Al-sharif and Pradhan 2013; Omar et al. 2014) after simulating the 2015 LU/LC conditions using the 1985 and 2000 classified images. Then, the simulated and the actual 2015 LU/LC maps were compared using the “Relative Operating Characteristic (ROC)” (Pontius and Schneider 2001) tool accessible in the IDRISI module. Furthermore, Kappa indexes (Mosammam et al. 2016) such as Kappa for no information (Kno), Kappa for location (Klocation), Kappa for stratum-level location (KlocationStrata) and Kappa for standard (Kstandard) (Clark Labs 2012; Omar et al. 2014; Mosammam et al. 2016) were also used to compare the agreements of the two maps. In addition, comparisons of the simulated and the actual area of each LU/LC class were also done.
Land use/land cover change
Change analysis is usually done to demonstrate the patterns of changes and to make useful decisions. After classification of images and projection of the 2030 and 2045 conditions, comparisons (Gete and Hurni 2001; Belay 2002; Schulz et al. 2010; Abate 2011; Rawat and Kumar 2015; Mosammam et al. 2016) between the subsequent periods were made to illustrate the changes between the periods. Conversion matrix (Mekuria 2005; Giriraj et al. 2008; Diress et al. 2010; Teferi et al. 2010; Abate 2011; Rientjes et al. 2011) between 1985 and 2000, 2000 and 2015, 2015 and 2030, and 2030 and 2045 periods were also done to distinguish the changes of each category at the expense of others. In addition, percent of change (Ebrahim and Mohamed 2017) and rate of change (Abate 2011; Temesgen et al. 2014a) was also computed to demonstrate the magnitude of the changes experienced between the periods using Eqs. 3 and 4, respectively.
where, X is area of LU/LC (ha) in time 2, Y is area of LU/LC (ha) in time 1, Z is Time interval between X and Y in years.
Furthermore, trends of LU/LC changes from 1985 to 2045 periods were also illustrated.
Methods of exploring the drivers of land use/land cover changes
It is generally apparent that LU/LC changes are driven by the interaction of natural and human forces (Meyer and Turner 1994; Belay 2002). The natural drivers such as climate change are felt after extended periods of time (Woldeamlak 2002; Ebrahim and Mohamed 2017). However, human drivers are immediate and often radical (Woldeamlak 2002). Whatever their speed and magnitude, most LU/LC changes have occurred due to human drivers (Gete and Hurni 2001). Hence, exploring the trend of population growth and its association with the observed LU/LC changes is very much crucial.
Thus, to appreciate the trend of population growth and its association with the observed LU/LC in the studied watershed, population data of the 1994 and 2007 census reports (CSA 1994, 2007) were used. As it is true in most of the time, obtaining population data of the entire studied watershed was indisputably difficult (Woldeamlak 2002; Ebrahim and Mohamed 2017). Due to this, the smallest administrative units called Kebeles whose entire areas are within the studied watershed were used for this purpose. Accordingly, ten Kebeles from the three administrative Districts (Bahir Dar Zuria, Mecha and Yilmana Densa) were purposefully selected. Using the data, various statistics such as growth between 1994 and 2007, rate of growth (%) and doubling time in years were computed, and associated with the observed LU/LC changes. Rate of growth (%) and doubling time in years was calculated using Eqs. 5 and 6 (Woldeamlak 2002; Abate 2011), respectively.
where, r is growth rate in percent, Pt2 is the population at time 2, Pt1 is the population at time 1, n is the number of years between time 1 and time 2, and DT is doubling time in year.
To asses other drivers of LU/LC changes, focus group discussions with local elders and agricultural development agents were also carried out.
Results and discussion
Land use/land cover change analysis
Analysis of LU/LC patterns in the studied watershed indicated the growth of cultivated land and built-up area at the expense of vegetative cover types over the last three decades (Table 5; Fig. 3). During these periods, cultivated land has expanded from 62.7% in 1985 to 73.1% in 2000, and again to 76.8% in 2015 (Table 5). Between 1985–2000 and 2000–2015 periods, it was increased by 16.6 and 5.1%, respectively. The rate of increment during 1985–2000 and 2000–2015 periods were 407 and 145.5 ha/year, respectively (Table 6). Similarly, built-up area had also increased from 0.1 to 0.2% and to 1.1% in 1985, 2000 and 2015 periods, respectively (Table 5). During 1985–2000 and 2000–2015 periods, built-up area increased by 192.5 and 562.8%, respectively (Table 6). The rapid percent change taken place in built-up area during these periods is associated with the nearest location of Bahir Dar town to the study site as can be seen most of the increased in built-up area occurred in the north-eastern areas of the studied watershed. This reason was also mentioned during focus group discussions of agricultural development agents. In contrast, forest, shrubland and grasslands had decreased in the whole study periods. For example, forest coverage decreased from 3.5% in 1985 to 2.6% in 2000 and to 1.9% in 2015 (Table 5) with the annual diminishing rate of 37.6 and 24.4 ha/year between 1985–2000 and 2000–2015 periods, respectively (Table 6). Similarly, shrubland and grasslands also decreased at a rate of 328.7 and 97 ha/year, and 45.2 and 62.2 ha/year, respectively, between 1985–2000 and 2000–2015 periods. During 1985 to 2015 periods, forest, shrubland and grassland have shown a reduction in size (Table 6).
The finding of this research is consistent with other studies carried out by Gete and Hurni (2001) in Dembecha area of northwestern Ethiopia; Gessesse and Kleman (2007) in South Central Rift Valley Region of Ethiopia; Rientjes et al. (2011) in Upper Gilgel Abbay catchment of Blue Nile basin; Gebremicael et al. (2013) in Blue Nile basin; Temesgen et al. (2014a) in Dera District of northwestern Ethiopia; Solomon et al. (2014) in Birr and Upper-Didesa watersheds of Blue Nile basin, where the agricultural land increased significantly where forest land was shrinking. A study in Shomba and Michity catchments of Kefa zone (Mekuria 2005) also indicated the conversion of vegetative lands into non-vegetative lands between 1987 to 2001 periods mainly for the expansion of cultivated land and settlement. Studies by Ebrahim and Mohamed (2017) in Geleda catchment and Solomon et al. (2010) in Koga watershed also reported the growth of cultivated lands at the reduction of forest cover in the respective study periods.
Analysis of land use/land cover conversions
Conversions of LU/LC from one category to another are common phenomena in LU/LC studies (Belay 2002). The conversions of one LU/LC category to another between 1985–2000 and 2000–2015 periods are presented in Tables 7 and 8. The diagonals in the matrix from the tables are the persistence while the off-diagonals are the conversions from one category to the others. The change detection analysis indicated the significant conversions in LU/LC in both periods.
Conversions between 1985 and 2000
In these periods, 8216, 2230, 547 and 1 ha of cultivated land were converted from shrubland, grassland, forest and built-up area, respectively. While cultivated land gained from other LU/LC categories, a significant area of cultivated land were also reverted to shrubland, grassland, forest and built-up area (Table 7). During these time, some area of built-up were also converted from cultivated land (47 ha), grassland (26 ha) and shrubland (2 ha). Although it is a small proportion, 5, 3 and 1 ha of built-up area were also in reverse converted to grassland, shrubland and cultivated land, respectively. Gains and losses in forest, shrubland and grassland were also taken place during these periods (Table 7). For example, 8216 ha of cultivated land, 651 ha grassland, 408 ha of forest and 2 ha of built-up area were altered from shrubland. In reverse, a considerable area of shrubland were also reverted from cultivated land (3009 ha), grassland (794 ha), forest (542 ha) and built-up area (3 ha).
Conversions between 2000 and 2015
During these periods, 4598, 322, 207 and 148 ha of shrubland were converted to cultivated land, forest, grassland and built-up area, respectively. About 2484, 513, 110 and 16 ha of grassland were also reverted to cultivated land, shrubland, built-up area and forest, respectively. An estimated 575, 297, 12 ha of forest were also converted to shrubland, cultivated land and grassland, respectively. Similarly, shrubland, grassland and forest were also gained from other LU/LC categories (Table 8). In these periods, a significant area of cultivated land were converted from shrubland (4598 ha), grassland (2484 ha), forest (297 ha) and built-up area (9 ha). In reverse, there was also a considerable conversion of cultivated land to other categories. A significant amount of gains and losses in built-up area was also occurred in these periods (Table 8).
Generally, during the two periods the conversion of cultivated land from shrubland is higher than any other category of conversions. Studies conducted in various parts of the country also reported the conversions of one category to others. For instance, a study in northern Afar rangelands (Diress et al. 2010) reported the conversions of scrubland, bushy grassland, and grassland to cultivated land between 1972 and 2007 periods. Woodlands were converted to bushland, scrubland and bushy grassland. Similarly, a study in Upper Gilgel Abbay catchment of Blue Nile basin between 1973 and 2001 periods (Rientjes et al. 2011) also reported the conversions of one category to the other and the gains and losses between each category. During 1973 to 1986 and 1986 to 2001 periods, the greatest conversions taken place were the conversions of shrubland, forest and grassland into cultivated lands.
Drivers of land use/land cover changes
It is quite observed that, population has grown in the studied watershed through 1994 to 2007 periods. The increase of population between these periods varied from 11% in Wetet Ber Kebele to 51% in Kimbaba Kebele (Table 9). Population growth increases demands of more cultivated land, fuel wood, charcoal and infrastructural development; which leads to vegetative cover losses. Hence, the increased of population number is certainly the primary driver of LU/LC changes in the studied watershed as observed in the sample Kebeles (Table 9), which was manifested largely through the expansion of cultivated lands at the expense of vegetative land covers (Table 4). A result from focus group discussions has also confirmed that the reduction of land productivity, which leads the intension of the people for getting new fertile cultivable lands, is the other important driver to these changes.
Similar with the conclusion drawn above, population growth was also the most important factor for the LU/LC dynamics in the Chemoga watershed of northwestern highland of Ethiopia (Woldeamlak 2002); Dendi District, Ethiopia (Berhan 2010); Dera District of northwestern Ethiopian highlands (Temesgen et al. 2014b) and Geleda catchment of northwestern highlands (Ebrahim and Mohamed 2017) to mention among others. Similarly, Hurni et al. (2005) also noted that population increment from the mid to the turn of the twentieth century at the country level (Ethiopia) accelerated deforestation and intensified cultivation.
Future land use/land cover change
CA-Markov model validation
Visual comparison of the 2015 simulated and actual maps (Fig. 4a, b) are reasonably similar. The area extent of the two maps (Table 10) also illustrates the acceptable range of decisions, in which agriculture accounts 77.1 and 76.8% in the simulated and actual maps, respectively. In addition, there are also approximately equal proportions of forest, shrubland and grasslands between the simulated and the actual maps. The less effective projected LU/LC category is built-up area. This is due to models’ less ability in capturing the randomly new developed areas. Validations through different statistics have been also done, and ROC value of 89.5% was achieved. All the kappa statistics such as Kno (87.7%), Klocation (82.2%), KlocationStrata (82.2%) and Kstandard (81.6%) were also above 80% (Table 10), which indicates the good performance of the model in simulating future LU/LC conditions (Singh et al. 2015; Mosammam et al. 2016).
The predicted land use/land cover
The projected 2030 and 2045 LU/LC conditions using CA-Markov model are indicated in Fig. 5a, b while their area coverage is presented in Table 11. According to the resulted maps, the area of cultivated land has grown from 76.8% in 2015 to 83.3% in 2030 and to 85.8% in 2045. A continuous increase of built-up area have been also observed from 2015 (1.1%) to 2030 (2%) and to 2045 (5.9%) periods. In contrast, a diminishing of forest cover from 1.9 to 1.5% and to 1.3%; shrubland from 15.3 to 9.2% and to 4.7% and grassland from 4.9 to 4% and to 2.3% were observed through 2015, 2030 and 2045 periods, respectively. In general, the increased of cultivated land and built-up area at the cost of vegetative LU/LC classes will be observed through 2015, 2030 and 2045 periods.
Future land use/land cover conversions
Conversions between 2015 and 2030
Within these periods, 3890, 1218, 113 and 36 ha of cultivated land were gained from shrubland, grassland, forest and built-up area, respectively. About 389, 135 and 39 ha of built-up area were also reverted from cultivated land, shrubland and grassland, respectively. Although cultivated and built-up area gained from other categories, there was also a significant loss of cultivated and built-up area to other categories (Table 12). Between these periods, there was a huge loss of forest, shrubland and grassland to other categories. For example, about 3890, 135, 69 and 4 ha of shrubland were converted to cultivated land, built-up area, grassland and forest, respectively. About 141, 113 and 5 ha of forest were also converted to shrubland, cultivated land and grassland, respectively. The grassland converted to cultivated land, built-up area, shrubland and forest were 1218, 39, 11 and 1 ha, respectively. In contrast, forest, shrubland and grassland were also gained from other categories (Table 12).
Conversions between 2030 and 2045
A significant conversion of LU/LC categories has been also taken place between these periods (Table 13). For example, 2466, 1083, 103 and 20 ha of cultivated land were reverted from shrubland, grassland, forest and built-up area, respectively. A considerable area of cultivated land (2115 ha), shrubland (117 ha) and grassland (41 ha) were also converted to built-up area. A substantial area of cultivated land and built-up area were also reversed to other categories. In contrast, there was also a huge conversion of forest, shrubland and grassland to other categories. For instance, 103, 34 and 2 ha of forest were converted to cultivated land, shrubland and grassland, respectively. An estimated 2466, 117, 80 and 2 ha of shrubland were also converted to cultivated land, built-up area, grassland and forest, respectively. The grassland reverted to cultivated land, built-up area and shrubland were 1083, 41 and 2 ha, respectively. Even though, forest, shrubland and grasslands experienced a significant losses (Table 13), there was also a gain of forest, shrubland and grassland from other categories.
Generally, cultivated land and built-up area have shown an increasing trend from 1985 to 2045 periods. In contrast, vegetative land cover types such as forest, shrubland and grasslands are in a decreasing trend (Fig. 6).
Significant amount of LU/LC conversions had occurred from 1985 to 2015 periods in the Andassa watershed. Cultivated land and built-up area had increased in the period of 1985 to 2015. In contrast, forest, shrubland and grassland have decreased in coverage. The trend of increasing in cultivated land and built-up area and decreasing in forest, shrubland and grassland LU/LC categories are expected continued in 2030 and 2045 periods. Population growth and reduction of land productivity are the drivers of such changes. If the trends of LU/LC changes continued, it will have implications on increasing soil loss and impacting the hydrology of the studied watershed in particular and the Blue Nile basin in general. Hence, reversing the projected conditions is very much important for maintaining the productivity of the studied watershed.
It is recommended that experts of environment, conservation, resource management and sustainability, ecology, biodiversity and eco-system are required to develop a plan for sustainable use of the site to ensure its functions for the next generations.
Central Statistics Agency
Digital Elevation Model
- ETM+ :
Enhanced Thematic Mapper
Geographic Information System
Global Positioning System
Maximum Likelihood Classification
Operational Land Imager–Thermal Infrared Sensor
Relative Operating Characteristic
Iterative Self-Organizing Data Analysis
Weighted Liner Combination
Abate S (2011) Evaluating the land use and land cover dynamics in Borena Woreda of South Wollo highlands, Ethiopia. J Sustain Dev Afr 13(1):87–105
Abineh T, Bogale T (2015) Accuracy assessment of land use land cover classification using google earth. AJEP 4(4):193–198
Adhikari S, Southworth J (2012) Simulating forest cover changes of Bannerghatta National Park based on a CA-Markov model: a remote sensing approach. Remote Sens 4:3215–3243
Al-sharif AA, Pradhan B (2013) Monitoring and predicting land use change in Tripoli Metropolitan City using an integrated Markov chain and cellular automata models in GIS. Arab J Geosci 7:4291–4301
Amare B (2007) Landscape transformation and opportunities for sustainable land management along the Escarpment of Wello, Ethiopia. Ph.D. Thesis, Bern University, Bern
Amare B, Hurni H, Gete Z (2011) Responses of rural households to the impacts of population and land-use changes along the Eastern Escarpment of Wello, Ethiopia. Nor J Geogr 65:42–53
Ariti AT, Vliet JV, Verburg PH (2015) Land-use and land-cover changes in the Central Rift Valley of Ethiopia: assessment of perception and adaptation of stakeholders. Appl Geogr 65:28–37
Arsanjani JJ, Kainz W, Mousivand AJ (2011) Tracking dynamic land-use change using spatially explicit Markov Chain based on cellular automata: the case of Tehran. Int J Image Data Fusion. doi:10.1080/19479832.2011.605397
Belay T (2002) Land-cover/land-use changes in the Derekolli catchment of the South Welo Zone of Amhara Region, Ethiopia. EASSRR 18(1):1–20
Berhan G (2010) The role of geo-information technology for predicting and mapping of forest cover spatio-temporal variability: Dendi district case study, Ethiopia. J Sustain Dev Afr 12(6):9–33
Boakye E, Odai SN, Adjei KA, Annor FO (2008) Landsat images for assessment of the impact of land use and land cover changes on the Barekese Catchment in Ghana. Eur J Sci Res 22(2):269–278
Congalton R (1991) A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens Environ 37:35–46
Congalton RG (2005) Thematic and positional accuracy assessment of digital remotely sensed data. Proceedings of the seventh annual forest inventory and analysis symposium. pp 149–154
Congalton RG, Green K (1999) Assessing the accuracy of remotely sensed data: principles and practices. Lewis Publishers, Boca Raton, p 137
CSA (Central Statistics Authority) (1994) The 1994 population and housing census of Ethiopia: results for Amhara region. Volume I, Part I. Statistical report on population size and characteristics. Central Statistics Authority, Addis Ababa
CSA (Central Statistics Authority) (2007) Report of the 2007 population and housing census of Ethiopia. Central Statistics Authority, Addis Ababa
Diress T, Moe SR, Vedeld P, Ermias A (2010) Land-use/cover dynamics in northern Afar rangelands, Ethiopia. Agric Ecosyst Environ 139:174–180
Eastman J (2012) IDRISI Selva manual, version 17. Clark University, Worcester
Ebrahim EA, Mohamed A (2017) Land use/cover dynamics and its drivers in Gelda catchment, Lake Tana watershed, Ethiopia. Environ Syst Res 6(4):1–13
Foody GM (2002) Status of land cover classification accuracy assessment. Remote Sens Environ 80:185–201
Fu B, Chen L, Ma K, Zhou H, Wang J (2000) The relationships between land use and soil conditions in the hilly area of the Loess Plateau in northern Shaanxi, China. CATENA 36:69–78
Gebremicael TG, Mohamed YA, Betrie GD, van der Zaag P, Teferi E (2013) Trend analysis of runoff and sediment fluxes in the Upper Blue Nile basin: a combined analysis of statistical tests, physically-based models and land use maps. J Hydrol 482:57–68
Gessesse D, Kleman J (2007) Pattern and magnitude of deforestation in the South Central Rift Valley Region of Ethiopia. Mt Res Dev 27:162–168
Gete Z, Hurni H (2001) Implications of land use and land cover dynamics for mountain resource degradation in the northwestern Ethiopian Highlands. Mt Res Dev 21(2):184–191
Giriraj A, Irfan-Ullah M, Murthy MS, Beierkuhnlein C (2008) Modeling spatial and temporal forest cover change patterns (1973–2020): a case study from South Western Ghats (India). Sensors 8:6132–6153
Hurni H, Kebede T, Gete Z (2005) The implications of changes in population, land use, and land management for surface runoff in the Upper Nile Basin area of Ethiopia. Mt Res Dev 25(2):147–154
Kamusoko C, Aniya M, Adi B, Manjoro M (2009) Rural sustainability under threat in Zimbabwe–simulation of future land use/cover changes in the Bindura district based on the Markov-cellular automata model. Appl Geogr 29:435–447
Kebrom T, Hedlund L (2000) Land cover changes between 1958 and 1986 in Kalu District, Southern Wello, Ethiopia. Mt Res Dev 20(1):42–51
Labs Clark (2012) Idrisi Selva help system. Clark University, Worcester
Lambin E, Geist H, Lepers E (2003) Dynamics of land use and land cover change in tropical regions. Annu Rev Environ Resour 28:206–232
Li S, Jin B, Wei X, Jiang Y, Wang J (2015) Using CA-Markov model to model the spatiotemporal change of land use/cover in Fuxian Lake for decision support. International workshop on spatiotemporal computing, 13–15 July 2015, Fairfax, Virginia, USA
Lillesand T, Kiefer R (2000) Remote sensing and image interpretation, 4th edn. Wiley, New York
MEA (Millennium Ecosystem Assessment) (2005) Ecosystems and human well-being: Synthesis. Island Press, Washington, DC
Mekuria A (2005) Forest conversion-soil degradation-farmers’ perception nexus implications for sustainable land use in the southwest of Ethiopia. In: Vlek P, Denich M, Martius C, Rodgers C and Giesen N (ed) Ecology and development series no. 26, p 161
Meyer WB, Turner BL (eds) (1994) Changes in land use and land cover: a global perspective. Cambridge University Press, Cambridge
Monserud RA (1990) Methods for comparing global vegetation maps, Report WP-90-40. IIASA, Laxenburg
Mosammam H, Nia J, Khani H, Teymouri A, Kazem M (2016) Monitoring land use change and measuring urban sprawl based on its spatial forms: the case of Qom city. Egypt J Remote Sens Space Sci. doi:10.1016/j.ejrs.2016.08.002
Muluneh W (2003) Population growth and environmental recovery: more people, more trees; lesson learned from western Gurageland. Ethiop J Soc Sci Humanit 1(1):1–33
Munro R, Deckers J, Mitiku H, Grove A, Poesen J, Nyssen J (2008) Soil landscapes, land cover change and erosion features of the Central Plateau region of Tigrai, Ethiopia: photo-monitoring with an interval of 30 years. CATENA 75:55–64
Omar NQ, Ahamad MS, Hussin WM, Samat N, Ahmad SZ (2014) Markov CA, multi regression, and multiple decision making for modeling historical changes in Kirkuk City, Iraq. J Indian Soc Remote Sens 42(1):165–178
Pontius RG, Schneider LC (2001) Land-cover change model validation by an ROC method for the Ipswich watershed, Massachusetts, USA. Agric Ecosyst Environ 85:239–248
Rawat JS, Kumar M (2015) Monitoring land use/cover change using remote sensing and GIS techniques: a case study of Hawalbagh block, district Almora, Uttarakhand, India. Egypt J Remote Sens Space Sci 18:77–84
Rientjes TH, Haile AT, Kebede E, Mannaerts CM, Habib E, Steenhuis TS (2011) Changes in land cover, rainfall and stream flow in Upper Gilgel Abbay catchment, Blue Nile basin-Ethiopia. Hydrol Earth Syst Sci 15:1979–1989
Sang L, Zhang C, Yang J, Zhu D, Yun W (2011) Simulation of land use spatial pattern of towns and villages based on CA–Markov model. Math Comput Model 54:938–943
Schulz JJ, Cayuela L, Echeverria C, Salas J, Rey Benayas JM (2010) Monitoring land cover change of the dryland forest landscape of Central Chile (1975–2008). Appl Geogr 30:436–447
Singh SK, Sk Mustak, Srivastava PK, Szabó S, Islam T (2015) Predicting spatial and decadal LULC changes through Cellular Automata Markov Chain models using earth observation datasets and geo-information. Environ Process 2:61–78
Solomon G, Ayele T, Bishop K (2010) Forest cover and stream flow in a headwater of the Blue Nile: complementing observational data analysis with community perception. Ambio 39(4):284–294
Solomon G, Woldeamlak B, Ga¨rdena¨s AI, Bishop K (2014) Forest cover change over four decades in the Blue Nile Basin, Ethiopia: comparison of three watersheds. Reg Environ Change 14:253–266
Teferi E, Uhlenbrook S, Bewket W, Wenninger J, Simane B (2010) The use of remote sensing to quantify wetland loss in the Choke Mountain range, Upper Blue Nile basin, Ethiopia. Hydrol Earth Syst Sci 14:2415–2428
Temesgen G, Amare B, Abraham M (2014a) Evaluation of land use/land cover changes and land degradation in Dera District, Ethiopia: GIS and remote sensing based analysis. Int J Sci Res Environ Sci 2(6):199–208
Temesgen G, Amare B, Abraham M (2014b) Population dynamics and land use/land cover changes in Dera District, Ethiopia. GJBAHS 3(1):137–140
Temesgen G, Taffa T, Mekuria A (2017) Erosion risk assessment for prioritization of conservation measures in Geleda watershed, Blue Nile basin, Ethiopia. Environ Syst Res 6(1):1–14
Wang SQ, Zheng XQ, Zang XB (2012) Accuracy assessments of land use change simulation based on Markov-cellular automata model. Procedia Environ Sci 13:1238–1245
Woldeamlak B (2002) Land cover dynamics since the 1950s in Chemoga watershed, Blue Nile basin, Ethiopia. Mt Res Dev 22(3):263–269
TG carried out designing the research idea, method design, field data collection, data analysis and interpretation, prepare draft of the manuscript, and structuring the report; TT, MA and AW participated in method design, data analysis and interpretation, and structuring the report. All authors read and approved the final manuscript.
Temesgen Gashaw: is a PhD candidate in Environmental Science at Addis Ababa University and lecturer at department of Natural Resource Management in Adigrat University. He has given watershed management, land degradation and rehabilitation, land use planning, and GIS and remote sensing courses, and also published more than 16 articles in the internationally peer reviewed journals.
Taffa Tulu: is a professor of Hydrology and Watershed Management in Addis Ababa University in Center of Environment and Development under College of Development Studies. He has authored five books and more than 40 publications. His spheres of professional expertise are Agricultural Engineering; Land Improvement and Water Management; Hydrology; and Irrigation and Water Engineering.
Mekuria Argaw (PhD): is associate professor of Environmental Science at the College of Natural Science in Addis Ababa University. He specializes in Ecology and Natural Resources Management. He teaches courses on watershed management and land degradation. Dr. Mekuria has published several peer reviewed papers on soil erosion, land degradation, biodiversity, watershed processes and climate change impacts.
Abeyou Wale (PhD): is a researcher at Texas A & M University, USA. He did his B.Sc degree in Hydraulic Engineering from Arba Mench University (Ethiopia), his Masters in Integrated Watershed Modeling and Management from faculty of Geo-information science (ITC), Ithaca, the Netherlands, and his PhD in Biological and Environmental Engineering from Cornel University, USA.
The Authors greatly acknowledge Center for Environmental Science, Addis Ababa University and Adigrat University for financial support; the GIS department of Ministry of Water and Energy for providing river and road data; Ethiopian Central Statistics Agency for providing population data; agricultural development agents and local elders of the watershed for their interest to focus group discussions.
The authors declare that they have no competing interests.
Availability of data and materials
The three landsat images were downloaded from U.S Geological Survey (USGS) Center for Earth Resources Observation and Science (EROS). ASTER GDEM was obtained from Aster Global Digital Elevation Map, and river and road data were collected from Ministry of Water and Energy. In addition, primary data were also obtained through extensive field works and in-depth focus group discussions with agricultural development agents and local elders.
This research was funded by Addis Ababa University and Adigrat University.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.