Introduction

Land use is describe as human uses of the land, or immediate actions modifying or converting land cover (Lambin et al. 2003). Whereas, land cover refers to the natural vegetative cover types that characterize a particular area. But these terms sometimes are too difficult to differentiate (De Sherbinin 2002). Similarly, Gomarasca (2009) stated land cover as the physical surface of the earth, including various combinations of natural and cultivated vegetation and manmade infrastructures. Land use, instead, includes both the way in which the Earth’s biophysical attributes are modified and the reasons for which they are altered (Messay 2011). The land use is affected by human activities, especially with reference to those who decide about land management, institutions included.

The process of land cover and use has been a continuously occurred at all times in the past, are presently ongoing, and are likely to continue in the future (Gomarasca 2009). However, over two decades, the issue of LULC change has gained considerable attention in a number of fields, for example remote sensing, political ecology and biogeography (Getachew et al. 2011). This is partly because it is the most important component of global change (Chamaporn and Tonny 2011). There are many driving forces contributing to LULC change which could be social, economic, institutional or geographic in nature. The global LULC change is a result of two main factors such as natural, and human forced, such as population growth, economic policies and domestic behaviour (Bezuayehu and Sterk 2008). Nevertheless, global population growth and increased economic development over the last century have been major causes of land use changes. Natural land covers have been transformed to agricultural lands and urban areas to meet the increasing demands for food, water, consumption of goods and other public utilities for over six billion people worldwide (Amare and Kameswara 2011).

Land use and land cover changes have many ecological, social and economic impacts on biotic diversity, soil degradation, and the ability of biological systems to support human needs (Hussein 2009). Land use and land cover changes also determine, in part, the vulnerability of places and people to climatic, economic, or sociopolitical perturbations. When aggregated globally, LULC changes significantly affect central aspects of earth system functioning (Lambin et al. 2003). LULC change plays an important role in global environmental change and sustainability, including response to climate change, effects on ecosystem structure and function, species and genetic diversity, water and energy balance, and agro-ecological potential (Hussein 2009).

There are many driving forces contributing to LULC change which could be social, economic, institutional or geographic in nature (Lepers et al. 2005). Natural land covers have been transformed to agricultural lands and urban areas to meet the increasing demands for food, water, consumption of goods and other public utilities for over seven billion people worldwide (Chamaporn and Tonny 2011 ). For example, Europe and eastern China experienced first the most rapid cropland expansion during the eighteenth century. A very gradual cropland expansion occurred in Africa, South and Southeast Asia, Latin America, and Australia until 1850, but since then, these regions have experienced dramatic increases in cropland, especially during the second half of the twentieth century. The greatest cropland expansion in the twentieth century occurred in South and Southeast Asia (Lambin et al. 2003).

Like in many other developing countries, most of the population of Ethiopia live in rural areas and depend directly on the land for its livelihood (Woldeamlak and Sterk 2005). In this regard, Kiros (2008) indicated that in response to the increasing demands for food production, agricultural lands are expanding at the expense of natural vegetation and grass lands. According to Hussein (2009) and Amare (2013a, b) in their study in the Ethiopian highlands concluded that LULC changes were mainly driven by human actions. Consequently, LULC changes could lead to a decreased availability of different products and services for human, livestock, agricultural production and damage to the environment as well.

Thus, comprehensive knowledge of these dynamics is useful to reconstruct past LULC changes and also predict future changes, and to elaborate sustainable land resource management practices aimed at preserving essential landscape functions (Hietel et al. 2004). As a result, there is the need to understand LULC change: its dynamics, the possible causes, its effect on the environment, the economic and social impacts as well their role on watershed degradation. There is also a continuing demand for accurate and up-to-date LULC change information for any kind of sustainable development programs where LULC serves as one of the major input criteria (Das 2009). However, to study LULC change, it is necessary to employ advanced technologies which have the potential to provide timely, accurate and reliable information. GIS and remote sensing in this perspective have been proved to be important tools in LULC change studies. These days GIS and remote sensing capabilities are utilized in various fields. LULC change detection is one of the GIS and remote sensing applications in environmental change studies.

Investigating LULC change has become an important tool to generate evidence for decision-makers, spatial planners, local communities, or actors who are operating within a given landscape to formulate appropriate policies and strategies. Hence, LULC change analysis helps decision-makers ensure sustainable development and understand the dynamics of the changing environment (Carmel and Kadmon 1998). Identifying the driving forces behind LULC changes, and developing appropriate measures to minimize their ecological effects have great deal of importance for land use planning. Specifically, this study determines the land cover and use status with special reference to its causative factors through time in Koga watershed, one of the most densely populated areas of Ethiopia. Thus, this study was intended to assess the LULC change and causative factors in Koga watershed by using GIS and remote sensing techniques for the past thirty eight years (1973–2011). Specifically the study was, amid to investigate LULC change drivers between 1973 and 2011 at the watershed level.

Research methods

Description of study area

Koga watershed, is situated in Northwestern part of Ethiopia, southern part of Lake Tana catchment, between 110 10′ and 110 25′ North latitude and 370 2′ and 370 17′ East longitude (Fig. 1). The watershed consists of land drained by the Koga River above its confluence with the Gilgel Abay. The Koga River flows south to Northwest with a total length of 49 km (Fikru 2009). Amare and Kameswara (2011) also states that, the Koga River is a tributary of the Gilgel Abbay and it joins to the Gilgel Abbay River downstream of the town of Wetet Abbay, at an altitude of 1985 m a.s.l. The watershed is characterized by tapered, strongly dissected highlands to the south, and a relatively flat plateau in the north. The elevation ranges from 1800 to 3200 m above sea level (masl).

Fig. 1
figure 1

Map of Koga River watershed

Full size image

The rainfall in the Koga watershed area has a uni-modal characteristic with one rainy and one dry season. The dry season occurs between November and April, while the wet season is observed between the months of May and October. Some rain is also received during the months of April and May. Mean annual rainfall is approximately 1560 mm, and a mean daily temperature is between 16 and 200c (Tilashwork 2009). The annual mean temperature in Koga watershed is about 220c. The typical economic activity in the study area is a mixed crop-livestock small holder farming system.

Research methods

Primary data were acquired from field survey using the Garmin GPS and structured household questionnaire, focus group discussions and interview with key informants. Landsat imageries used for this study were selected based on availability, percentage of cloud cover and correspondence with years of major events in the study area. All Landsat imageries were accessed free of charge from the US Geological Survey (USGS) Center for Earth Resources Observation and Science (EROS) via http://glovis.usgs.gov/. The nonaligned scenes were geocoded to the underlying base layers by using control points. Re-projection to the local projection system was made (UTM, map projection; Clarke 1880, Spheroid; and Adindan Datum). A freely available IDL and ERDAS software used for the correction. Atmospheric and radiometric correction of all Landsat images acquired for the different times were also made using a fully image-based technique developed by Chavez (1996) known as the COST model; this model derives its input parameters from the image itself. In a next step, temporal normalization of all the images was achieved by applying regression equations to the 1973, 1986, 1995, and 2011 imageries which predict what a given Brightness Value (BV) would be if it had been acquired. This pre-processing step reduces pixel brightness value variations caused by non-surface factors and enhances variations in pixel BV between dates related to actual changes in surface conditions. The 2011 TM image was selected as the reference scene to which the other images were normalized, because it was the only year for which quality in situ ground reference data were available.

A hierarchical land use classification was derived from the authors’ prior knowledge of the study area and based on the most popular scheme of the U.S. Geological Survey Land Use/Cover System devised by Anderson et al. (1976). A hybrid unsupervised/supervised classification approach was integrated with successive Geographic Information System (GIS) operations (spatial analysis) to classify the imageries of 1973 (MSS), 1986 (TM), 1995 (TM) and 2011 (TM) to land classes (Table 1).. First, to determine the spectral classes Iterative Self-Organizing Data Analysis (ISODATA) clustering was performed. Ground truth (reference data) was collected to associate the spectral classes with the cover types of the already defined classification scheme for the 2011 (TM) imagery from field observations.

Table 1 Description of LULC types in the Koga watershed
Full size table

The ground truth data collection was carried out for training and ground verification for image classification and accuracy assessment respectively. To assess the accuracy of thematic information derived from 1973 (MSS), 1986 (TM), 1995 (TM) and 2011 (TM) imageries, the design-based statistical inference method was employed which provides unbiased map accuracy statistics (Jensen 2005). Among other sources of information, reference data was collected from previous maps for the 1995 image (Zeleke and Hurni 2001). A total of 2277 reference data points for each year were obtained, of which 759 points were used for accuracy assessment and 1518 points were used for classification. Training sites were developed from the ground truth data collected to generate a signature for each land cover type. The number of samples required for each class was adjusted based on the proportion of the class and inherent variability within each category. Discrete multivariate analytical techniques were used to statistically evaluate the accuracy of the classified maps (Foody 2002). As there is no single universally accepted measure of accuracy for this study a kappa index was accepted (Stehman 1997).

The Post classification comparison was employed using separately classified landsat images and then a comparison was made for the LULC maps of 1973–1986, 1986–1995 and 1995–2011. Hence, change matrix was applied in ERDAS Imagine 10 to determine type of LULC changed and the direction of LULC change in the study area. In this study the farm and settlement areas were included in the same land cover, because settlements in the rural areas are near their farmlands and difficult to separate from low resolution imageries.

In order to acquire supplementary data related to LULC change in the study area, the structured questionnaire was used and data related to demographic features of households, land holding size, crop production, livestock production, sources of energy etc. The intensive interviews and focus group discussion were held with individuals who have lived at least 20 and above years in the study area. The focus group discussion was organized to obtain additional information on historical and current LULC changes in the study watershed. Two sample kebeles were taken from 14 kebeles by using randomly sampling technique. Subsequently, sample households were selected from each cluster using simple random sampling technique. Accordingly, a total sample size of 138 household heads were taken from the whole clusters in which 23 households were selected on average from each cluster. Finally, a structured household survey questionnaire was administered to sample households. Interviews were conducted to the household heads to gather data on demographic characteristics, household assets (land holding, livestock, etc.), crop production, household supply (fuel, fodder).

The socio economic data that was obtained from the household survey was entered and analyzed using Statistical Package for the Social Sciences (SPSS, version 16). Hence, to attain the objectives of the study, descriptive analysis methods were done by using the above statistical package software. Moreover, relevant issues collected through focus group discussion, interviews and observation methods were summarized and presented to supplement household survey data analysis (Fig. 2).

Fig. 2
figure 2

Methodological flow of the study variables

Full size image

Result and discussion

Land use and land cover change in the Koga watershed

During the study periods cultivated and settlement areas has been increased by 26.4% between 1973 and 1986. While, forest cover and bush lands, have shown a reverse trend and reduced by 76.1% and 44.5% respectively at the same period. Similarly, the area under wetland and grass land has declined by 87.3% and 10.4%, respectively in the same time (Table 2). This indicates that cultivated and settlement areas were expanded at the expense of other LULC types between 1973 and 1986. At the same time, cultivated and settlement area have increased by 12.7% between 1986 and 1995. While, forest cover and bush land have been declined by 10.1% and 8.9% respectively during the same time. Likewise, grass land has declined by 40.7% in this period. On the other hand, wetland has increased by 33.6% (Table 2). Similar to the previous period, cultivated and settlement has shown a remarkable increment. Thus, it is possible to say that cultivated and settlement land has expanded at the cost of other LULC types, mainly grass and bush lands.

Table 2 Land use/cover change from 1973 to 2011 in the Koga Watershed
Full size table

Between 1995 and 2011, the area of cultivated and settlement area has increased slightly to 206.7 ha as compared to 2595.6 in the previous period. However, forest cover has shown a slight increment (Table 2). This was due to the plantation of eucalyptus tree in the watershed, as response to scarcity of wood for fuel and other demands. In the same way, study by Woldeamlak (2002) confirmed that as there was persistent increase of forest cover which includes trees planted around homesteads (Most of the tree species planted were varieties of Juniper and Eucalyptus) in the Chemoga Watershed led to increment of forest cover. Getachew et al. (2011) also reported a slight increase in eucalyptus tree between 1982 and 2008 in the Debre Mewi watershed near to the study watershed. On the other hand, grass land has diminished drastically (reduced by 54%) being pushed by the expanding cultivated land and the construction of the Koga dam that was observed on 2011 imagery. Similarly, bush and wetland lands have declined by 29% and 31% respectively in this period. Water body has shown a considerable expansion over the same period (Table 2 and Fig. 3). This was due to the construction of the Koga dam in the watershed.

Fig. 3
figure 3

Land cover change from 1973 to 2011

Full size image

Generally, the result indicated that there was a series of LULC change for the last thirty eight years (1973–2011). In this regard, the trend has shown a tendency towards more land being brought under cultivated and settlement area. At the same time, forest (i.e. eucalyptus plantations) became more important in the study watershed and has shown a slight expansion for the past sixteen years. This finding has similar results to Amare (2013a, b) that confirmed the expansion of agriculture and settlement land at the expense of other land uses. The finding was also consistent with the findings of Getachew et al. (2011) and Abate (2011) who reported a significant increase of agricultural land at the expense of shrub and grass land.

Trends of change in land use and land cover

Taking into consideration the overall study period, cultivated and settlement land has shown a remarkable areal increment; by 44%. Forest cover has shown relative increment in coverage, by 10% in the same period. Contrary to this, wetland and grass lands have diminished by 88% and 76% respectively in this period. Similarly, bush land has declined by 64% (Table 3 and Fig. 4).

Table 3 Trend of LULC change from 1973 to 2011and expected in 2021
Full size table
Fig. 4
figure 4

LULC changes from 1973 to 2011

Full size image

Generally, the result has indicated a series of LULC change for the last thirty eight years (1973–2011). In this regard, the trend has shown a tendency towards more land being brought under cultivated and settlement area. At the same time, forest (i.e. eucalyptus plantations) became more important in the study watershed and has shown a slight expansion for the past sixteen years. To the contrary, more and more land covered by bush, grass and wetlands had transformed to cultivated and settlement areas and declined severely. This is consistent with the findings of Getachew et al. (2011) and Abate (2011) who reported a significant increase of agricultural land at the expense of shrub and grass land.

The overall accuracy assessment (Table 4) indicated that the nature of the classification error used in many other research works not far from this result (Abate 2011). The overall accuracy and kappa coefficient is 89.55% and 0.635 respectively (Table 4). According to Ismail and Jusoff (2008) the agreement criteria for Kappa statistics (K) was defined as: poor when K < 0.4, good when 0.4 < K < 0.7 and excellent when K > 0.75. Thus, according to these agreement scales, the LULC change classification for 2011 in this study denotes as better represented. The finding was also similar to Diallo et al. (2009) and Reis (2008) who reported a satisfactory overall accuracy of 85.1% and 87.1% respectively. Hence, the overall accuracy is considered acceptable for this study.

Table 4 Accuracy assessment of LULC classification, 2011
Full size table

Causes of land use and land cover change in the Koga watershed

In the study area there was continuous LULC changes. The changes have been due to different factors which may lead to an irreversible environmental problems and natural resource degradation (Lepers et al. 2005). These include the institutional, socioeconomic and demographic conditions of the surrounding community. The human activities are to be responsible for the major LULC changes in the study area. As in other areas of the country, the number of people in the Koga watershed is growing rapidly from time to time. The population of the Koga watershed was growing at a rate of 2.4% per annum between 1994 and 2007 (Table 5). Moreover, the population of the watershed doubles its size within 29 years. Hence, the population of the Koga watershed will increase twofold within a span of 29 years. As a result of such unparalleled increases in population size, the pressure on existing natural resources is expected to intensify in the study area (Amare 2013a, b). The crude density in the study area was about 235 persons per km2 in 1994 and about 321 persons per km2 in 2007 (Table 5).

Table 5 Population distribution in the Koga watershed by kebele and sex, 2007
Full size table

This finding is in line with that of Braimoh and Vlek (2005) who indicated that population density is important determinant where the dominant change process is extensification. An increase in population density between 1984 and 2007 significantly decreased the probability of observing the grassland-to-cultivated land conversion. This result suggests that an increasing population pressure could rather lead to intensified cultivation of land as a land use change process than an agricultural extensification. This indicates that the study watershed is a densely populated area, which implies that there is very high pressure on the natural resources in the study area.

Land is the most important natural resource for rural households. However like most highland areas of Ethiopia, the farm size of the sample households is on average is less than 1 ha. Based on the household survey, the average land ownership for a household was 0.69 ha which ranged from. 0.5 to around 1. ha (Fig. 5). This shows that there is acute shortage of farm lands in the study area. In this regard, Fikru (2009) and Amare (2013a, b) states that because of shortage of farm size, fallow lands are not common and there is also a shortage of grazing land in the koga watershed. Thus, it is possible to say that land holding size has probably adversely influenced LULC changes in the study area. The major LULC change observed in the study area was the remarkable expansion of crop land at the expense of forest cover, bush land, grassland and wetland. The farmers in the study area depend on rainfed crop production, where production is carried out on a subsistence basis using the traditional farming system. Thus to feed the growing population, the farmers increased crop production through aerial expansion of cultivated land towards forest cover, bush and grass lands. The field observation confirmed that, these days almost all the lands which are suitable for crop production are already cultivated. Overuse of the agricultural land without appropriate conservation measures is facilitating erosion, which is reflected in the sediment laden runoff from the ploughed agricultural fields in the catchment. So, the only chance to increase crop production is to increase productivity on every plot of crop land.

Fig. 5
figure 5

Land size of respondents in Ha,Upper Watershed (UW) and Lower Watershed (LW)

Full size image

As in all other parts of the Ethiopian highlands, oxen are the major source of draught power in the study watershed (Fig. 6a). However, 78 (56.5%) have only one ox per household and 23 (16.7%) have only two oxen per household. The rest 37 (26.8%) of the households do not have any oxen. In the Koga watershed livestock population is exerting much pressure on the environment resulting in degradation of the land that in turn is reflected by the increasing tendency of the farmers to shift to plantation of eucalyptus trees on their crop farms, and deforestation and degradation of the natural woody vegetation cover as confirmed by the key informants. The wood demand as source of energy is also another important factor that causes LULC change in the study area (Fig. 6b). In the study area, the major source of domestic energy for cooking and heating comes from fuel wood. Moreover, the demand of wood products for the construction of house has provoked the destruction of forest in the study area. According to the information acquired through focus group discussion, cutting down trees for house construction is, to some extent, considered to be the causes of LULC change in the study area.

Fig. 6
figure 6

Live stocks consuming maize stack (a) and women collecting fire wood (b)

Full size image

Conclusion

The has identified that the main change observed for the stated time was a substantial areal expansion of cultivated and settlement; increased approximately by 7054.6 ha at the expense of grass land, bush land and natural forests. The major responsible for this change were change in population dynamics and socioeconomic conditions (i.e. land holding size, crop and livestock production and wood demand for source of energy and construction material). Hence, there should be family planning programs to balance the existing natural resources with population, develop and expand appropriate agricultural technologies and rural infrastructure so as to improve agricultural productivity. Since, the major source of domestic energy for cooking and heating in the study area comes from fuel wood, cutting trees for fuel wood partly causes LULC changes in the study area. Therefore, concerned authorities should provide alternative energy sources and energy saving technologies for households in the study area. The appropriate land use plan should be also put in place.