Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed,...

15
10 Applied Research Journal of Geographic Information System Vol 1(1), pp. 10-24, December, 2018 International Institute for Applied Research Article number: se-j-arjgis-2018.0101002 Skies Educational https://skies.education/journal-arjgis/ Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia Asirat Teshome Tolosa Department of Hydraulic and Water Resources Engineering, Debre Tabor University, Ethiopia Correspondence to: [email protected]; [email protected] Abstract: Land use / land cover change (LULC) is influenced by human activities and natural processes. The increase in the population increased the demand for utilizing natural resources, which in turn resulted in land degradation. Biodiversity losses, environmental pollution, and climatic changes are the negative consequences of Land Cover Change (LCC). This study aimed at detecting and analyzing LCC. The study was conducted in the highlands of South Wollo, Yewoll Watershed, Blue Nile Basin, Ethiopia. Three Landsat images (1986, 2000 and 2016) were used to analyze the LCC. Supervised classification using maximum likelihood algorism was used to analyze the LCC. In addition, socio- economic data were collected to support the satellite image analysis. The view of residents was used to develop a historical trend of land cover and to understand the knowledge and the perception of the residents in the watershed. Four land cover types (LCTs) were defined. These are Cropland, Forest, Grassland and Shrubland. The result showed that Cropland and Grassland increased from 41.6% and 15.4% in 1986 to 58.8% and 28.3% in 2016, respectively. However, shrub-land and Forest land declined from 32.3% and 10.6% in 1986 to 5.6% and 7.3% in 2016, respectively. The driver of change is the increase in human and livestock population in the study area. The socioeconomic survey analysis also indicated that forest is converted to cropland and shrub-lands were used for grazing. Generally, the results of the study were verified by field data collected and the judgment of the experts. Keywords: Land use land covers, GIS, Landsat, Remote sensing, supervised classification, INTRODUCTION Natural resources are the basis for economic and social development. Human beings have purposefully managed and converted the landscape to utilize natural resources in order to obtain basic needs such as food, shelter, water, and other products (Goldewijk & Ramankutty, 2004). The human activities in general and agriculture, in particular, modify or change the environment of the given landscape. Human activities such as crop production, animal husbandry, and other related agricultural activities are the dominant causes of a landscape change in human history (Dymond & Johnson, 2002). Land cover change is the central driver and the most dynamic phenomenon that is caused by the interface between the human and ecological system (Manson, 2005). With particular to Ethiopia different studies particularly in the highlands of Ethiopia indicated considerable Land cover change (LCC) is a continuous process due to increase in human and livestock population (Zeleke & Hurni 2001; Wondie et al., 2011; Alemu, 2015; Wondie et al., 2016; Halefom et al., 2018). The major land cover conversions are from forests into other land cover types (LCTs) such as into cultivated land, settlement, and grassland (Fisseha et al., 2011; Wondie et al., 2011; Lunetta et al., 2002; Wondie et al., 2016; Molla, 2015; Halefom et al., 2017). These changes and modification of a landscape can be described using field data or remote sensing approach to support the agriculture-related decision and policy-making process (Ahmad, & Prasad, 2011). The land cover dynamics or modifications are more aggravated by the socioeconomic and biophysical conditions of a given landscape. The ecosystem changes and modification are triggered more by socio-economic conditions than any other factors (Meyer & Turner, Copyright © 2018 by Author(s) and Skies Educational

Transcript of Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed,...

Page 1: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

10

Applied Research Journal of Geographic Information System

Vol 1(1), pp. 10-24, December, 2018 International Institute for Applied Research Article number: se-j-arjgis-2018.0101002 Skies Educational https://skies.education/journal-arjgis/

Evaluating the Dynamics of Land Use / Land Cover Change

Using GIS and Remote Sensing Data in Case of Yewoll

Watershed, Blue Nile Basin, Ethiopia

Asirat Teshome Tolosa

Department of Hydraulic and Water Resources Engineering, Debre Tabor University, Ethiopia

Correspondence to: [email protected]; [email protected]

Abstract: Land use / land cover change (LULC) is influenced by human activities and natural

processes. The increase in the population increased the demand for utilizing natural resources, which in

turn resulted in land degradation. Biodiversity losses, environmental pollution, and climatic changes are

the negative consequences of Land Cover Change (LCC). This study aimed at detecting and analyzing

LCC. The study was conducted in the highlands of South Wollo, Yewoll Watershed, Blue Nile Basin,

Ethiopia. Three Landsat images (1986, 2000 and 2016) were used to analyze the LCC. Supervised

classification using maximum likelihood algorism was used to analyze the LCC. In addition, socio-

economic data were collected to support the satellite image analysis. The view of residents was used to

develop a historical trend of land cover and to understand the knowledge and the perception of the

residents in the watershed. Four land cover types (LCTs) were defined. These are Cropland, Forest,

Grassland and Shrubland. The result showed that Cropland and Grassland increased from 41.6% and

15.4% in 1986 to 58.8% and 28.3% in 2016, respectively. However, shrub-land and Forest land declined

from 32.3% and 10.6% in 1986 to 5.6% and 7.3% in 2016, respectively. The driver of change is the

increase in human and livestock population in the study area. The socioeconomic survey analysis also

indicated that forest is converted to cropland and shrub-lands were used for grazing. Generally, the

results of the study were verified by field data collected and the judgment of the experts.

Keywords: Land use land covers, GIS, Landsat, Remote sensing, supervised classification,

INTRODUCTION

Natural resources are the basis for economic and social development. Human beings have purposefully managed and converted the landscape to utilize natural resources in order to obtain basic needs such as food, shelter, water, and other products (Goldewijk & Ramankutty, 2004). The human activities in general and agriculture, in particular, modify or change the environment of the given landscape. Human activities such as crop production, animal husbandry, and other related agricultural activities are the dominant causes of a landscape change in human history (Dymond & Johnson, 2002). Land cover change is the central driver and the most dynamic phenomenon that is caused by the interface between the human and ecological system (Manson, 2005).

With particular to Ethiopia different studies particularly in the highlands of Ethiopia indicated considerable Land cover change (LCC) is a continuous process due to increase in human and livestock population (Zeleke & Hurni 2001; Wondie et al., 2011; Alemu, 2015; Wondie et al., 2016; Halefom et al., 2018). The major land cover conversions are from forests into other land cover types (LCTs) such as into cultivated land, settlement, and grassland (Fisseha et al., 2011; Wondie et al., 2011; Lunetta et al., 2002; Wondie et al., 2016; Molla, 2015; Halefom et al., 2017). These changes and modification of a landscape can be described using field data or remote sensing approach to support the agriculture-related decision and policy-making process (Ahmad, & Prasad, 2011).

The land cover dynamics or modifications are more aggravated by the socioeconomic and biophysical conditions of a given landscape. The ecosystem changes and modification are triggered more by socio-economic conditions than any other factors (Meyer & Turner, Copyright © 2018 by Author(s) and Skies Educational

Page 2: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Applied Research Journal of Geographic Information System

11

1994; Falcucci & Maiorano, 2007). These dynamics affect the ecological functions, social conditions and economic situation of the respective area (Hurni & Ludi, 2000).

The trend of landscape change and modification are interesting to a comprehensive understanding of the overall changes to make proactive actions. Land cover mapping, modeling, and monitoring of the environment are extracted from remote sensing data. It is because remote sensing technology provides multi-spectral and multi-temporal data biophysical features of the ecosystem (Rogan & Chen, 2004; Rosenqvist et al., 2003). Nowadays, the use of remote sensing data and GIS is increasing over time for mapping land cover, land cover change detection and monitoring of different ecosystems since 1972 (Lunetta et al., 2002; Lunetta & Elvidge, 1999; Sisay et al., 2017).

The existing socioeconomic activity, situation of land cover and environmental changes in the highlands of rural farming communities are the principal subject of this study. Livelihood strategy is linked to land use and is manifested as a land cover as part of the environment (Godfray et al., 2010).

MATERIALS AND METHODS

Description of the Study Area

Yewoll is situated approximately between the Geographic coordinate system of 10°46'29.83" and 10°55'11.21"’N, and between 39°24'35.38" and 39°28'20.60" ’E (Figure 1). The nearest town is Dessie in the South Wollo Zone of Amhara Region. It is about 460 km North-East of Addis Ababa. The study area comprises three districts namely; Werraeilu, Lega ambo, and Dessie zuria.

Topographical location of study area ranges with an elevation difference from 2731m a.s.l just at the outlet of Selig River to 3847 ma.s.l at the top of the Yewoll Mountain (Figure 2). The main River Selig and the tributaries Aba Tisha river and Abale river drain the upper

Figure 1: Location map of the study area

Page 3: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Asirat Teshome Tolosa (2018): Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia

12

parts of the study watershed. The total area of the watershed is about 8203ha. The study watershed is characterized by different landforms which range from high mountains too steep lands to flat slope at the outlet. The watershed is drained to Selig River which is emanated from Spring of Yewoll Mountain.

Yewoll Mountain is the peak for the study area and the landmass of the study area slopes down from the mountain towards the outlet of the watershed. As a result of the topography, soil erosion problems, gully formation and flooding on the bottomlands of the study area are common occurrences.

Figure 2: The topographic condition of the study area (Elevation and slope)

Methods

The methods used in this study were field data collection such as socioeconomic data, and satellite image analysis. Three satellite images (Landsat8 OLI, L7 ETM+, and L5 TM) were acquired from USGS website http://earthexplorer.usgs.gov. These data were used to analyze the status and changes of land cover in the different time span. The field data collection conducted from (Jan – Sept 2016) was used to validate the image classification and understanding different features of the study area.

Data Collection

A DEM supplied from SRTM (Shuttle Radar Topography Mission) data was used to obtain slope and elevation information. The SRTM DEM with a spatial resolution of 30mx30m was downloaded from USGS website, http://earthexplorer.usgs.gov. Field data collection were carried out to obtain ground control points (GCPs) using GPS and to collect biophysical data. The GCPs were used to understand the features of the different LCTs, to support the visual interpretation of the images and to select reference areas (area of interest) and accuracy assessment. All reference areas were localized using GERMIN76 GPS. Representative samples were taken from Cropland, Forest land, Grassland, Shrub-land, and Gully sites for image analysis. In addition, the samples were used for accuracy assessment. The numbers of representative samples collected by GPS during fieldwork for Cropland, Forest, Grassland, Shrub-land, and Gully sites were 38, 34, 39, 11 and 15, respectively. As it was said earlier GPS readings were taken for each sample point with an accuracy varying from 3 to 7.5

Page 4: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Applied Research Journal of Geographic Information System

13

meters. Physical description of the land cover types was carried out to support image analysis. The local development agencies and the farmers residing in the watershed assisted to describe different Land Cover Types (LCTs) and the biophysical condition of the area. A local expert provided detailed information about the land cover types. The land cover categorization scheme was set following the methods of Amsalu et al. (2007) and Wondie et

al. (2016) (Table 1).

Table 1: Land covers categories for change detection Cover class Characterization, features

Cropland Cultivated and fallow land has a characteristic pattern, for example, sharp edges between fields. Dark to grey color in the Landsat image (4, 3, 2 color composition), unless the land lies fallow.

Forest-land This category comprised of planted junipers, eucalyptus trees, Acacia and natural forest or areas covered by trees planted around homesteads and some public institution areas.

Grassland Land allocated as a source of animal feed. Land under permanent pasture and grassland.

Shrubland The areas covered with different species of shrubs, bushes and young tree species with widely varying density from one locality to another

The socioeconomic activities of an area play an important role in the land use land cover types as well as on land degradation. In relation to this, the economic activities of the study area were based on primary productions such as crop farming and animal husbandry. As a result of this, questionnaire was prepared to interview the local people about the historical description of the study area and to understand different features of the study area. The selection of interviewees was random. This was because the economic activities of almost all the farmers were based on agricultural and livestock husbandry. Most of the interviewees were farmers living in and around the watershed. A total of 30 interviewees were selected and interviewed to obtain the required information to find out the relationship of the socioeconomic situation with the land cover situation. These numbers of the selected interviewer are enough to get background information about the study area. The overall content of the questionnaire mainly focused on the knowledge of local people in relation to the trend of LCC and their livelihood strategy. Comparison of the knowledge of the local people and their socioeconomic situation was correlated with the image data using descriptive statistical methods and cross-tabulation. The statistical analysis of the socioeconomic data was carried out using the Statistical Package for Social Science (SPSS) software version 20.

Data analysis

Image preprocessing such as layer stacking and contrast stretching were carried out using ERDAS Imagine 2014. Contrast stretching using a linear method was conducted to enhance the image so as to improve the visual interpretability of the image. The formula used for this purpose is indicated (Lillesand et al., 2000) (Equation 1)

Contrast stretched DN = DN’ =

−−

MINMAX

MINDN*255 ………………………………………………….(1)

Page 5: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Asirat Teshome Tolosa (2018): Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia

14

where:- DN’ is the digital number assigned to a pixel in the new (output image), DN is the original digital number of the pixel in the input image, MIN is the minimum value of the input image digital number, and MAX is the maximum value of the input image digital number. After this normalization, the classification is performed and the output was displayed. The three images used for analysis have a pixel size of 30m.

The method used to classify the Landsat images was supervised classification. The decision rule used in supervised classification was the maximum likelihood classifier algorithm. Area of interests (AOIs) was selected and collected as training areas for classification of the pre-defined LCTs. Pixels were clustered into the categories of Cropland, Forest land, Grass land, and Shrubland. Variable numbers of AOIs were used to classify the images of the three dates (1986, 2000 and 2016). In addition, size and distribution of individual training polygons (AOIs) were variable within and between LCTs depending on the location and availability. The AOIs were distributed in the area of each LCT. The AOIs were selected based on the knowledge of the area obtained from field work and visual interpretation of the images. During selection of AOIs, cropland was differentiated from grassland based on differences in pattern and texture. After classification, feature space was developed to understand the separation. AOIs for Cropland, Forest, Grassland and Shrub-land in 1986 were 36, 47, 26 and 31 and in 2000, it was 46, 40, 22 and 23, while in 2016, it was 48, 40, 30 and 17 respectively.

The accuracy of the image classification was assessed using field data. Based on the field survey, an error matrix was compiled showing field data versus automated classification. The overall accuracy of the classification, producer´s, user´s accuracy and kappa coefficient were calculated from the error matrix.

The results of supervised classification satellite images were evaluated using overall accuracy assessment and kappa coefficient. The kappa ranges between +1 to -1 (Congalton & Green, 2008). The user’s and producer’s accuracy, as well as elements of the error matrix, were calculated to assess error patterns of the respective classification. The results of accuracy statistics provided whether it provides reasonable result or about the quality of land cover classification. The error matrix tables produce many statistical measures of thematic accuracy including overall classification accuracy, the percentage of omission and commission error and the kappa coefficient (Congalton & Green, 2008). The error of omission is the percentage of pixels that should have been put into a given class but were not. The error of commission indicates pixels that were placed in a given class when they actually belong to another. The drawback of confusion matrix and kappa coefficient is that it does not provide a spatial distribution of the errors (Foody, 2002).

The kappa coefficient was indicated to show how strong the agreement using the formula given by Congalton & Green (2008). It is a discrete multivariate technique of use in accuracy assessment. Kappa > 0.80 represent strong agreement and good accuracy. 0.40 - 0.80 is middle, < 0.40 is poor.

)2..(.................... 1

)*(

)*(

1

2

1 1

agreementchance

agreementchanceaccuracyobserved

xxN

xxxN

Kr

i

ii

r

i

r

i

iiii

−−=

−=

∑ ∑

=++

= =++

where:

r = the number of rows in the error matrix,

Xii = the number of observations in row I column I (along with the major diagonal),

Xi+ = is the marginal total of row I (right of the matrix),

Page 6: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Applied Research Journal of Geographic Information System

15

X+i = the marginal total of column I (bottom of the matrix),

N = the total number of observations included in the matrix

Land cover change detection analysis

Post classification comparison was carried out for independent images (thematic maps) which are the most proven technique to deal with change detection. Before LCC detection, consistency of classification and approach was checked to reduce human-induced misclassification. ERDAS modeler was used to detecting the change between two datasets (Figure 3).

(a) (b)

(c)

Figure 3: Land covers change between 1986 and 2016

Page 7: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Asirat Teshome Tolosa (2018): Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia

16

The conditional formula to detect the change is as follows:

Difference of the end year to the beginning = CONDITIONAL {(<test1>) <arg1>, (<test2>) <arg2>,…...}

The result of the CONDITIONAL operation is a change map giving the change category for every pixel. There is 16 change categories date including “no change” corresponding to the transition from each of the 4 LCTs of the one date to each one of the other.

RESULTS AND DISCUSSION

Land cover maps derived from remote sensing always contain errors sourced from different factors. To test whether the classification was reasonable or not, accuracy assessment was carried out to describe the errors quantitatively.

Producer’s and user’s accuracy is shown below (Table 2). The accuracy assessment is not conducted for old images (1986 and 2000 years) because the accuracy assessment of image classification should be based on ground truth and known reference pixels of that time. The accuracy assessment was conducted for the 2016 year image. Landscape often changes rapidly. Therefore, it is best to collect the ground reference as close to the date of remote sensing data acquisition as possible. The number of random samples for this study was 122 from Cropland, Forestland, Grassland, and Shrub-land. This samples evaluated by using the error matrix and the level of accuracy for the true land cover category was presented in Table 3.

The overall accuracy of image classification is 87.7%. The result revealed that shrubland has 100% in producer’s accuracy. Forest is the least accurately classified class compared to the other LCTs. The reason for lower accuracy may be due to the accuracy of GPS data during field collection, land covered by scattered trees were not identified and considered as forestry. The wider spacing between scattered trees on farmland resulted in a spectral

Table 2: Error matrix of field data versus landsat8 OLI 2016 for accuracy Field data

Automated classification

Cropland Forest land

Grassland Shrubland Row total

Cropland 35 3 2 - 40 Forest - 26 1 - 27 Grassland 2 3 35 - 40 Shrubland 1 2 1 11 15 Column total 38 34 39 11 122 Table 3: The accuracy level of each true LC category Land cover classes Producer’s accuracy (%) User’s accuracy (%)

Cropland 92.1 87.5

Forest-land 76.5 96.3

Grassland 89.7 87.5

Shrubland 100 73.3

Page 8: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Applied Research Journal of Geographic Information System

17

signature similar to that of the closest LCTs. Some areas were identified as grassland or cropland. The Kappa coefficient is calculated to be 82.9%. (Equation 2) according to Congalton & Green (2008).This kappa value shows strong agreement.

Major LCTs in the study area in the year 1986 to 2016 include Cropland, Forestland, Grassland and Shrubland. As indicated by the following (Table 4 and Figure 4). For each classified image, the area of each land use/cover class was computed and compared statistically if there are differences between the images. The decreasing of grassland and shrubland has implication on land degradation particularly soil erosion. But the increase of aerial coverage for cropland and grassland was due to an increase of population pressure, demand for cultivated land in the highland and intervention of soil conservation practice by different NGOs and Governmental Organization increase the areal coverage of grassland. This information was obtained during field data collection from farmers living in and surrounding the study area for a long period of time. Generally from the results of image classification Cropland and grassland increase in their areal extent and the rest LCT decrease in their areal extent in the year from 1986 to 2016.

Table 4: Area coverage of each LCT from 1986 - 2016 LCT

1986 2000 2016 Area (ha) Area (%) Area (ha) Area (%) Area (ha) Area (%)

Cropland 3414.78 41.6 5196.6 63.3 4825.26 58.8 Forestland 873 10.6 1348.56 16.4 596.25 7.27 Grassland 1261.62 15.4 1241.19 15.13 2323.98 28.33 Shrubland 2653.65 32.3 416.7 5.08 457.56 5.58 Total 8203.05 100 8203.05 100 8203.05 100

Figure 4: Areal coverage of different categories (1986, 2000 & 2016 years)

0

1000

2000

3000

4000

5000

6000

1986 2000 2016

Are

al c

over

age

in (

ha)

Years

Crop land

Forest land

Grass land

Shrubland

Page 9: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Asirat Teshome Tolosa (2018): Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia

18

In the year 1986 the Forestland, Grassland, Shrubland, and Cropland are arranged in order of increase in areal coverage (Fig 4 and Table 4). However, in 2016 the land cover changed due to the shift of landscape.

Between 1986 and 2016, forest declined from 10.6% to 7.27% (Table 4). A contributing factor to this decline may be the increase of population. The increase of population in turn increased food demand which resulted in the decline of the forest. The result computed in the above tables showed the areal extent of each land cover types for 30 years. Therefore cropland and grassland increase their areal extent but forestland and shrub-land decrease their areal extent in the span of 30 years. The results of the current study show that cropland expansion is a similar trend with the other regions, as illustrated by some studies conducted in the highlands of Amhara (Tekle & Hedlund, 2000; Zeleke & Hurni,

2001; Wondie et al., 2016). Most of the studies in the Ethiopian highlands showed a decline of the forest, which is a similar situation due to the increase in human and livestock population. The increase of human and livestock population increased the demand for food. This urged the conversion of shrub and forestland into crop and grassland.

Land use land cover changes

Cropland and Forestland in the study area during observation from 1986 to 2000 have increased by 1781.82ha and 475.56 ha respectively. But Grassland and Shrubland have decreased by 20.43ha and 2237ha respectively. As indicated by Table 4 and Figure 5.

The impact of decreased in grassland and shrubland in the study area between 1986 and 2000 have resulted in land degradation particularly soil erosion. Clearing of shrubs exposes soils to compaction by farm machinery and by animal hoofs during grazing; consequently, the topsoil compacts and reduces infiltration of water thereby increasing surface runoff. Generally, change of one LCT to another has been shown by the following Figures 6 (a) to (c)

All the four LCTs generated a total of 16 possible combinations or transformations including the four “no change”. From the total area of the watershed, 46.5% of the area remained unchanged and 53.5% of the area changed from one land cover to another category within 30 years.

The rate of change of land cover

From the analysis of Landsat image with verification of field data, the rate of change of land cover has been presented (Table 5). The result revealed that there were gains and losses of LCT in 30 years.

It is known that agriculture and grazing are the main livelihood strategies of the population living in and around the Yewoll watershed. In 1986, shrubland was the dominant cover which is 2653.65 ha. Croplands followed by grassland are the top two dominant land cover classes of the watershed in 2016 (Fig 5). Figure 7, below shows the net gain and loss of land cover in percent. The overall area of cropland and grassland is increased from 1986 to 2016. The increase may be due to the increase in demand for food and other animal related products.

As indicated by the figure above all the land cover classes have gains and losses. The cropland is increased by 17.2%, whereas shrub land declined by 26.72% in 30 years period (1986 – 2016).

Page 10: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Applied Research Journal of Geographic Information System

19

Socioeconomic data analysis

The economy of farmers is based on crop and livestock. Average land holding size of farmers interviewed varied from 0.25 to 2ha per household. Age has a factor for landholding size. Older farmers have larger land size than younger ones (Table 6).

(a) (b)

(c)

Figure 5: Land cover map of the (a) 1986, (b) 2000 and (c) 2016 years

Page 11: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Asirat Teshome Tolosa (2018): Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia

20

(a) (b)

(c)

Figure 6: Land cover changes between (a) 1986 and 2000 (b) 2000 and 2016 and (c) 1986 and 2016

Table 5: Summary of land use land cover change distribution over years LCT 1986 2000 2016 The rate of change (%)

Area(ha) Area(%) Area(ha) Area(%) Area(ha) Area(%) 1986-2000

2000-2016

1986-2016

Cropland 3414.78 41.6 5196.6 63.35 4825.26 58.80 21.75 -4.55 17.20 Forestland 873 10.6 1348.56 16.44 596.25 7.27 5.84 -9.17 -3.33 Grassland 1261.62 15.4 1241.19 15.13 2323.98 28.33 -0.27 13.20 12.93 Shrubland 2653.65 32.3 416.7 5.08 457.56 5.58 -27.22 0.50 -26.72

Legend

Watershed

no data

from cropland to cropland

from cropland to forestland

from cropland to grassland

from cropland to shrubland

from forestland to cropland

from forestland to forestland

from forestland to grassland

from forestland to shrubland

from grassland to cropland

from grassland to forestland

from grassland to grassland

from grassland to shrubland

from shrubland to cropland

from shrubland to forestland

from shrubland to grassland

from shrubland to shrubland

Page 12: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Applied Research Journal of Geographic Information System

21

Based on the farmers’ knowledge lived in and around the study area, the land use changes dominantly from forest to cropland. This judgment was compared and validated with the result of remote sensing data.

The farmers’ description of the change of one LCT to another is indicated in Table 7. According to the interviewed people’s response, 55% of the respondents observed the main change is from forest to cropland. The driver of change was described by the farmers’ residing in the watershed. The major drivers of change are an increase of human population. Some of the farmers also described animal population as the cause for land cover change. It is because farmers’ livelihood is partly based on animal husbandry.

Figure 7: The net gain or loss of individual categories from 1986 to 2016

Table 6: Average lands holding in each age category Age category of farmers (years) Land holding size for agriculture (ha) ≤ 30 31- 40 41- 65 Greater than 65 0 to 0.25 - - - - 0.3 to 0.5 1 - - - 0.6 to 1 2 1 2 1 Greater than 1 3 7 9 4

Table 7: The main change of LCC as perceived by interviewees (%) Respondents (%) in age

categories Total of respondents (%)

Direction of land conversion <30 31- 40 41- 65 >65

Forest to cropland 13 20 12 10 55 Forest to grassland 0 0 0 0 0 Cropland to Grassland 0 0 0 0 0 Grassland to Cropland 18 12 8 7 45 No observation 0 0 0 0 0

Total of each age category of

(%) 31 32 20 17 100

17.2

-3.4

13

-26.8-30

-25

-20

-15

-10

-5

0

5

10

15

20

1

Ind

ivid

ua

l ca

teg

ory

ch

an

ges

(%)

cropland

forestland

grassland

shrubland

Page 13: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Asirat Teshome Tolosa (2018): Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia

22

CONCLUSIONS AND RECOMMENDATIONS

Conclusions

By analyzing remote sensing data from a period of 30 years (1986 - 2016), this study demonstrates the magnitude of change of LULC. Multi-temporal remote sensing data analysis shows the interrelationship between human demand, the dynamics of changes in land cover in the watershed ecosystems, and the areal extent of soil erosion in the study area. The landscape is modified to provide various types of ecosystem services, such as food, feed, and wood for energy. This shows how land cover change is influenced by the demand of the farmers.

The four major land use/land cover units classified on the basis of 2016 image classification in the study area were forced to either an increase or a decrease in areal extent due to the pressure exerted by the high population growth of the study area and its surroundings. The quantitative evidence of land use / land cover change reflected that cropland and grassland showed 17.2% and 13 % increase in areal coverage between 1986 and 2016, respectively. The increase may be due to the increase in demand for food and other animal related products. On the other hand, forestland and shrub land showed 3.4% and 26.8% decrease in their areal extent respectively. This was due to a transformation of forestland and shrubland into other land use/land cover types. As revealed from the socio-economic survey and confirmed by GIS and RS analyses of satellite images, the land use changed dominantly from forestland to cropland. The major drivers of change in land cover of the study area are the increase in population and livestock production.

Quantitative description of remote sensing data shows that smallholding farmers in Yewoll Watershed have gradually changed their land from shrub and forest to cereal-based farming. The output of this study can be used as a basis for sustainable development of the study area. The information obtained from the classification of Landsat imagery is crucial for decision making. It quantitatively describes the state of the landscape and the base of the economic activity of the watershed, which is necessary for long-term planning, and for utilizing and managing land resources. The results of this study can provide information useful for designing land use planning to regulate the effect of a land cover change.

Recommendations

Change in land use contributes a lot to soil erosion so, monitoring of such changes using remote sensing, which gives accurate and timely information, is a sine qua non. Remote sensing data analysis provides information on the change trend of land resources, thereby facilitating monitoring of the environment and to support land use planning. The outputs of model results have been used as guideline for protecting the land from severe degradation because the real world is more complex to understand and to handle at a time. To facilitate effective management and market planning, knowledge of the rate of change of the farming system's evolution - specifically the forest resources is required.

REFERENCES

Ahmad, P. & Prasad, M., 2011. Environmental adaptations and stress tolerance of plants in the era of climate change.

Alemu, B., 2015. The Effect of Land Use Land Cover Change on Land Degradation in the Highlands of Ethiopia. Journal of Environment and Earth Science, 5(1), pp.1 - 13.

Amsalu, A., Stroosnijder, L. & Graaff, J. De, 2007. Long-term dynamics in land resource use

Page 14: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Applied Research Journal of Geographic Information System

23

and the driving forces in the Beressa watershed, highlands of Ethiopia. Journal of Environmental Management, 83, pp.448 - 459.

Congalton, R. & Green, K., 2008. Assessing the accuracy of remotely sensed data: principles and practices. Available at: https://books.google.nl/books? [Accessed April 12, 2017].

Dymond, C. & Johnson, E., 2002. Mapping vegetation spatial patterns from modeled water, temperature, and solar radiation gradients. ISPRS Journal of Photogrammetry and Remote, 57, pp.69 - 85.

Falcucci, A. & Maiorano, Æ.L., 2007. Changes in land-use / land-cover patterns in Italy and their implications for biodiversity conservation. Landscape Ecol, 22, pp.617 - 631.

Fisseha, G. et al., 2011. Analysis of land use/land cover changes in the Debre-Mewi watershed at the upper catchment of the Blue Nile Basin, Northwest Ethiopia. , 1(6), pp.184 - 198.

Foody, G., 2002. Status of land cover classification accuracy assessment. Remote sensing of environment, 80, pp.185 - 201.

Godfray, H. et al., 2010. Food security: the challenge of feeding 9 billion people. Goldewijk, K.K. & Ramankutty, N., 2004. Land cover change over the last three centuries

due to human activities : The availability of new global data sets. , 61, pp.335 - 344. Halefom, A., Sisay, E., Khare, D., Singh, L. and Worku, T. 2017. Hydrological Modeling of

Urban Catchment Using Semi-Distributed Model. Modeling Earth Systems and Environment, 3, 683 - 692.

Halefom, A., Teshome, A., Sisay, E. and Ahmad, I. 2018. Dynamics of Land Use and Land Cover Change Using Remote Sensing and GIS: A Case Study of Debre Tabor Town, South Gondar, Ethiopia. Journal of Geographic Information System, 10, 165 - 174. https://doi.org/10.4236/jgis.2018.102008

Hurni, H. & Ludi, E., 2000. Reconciling conservation with sustainable development: a participatory study inside and around the Simen Mountains National Park, Ethiopia.

Lillesand, T., Kiefer, R. & Chipman, J., 2000. Remote sensing and image analysis. John Wiley and Sons, New York. Available at: https://scholar.google.nl/scholar.

Lunetta, R. & Elvidge, C., 1999. Remote sensing change detection. Lunetta, R., Ediriwickrema, J. & Johnson, D., 2002. Impacts of vegetation dynamics on the

identification of land-cover change in a biologically complex community in North Carolina, USA. Remote Sensing of, 82, pp.258 - 270.

Manson, S.M., 2005. Agent-based modeling and genetic programming for ´ n modeling land change in the Southern Yucatan Peninsular Region of Mexico. Agriculture, Ecosystems and Environment, 111, pp.47–62.

Meyer, W. & Turner, I.B., 1994. Changes in land use and land cover: a global perspective. Available at: http://books.google.com/books.

Molla, M., 2015. Land Use/Land Cover Dynamics in the Central Rift Valley Region of Ethiopia: Case of Arsi Negele District. African Journal of Agricultural Research, 10, pp.434 - 449.

Rogan, J. & Chen, D., 2004. Remote sensing technology for mapping and monitoring land-cover and land-use change. Progress in planning.

Rosenqvist, Å. et al., 2003. A review of remote sensing technology in support of the Kyoto Protocol. Environmental Science.

Sisay, E., Halefom, A., Khare, D., Singh, L. and Worku, T. 2017. Hydrological Modelling of Ungauged Urban Watershed Using SWAT Model. Modeling Earth Systems and Environment, 3, 693-702. https://doi.org/10.1007/s40808-017-0328-6

Tekle, K. & Hedlund, L., 2000. Land cover changes between 1958 and 1986 in Kalu District, southern Wello, Ethiopia. Mountain research and development, 20, pp.42 - 51.

Wondie, M. et al., 2011. Spatial and temporal land cover changes in the semen mountains

Page 15: Evaluating the Dynamics of Land Use / Land Cover Change ... · Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia 12 parts of the study watershed. The total area of the watershed

Asirat Teshome Tolosa (2018): Evaluating the Dynamics of Land Use / Land Cover Change Using GIS and Remote Sensing Data in Case of Yewoll Watershed, Blue Nile Basin, Ethiopia

24

national park, a world heritage Site in northwestern Ethiopia. Remote Sensing, 3(4), pp.752 - 766.

Wondie, M. et al., 2016. Modeling the dynamics of landscape transformations and population growth in the highlands of Ethiopia using remote-sensing data. International Journal of Remote Sensing, 37(23), pp.5647 - 5667.

Zeleke, G. & Hurni, H., 2001. Implications of Land Use and Land Cover Dynamics for Mountain Resource Degradation in the Northwestern Ethiopian Highlands Implications of Land Use and Land Cover Dynamics for Mountain Resource Degradation in the Northwestern Ethiopian Highlands. , 21(2), pp.184 - 191.