The changes in land use and land cover are the most considerable factors for the decision makers, which can be precisely evaluated by using Geographic Information System (GIS) and Satellite Remote Sensing (SRS) technique. The present study was carried out in the Quetta Basin Watershed for evaluating the land use change of two decades during 1999-2009 and 2009-2018, using the Landsat imageries. The main land use types of study area include water body, agriculture, vegetation, barren and built-up land area. All these land use types were mapped for each time period. The technique of hybrid classification was used for the classification of the imageries by using ISODATA and Maximum Likelihood Algorithm (MLA) in ERDAS Imagine software 2011 v, while change detection maps of different land use land cover classes was generated by using the ArcGIS 10.2 v. The finding of study indicated that widespread and significant changes has occurred in study area during two decades in which remarkable increase occur in built up and vegetation land area, on the other hand reduction was observed in water body, barren and agriculture land area from year 1999 to 2018. During the whole study period, major land shifting was observed in agriculture to vegetation and vegetation to barren. The transformation in land cover use may pose severe risk to watershed resources. Therefore, proper management, effective conservation plans and strategies are necessary to designed for the protection of this natural resource.
Research Article
Sequential Analysis of Quetta Watershed Using Machine Learning Tool
https://doi.org/10.21203/rs.3.rs-3178851/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Classification
Change analysis
Remote sensing
Machine learning tool
A watershed can be defined as topographically delineated vicinity that tends to drain groundwater, rainwater and melting snow water into either a stream or river system [1]. Every part of land is an important element of the watershed through which water (along with nutrients, debris, chemicals etc) drains to its decrease point. Consequently, a typical watershed consists of multiple forms of land in terms of use and land cover (LULC), which can have certain impact on watersheds’ hydrological characteristics [2]. Watersheds can be big or small depending on the size of stream; the larger the stream the larger will be the watershed within areas of similar climate, geology and topography.
In Pakistan major part of the northern mountains and a part of the Pothwar plateau represent the primary watersheds that drains into the tributaries of the Indus rivers system whereas, the watersheds of Mangla dam and Tarbela dam and many other small dams will also fall in to this place [3]. During past centuries, these watersheds have sadly suffered excessive and extensive cutting of woody plants as well as overgrazing which create changes in land cover areas. According to Ahmad, 2001, the sediment in watershed is mainly deposited into channels, reservoirs and streams which require high investment for remedial measures to maintain up their useful lives [4].
The extent of LULC of a particular land area represents the socio-economic and natural elements of that place; along with their utilization in phases of time and space [5]. Remote sensing (RS) data had been one of the most important data resources for studying temporal and spatial changes. Especially, with the help of the techniques of geographic information system and remote sensing, it is possible to analyze and classify the changes occurring in the pattern of land cover at some stage over a long period of time; which is helpful for us to recognize the modifications inside the region of interest [6].
Collecting information about the land cover change is important to better understand the interaction and relationship between humans and natural environment. In order to evaluate all the changes occurred along the Earth’s surface caused by anthropogenic activities, land cover alteration can be found on a temporal scale. Spatio-temporal analysis presents evidence about the resources, development, environment and the socioeconomic factors. Anthropogenic processes are the major factors in impacting the land use and land cover change which significantly changing the surface of earth [7]. GIS technology coupled with satellite remote sensing (SRS) provided information for developing such database information that can be used to monitor watershed areas in order to protect them from impacts of anthropogenic activities. Through the invention of GIS and SRS technique, the mapping of LULC change has given a proper means of improving the selection of the areas that are designated as water bodies, agriculture and built up area of a specific region [8]. The primary factors which can be contributing to the extended use of satellite imageries for LULC change detection includes accurate geo-referencing system, precise spectral bandwidth as well as high or excessive temporal resolution. Upon gaining the understanding about the algorithms and deciding an appropriate method, a digital number (DN) value is allotted to one pixel by differentiating the other pixel with its own (DN) value. In reference to numerous supported strategies for accumulating data regarding watershed health as well as analyzing and mapping LULC change, the GIS and SRS techniques are usually applied and considered as a powerful tool which plays an important role for resolving the challenges related to various LULC change [9].
Information retrieved from multi temporal remote sensing for the purpose of change detection is applied for quantitatively examining the occurrence of historical consequences, which facilitates in figuring out the changes associated with LULC change properties with reference to the multi temporal datasets [3, 10–11]. The LULC change degrades vital ecosystem services of watersheds [12]. By attaining a better understanding of these impacts, the estimation, forecasting and modeling of watershed can be conducted at regional, local and international levels [13]. Many researches have been carried out on change analysis of watershed through spatio-temporal technique by using various change detection methods which includes supervised and unsupervised classification, hybrid and fuzzy classification etc. All these techniques are commonly applied for studying various LULC change patterns [14–16]. The aim of present study was to analyze the spatio-temporal changes occurred in last two decade (1999–2009, 2009–2018) in Quetta watershed using the Landsat imageries.
Quetta valley is the capital city of Balochistan province in Pakistan lie between 30.1798 0N and 66.9750 0E. It is basically a cup shaped valley, at an average elevation of 1,680 meters above sea level. It is basically a cup shaped valley. The whole city is surrounded by beautiful hills such as Chiltan, Murdar, Takatoo and Zarghoon hills. The over-all watershed area of Quetta valley is about 1757 km2 out of which 792 km2 area is mainly covered by alluvium [17]. Quetta valley falls in the Southern basin watershed which is also known as “Quetta Pishin sub basin”, and is a part of North East Pishin Lora Basin (NEPL). The Quetta valley is composed of three distinctive landforms. The first is the valley floor and is located in the center of the valley. The second is the piedmont region lying between the ground and the mountain. Moreover, the third is an elevated mountain zone with steep slopes including mountain ranges of Takatu, Chiltan, Daghari, Zarghoon and Murdar. Quetta valley has semi-arid climate, with great temperature variations endures in the valley. The city is blessed with all four seasons.
The first and most important step for examining the LULC change was to acquire the imageries of two decades 1999–2009, 2009–2018 with zero cloud cover having 30-meter and 15-meter resolution respectively were obtained by USGS Glovis. The imageries were preprocessed using ERDAS imagine software 2011 v. The technique of hybrid classification was used for the classification of the imageries by using ISODATA and Maximum Likelihood Algorithm (MLA). Total five classes were delineated which include water body, vegetation, agriculture, built-up and barren area. The accuracy assessment was accomplished using ancillary data. Total 200 random sample points were selected and monitored through Google Earth. Land shifting pattern was also monitored and major changes in land cover classes were analyzed. Figure 1 represents the geographical location map of study area.
The results of present research revealed that the classified imageries of year 1999, 2009 and 2018 shows an overall calculated accuracy of 92%, 91% and 95.5% respectively, while the results of kappa statistic value of the respective years were found 0.89%, 0.85% and 0.92% are shown in Table 1.
|
Reference years |
Classified imageries |
Overall Classification Accuracy |
Overall Kappa Statistics |
|---|---|---|---|
|
1999 |
Landsat 4–5 TM |
92.00% |
0.87 |
|
2009 |
Landsat 4–5 TM |
91.00% |
0.85 |
|
2018 |
Landsat 8 OLI/TRS |
95.50% |
0.92 |
Landsat image 4–5 TM was chosen for years 1999 and 2009; and Landsat 8 OLI image was used for year 2018 and classified it into five classes. The classified image of year 1999 depicted the area 0.042%, 87.97%, 4.90%, 5.97% and 1.10% covered by water, barren area, vegetation, agriculture and built-up (Fig. 2). In year 2009 water body covered an area of 0.03%, Vegetation and Agriculture covered 20.45% and 3.74%, while Barren and Built-up area covered 72.29%, and 3.46% respectively seen in Fig. 3. Whereas the classified image result of year 2018 showed that 3.51%, 3.79%, 20.57%, 72.09% and 0.03% area is covered by Built-up, Agriculture, Vegetation, Barren and Water respectively (Fig. 4).
According to the results presented in Table 2, in year 1999 barren land was the major class of study area which covered 188592.23 ha land followed by agriculture area which covers 12799.70 ha land area, third major class was vegetation that shares 10516.64 ha of land area. Built up was the fourth major class which covers 2375.26 ha area and remaining area was occupied by water body about 90.95 ha of land area shown in Fig. 2.
|
LULC classes |
Years |
|||||
|---|---|---|---|---|---|---|
|
1999 |
%age |
2009 |
%age |
2018 |
%age |
|
|
Water |
90.95 |
0.042 |
70.43 |
0.03 |
70.11 |
0.03 |
|
Agriculture |
12799.70 |
5.97 |
8035.13 |
3.74 |
8126.95 |
3.79 |
|
Barren |
188592.23 |
87.97 |
154988.39 |
72.29 |
154542.80 |
72.09 |
|
Vegetation |
10516.64 |
4.90 |
43848.07 |
20.45 |
44102.84 |
20.57 |
|
Built up |
2375.26 |
1.10 |
7432.76 |
3.46 |
7532.08 |
3.51 |
|
Total |
214374.78 |
214374.78 |
214374.78 |
LULC classes area of year 2009 and 2018 (Fig. 3 and Fig. 4) showed that first major class in both land use map was barren area which covered 154988.39 ha in 2009 and 154542.80 ha in the year 2018. The second major class was vegetation i.e. 43848.07 ha and 44102.84 ha in year 2009 and 2018 respectively. Agriculture land area was third major class which covered 8035.13 ha, 8126.95 ha respectively, in year 2009 and 2018. Whereas the fourth major class was built up in both land use maps of year 2009 and 2018; which covered 7432.76 ha and 7532.08 ha of land area. Comparatively, in both years (2009 and 2018) the area covered by water body is less than the area covered by other land cover classes. In year 2009 water body covered an area of 70.43 ha and about 70.11 ha area was covered in year 2018.
The LULC change of an area is always the result of aggregate or collective interaction among the demographic, socio-economic and biophysical factors. The change information and the area of transformation were analyzed for LULC classes in two study periods. Furthermore, by using cross tabulation, the classified imageries of consecutive years were compared. The cross tabulation matrices (Tables 3 and 4) represent the nature of change/shift in the land cover classes, which are required for determining the change in terms of quantitative and qualitative aspects.
Area shifting from year 1999 to 2009
The shifting of area can be observed in Fig. 5 which shows that all the land cover classes were shifted into another class. The change from year 1999 to 2009 indicated that the major portion of agriculture land was shifted into vegetation class (7031.05 ha) during year 1999–2009. Whereas, 3206.08 ha built up land area was also shifted into vegetation class. Similarly, the major portion of Barren land about 32198.03 ha area was transformed into vegetation class and about 4930.19 ha of vegetation land area was converted into agriculture class (Table 3). While major portion of water about 20.78 ha area was transformed into barren land area.
Table 3. Cross-tabulation from year 1999 to 2009 (unit ha)
Area shifting from year 2009 to 2018
Major land shifting was recorded in the class of vegetation, agriculture, barren and built up land area during the year 2009 to 2018 according to the statistical calculation (Table 4). Major land area of vegetation cover was 7525.77 ha which was shifted into barren area (Fig. 6). The class of agriculture was replaced by barren (285.23 ha) class. During the same year, barren class was shifted into vegetation which covered 1568.911 ha area. Similarly, during the same decade area of land, covered by water class was exceptionally small. Only 0.18 ha area was shifted into barren class while 68.07 ha area was remain as water during 2009–2018.
Table 4. Cross-tabulation from year 2009 to 20018 (unit ha)
The changes in the total area of Quetta basin watershed from year 1999, 2009 and 2018 can be seen in Fig. 5 and Fig. 6. Quetta basin covers an area of 1300km2. This basin lies in two districts one is Quetta and other is Mastung with an average elevation of 1680 m. The basin is bordered by disproportionate mountain range. Due to various environmental and socioeconomic activities, all the changes in the land cover areas are of different magnitude and nature.
During the concern years of study periods, major changes in land cover areas were observed. The LULC change of two consecutive decades (1999–2009 and 2009–2018) showed that water body, agriculture and barren covered area shows declining trend while increasing trend was observed in area covered by vegetation class and built up land class. The very prominent impacts of land use land cover change in study area are linked to the expansion due to increasing population and migration of people towards towns; which has resulted in the acquisition of most suitable agriculture land area for urban development. The decline of agriculture land cover area is mainly the result of very low precipitation during most years of study period.
During the year 1999–2009 water body is consider as more vulnerable class because high percentage 20.78% of water cover class is being transformed to barren class, as in 1999 the water body covers an area of about 90.95 ha and in 2009 it is decreased up to 70.43 ha and has shifted in to vegetation, barren and built up land area. While in year 2018 it reduced about 70.11 ha in which 72.09% has shifted into other class mainly in barren land area because of rapid urbanization and the large-scale migration of refugees from Afghanistan, the population of study area is extended continuously. Since the drought condition prevailing within last few years in surrounding area of Quetta number of people was pressured to migrate to Quetta city to earn their better livelihood causing drastic increase in population. According to Pakistan Meteorological Department Balochistan the reason behind the decline of water cover class is due to the drought condition prevailing within last few years in surrounding area of Quetta. The valley has faced several droughts since 1999 till 2009. Due to enormous gaps in regular rainfall patterns and an increase in temperature, the extended periods of the drought in the recent past have also played role in causing unbalancing of the natural recharge, which made the situation even worse. During drought the reservoirs of Spin Karez located in Northwest and Hana-Urak to the eastern side had become completely dry, which caused huge decline of water in Quetta valley aquifer, and created an exceptional burden of human population at the aquifer. This all contributed a lot in creating the issues of water supply and demand within the valley [17]. Precipitation is the predominant source of natural recharge to the aquifer (falling over Quetta sub basin’s catchment areas). Due to the existing drought condition in different part of Balochistan, well below average precipitation that is 60–150 mm receives high land areas for a continuous period of 4 to 5 years [18].
As shown in Table 3, about 20.78 ha and 1.17 ha from water class was transformed in to barren, and vegetation land area during the year 1999 to 2009, and about 1.33 ha of barren land in year 1999 was transformed into water body class. While major portion of agriculture class was shifted into vegetation, barren and built up. Since 2009 to 2018 major portion of vegetation class that is 7525 ha was shifted into barren area.
During same year, negative change has been observed in vegetation cover class as it has been shifted to other feature classes except water body class. Quetta valley has sparse vegetation, throughout the concerned years of study period vegetation area increased drastically over time, The area under vegetation increases from 4.90% in the year 1999 up to 20.45% and 20.57% in respective year 2009 to 2018, because wild plants are growing on mountains and on lower elevated areas because wild plants are growing on mountains and on lower elevated areas, The change detection matrix indicated that major portion of vegetation class was shifted in to agriculture, barren and built up land in both decades (1999–2009 and 2009–2018) shown in Fig. 3 and Fig. 4. There are several reasons behind the expansion and conversion of vegetation class in to other class which includes that vegetation cover is directly proportional to availability of water resources and indirectly proportional to the population growth.
For many economies, especially for developing countries agriculture is an important driving force for economic growth. The agriculture class covered 12799.70 ha area in the year 1999 and 8035.13 ha area covered in the year 2009 while 8126.95 ha area covered in the year 2018. From year 1999 to 2009 major portion of agriculture class was replaced by vegetation class i.e. 7931.05 ha area and about 2701.37 ha area was shifted in to barren class, while 1042.48 ha area converted in to build up land area (Table 3 In the year 2009 to 2018 major portion of agriculture land 285.23 ha area was shifted into barren class and small portion about 0.25 ha area was converted into vegetation cover class (Table 4). The possibilities of growth and development in the sphere of agriculture are almost negligible; and the reason behind this is extended demographic thrust (resulting is construction of increased number of housing societies, buildings, business and market centers etc). Nevertheless, depletion of groundwater and overall water scarcity are major constraints or hurdles in agricultural developments. Other factor which contributes in the decline of agriculture land area was the climatic condition and transformation of agriculture land to other land cover classes. As in year 1999 to 2009 major portion of agriculture class was occupied by vegetation, barren and built up class, while in decade 2009 to 2018 vegetation and barren area class replace major portion of agriculture land area
The area under barren land showed continuously decreasing trend. About 87.97% area was covered by barren land in the year 1999. During the year 2009 and 2018 barren land area was decreased up to 72.29% and 72.09%; as this change was brought by transformation of 1568.911 ha land area into vegetation and 2790.59 ha area in to build up. The massive decrease in barren land area was due to urban expansion all around and construction of new buildings and infrastructure.
The result of present study of spatiotemporal change analysis revealed that all land cover classes showed both increasing and decreasing trend throughout the whole study period. The main focus of this study is basically the water cover land area which is significantly decreases from 90.95 ha in 1999 to 70.43 ha to 70.11 ha land area in 2009 to 2018 and were shifted in to vegetation, barren and built up land cover areas. Vegetation and built up land area increase during the year of 2009–2018 while decreasing trend were observed in agriculture and barren land area. Generally, the area of LULC classes showed that continuously increasing trend was observed in vegetation and built up class, while decreasing trend was observed in water, agriculture and barren land area during the study time period. However, this haphazard situation of LULC change not only affects the socioeconomic condition of study area but also affects the entire valley as well. Therefore, there is a need for strong collaboration among the researcher and government authorities at local, regional and national level in order to control future threats regarding land use change and the degradation of useful natural resource in the watershed area.
- Funding: “There is no external funding for this work.”
- Competing Interests: “The authors have no relevant financial or non-financial interests to disclose.”
- Author Contributions: “Conceptualization, method designing and supervision was conducted by Rabail Urooj, formal analysis by Kinza Imtiaz, data review and edit by Anila Bahadur, Nafisa Kakar and Laraib Ehtasham. All authors have read and agreed to the published version of the manuscript.”
- Ethics Approval: “As this study does not involve any human or animal subject, therefore there cannot be any ethic approval statement.”
- Consent to participate: “There are no human subjects in this research.”
- Consent to publish: “There are no human subjects in this research.”
- Data Availability Statement: Data and associated metadata are available from corresponding author
- Tekle, K., & Hedlund, L. (2000). Land cover changes between 1958 and 1986 in Kalu District, southern Wello, Ethiopia. Mountain Research and Development, 20(1), 42–51.
- Haque, S. M. S. (2013). Watershed Management in Bangladesh. Degradation of Upland Watershed in Bangladesh Project. Bangladesh: Institute of Forestry and Environmental Sciences, University of Chittagong (IFESCU).
- Ahmad, F. (2012). Detection of change in vegetation cover using multi-spectral and multi-temporal information for District Sargodha, Pakistan. Sociedade & Natureza, 24(3), 557–571.
- Ahmad, S. (2001). Watershed management in Pakistan: achievements and issues. National seminar on water resources (pp. 1–17). Pakistan: Lahore.
- Rawat, J. S., & Kumar, M. (2015). Monitoring land use/cover change using remote sensing and GIS techniques: A case study of Hawalbagh block, district Almora, Uttarakhand, India. The Egyptian Journal of Remote Sensing and Space Science, 18(1), 77–84.
- Fichera, C. R., Modica, G., & Pollino, M. (2012). Land cover classification and change-detection analysis using multi-temporal remote sensed imagery and landscape metrics. European Journal of Remote Sensing, 45(1), 1–18.
- Dinka, M. O., & Chaka, D. D. (2019). Analysis of land use/land cover change in Adei watershed, Central Highlands of Ethiopia. Journal of Water and Land Development, 41(1), 146–153.
- Boori, M. S., Vozenílek, V., & Choudhary, K. (2015). Land use/cover disturbance due to tourism in Jeseníky Mountain, Czech Republic: A remote sensing and GIS based approach. The Egyptian Journal of Remote Sensing and Space Science, 18(1), 17–26.
- Zhang, S. Q., Zhang, S. K., & Zhang, J. Y. (2000). A study on wetland classification model of remote sensing in the Sangjiang Plain. Chinese Geographical Science, 10(1), 68–73.
- Seif, A., & Mokarram, M. (2012). Change detection of Gil Playa in the Northeast of Fars Province. Iran Am J Sci Res, 86, 122–130.
- Swallow, B. M., Johnson, N. L., & Meinzen-Dick, R. S. (2002). Working with people for watershed management. Water Policy, 6(3), 449–455.
- Adhikari, S., Shrestha, S. M., Singh, R., Upadhaya, S., & Stapp, J. R. (2016). Land Use Change at Sub-Watershed Level. Hydrology Current Research, 7(256), 1–5.
- Chen, L., Wang, J., Fu, B., & Qiu, Y. (2001). Land-use change in a small catchment of northern Loess Plateau, China. Agriculture. Ecosystems & Environment, 86(2), 163–172.
- Rundquist, D. C., Narumalani, S., & Narayanan, R. M. (2001). A review of wetlands remote sensing and defining new considerations. International Journal of Remote Sensing, 20(3), 207–226.
- Lu, D., Mausel, P., Brondizio, E., & Moran, E. (2004). Change detection techniques. International journal of remote sensing, 25(12), 2365–2401.
- Zoran, M., & Anderson, E. (2006). The use of multi-temporal and multispectral satellite data for change detection analysis of the Romanian Black Sea coastal zone. Journal of optoelectronics and advanced materials, 8(1), 252–256.
- Kakar, N., Kakar, D. M., Khan, A. S., & Khan, S. D. (2019). Land subsidence caused by groundwater exploitation in Quetta Valley, Pakistan. International Journal of Economic and Environmental Geology, 7(2), 10–19.
- Naz, M. S., Athar, M., & Qayyum, M. (2009). Effect of drought on rangeland productivity and animal performance in dryland region of Balochistan, Pakistan. Agriculturae Conspectus Scientificus, 74(2), 105–109.
No competing interests reported.