Data on spatiotemporal land use land cover changes in peri-urban Addis Ababa, Ethiopia: Empirical evidences from Koye-Feche and Qilinto peri-urban areas

Urban expansion is one of the key problems in Ethiopia resulting in displacement of the rural people inhabiting areas bordering the cities/towns. It is also resulting in land use land cover (LULC) changes affecting the livelihoods of the people and the ecosystems [1], [2]. The data presented in this article, therefore, shows the spatiotemporal LULC changes of peri-urban expansion areas known as Koye-Feche and Qilinto, around Addis Ababa City (the capital of Ethiopia). The data were generated from Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) images (with path/row numbers 168/054) by using ERDAS EMAGINE 2014 software. The precision of the images was verified by geolocation data collected from ground control points by using Geographic Positioning System (GPS) receiver. The data indicate that the built-up areas have increased by 1017.85 ha (10.178 km2) with 89.1%, 58.4%, 47% and 13% decline of plantation (mostly eucalyptus woodlots), grasslands, riverine vegetation (forestland) and cropland, respectively, between 1986 and 2016.


Specifications
Data accessibility The data is with this article

Value of the data
The data is helpful to Addis Ababa City administrators to speculate the extent of the spatiotemporal expansion of Addis Ababa and its potential impacts on the surrounding areas.
The data provides information on the status of urban expansion towards rural peri-urban areas around Addis Ababa.
The data is vital to model urban expansion towards rural peri-urban areas surrounding Addis Ababa to mitigate its adverse impacts on the livelihoods of the people inhabiting the area and the ecosystem.
The data is useful to researchers, urban planners and experts working in the field.

Data
The data in this article provides information on the spatiotemporal LULC changes in Koye-Feche and Qilinto urban expansion areas around Addis Ababa City between 1986 and 2016. Figs. 1-3 illustrate pictorially the spatiotemporal LULC classes of the area in 1986, 2000 and 2016. Cropland and grassland had dominated the land use in 1986 ( Fig. 1) with very few built up areas, plantation and forestland. In 2000 (Fig. 2), plantation was tremendously expanded and cropland was considerably reduced. In 2016 (Fig. 3), built up area was extremely enlarged, cropland was almost disappeared and some part of the cropland was replaced by built up areas. Table 1 Table 5 illustrates rate of LULC change in hectare per year.

Experimental design, materials and methods
Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM þ) images (with path/ row numbers 168/054) as well as GPS-based ground survey records were vital data sources for this data article. The analyses, such as data extraction and LULC classification, were done by using ERDAS IMAGINE Version 2014 software. The images were geo-referenced with World Geodetic System  Supervised and unsupervised image classifications techniques were employed to extract the data [3]. Supervised classification involved selecting pixels that represent land cover classes that are recognized by the analyst. Unsupervised image classification is more computer-automated. It enables the analyst to specify some parameters that the computer uses to reveal statistical patterns that are inherent in the data. These patterns are simply clusters of pixels with similar spectral characteristics. Due to similar spectral characteristics of grass, crop and bush lands, which were determined to be independent classes before classification, the application of unsupervised classification may not give good results. As a result, in the data extraction process, supervised image classification was used. After determining the land use features, the next step was LULC change detection. This was done through overlaying the classified satellite imageries and analyzing by image differencing algorithm. Finally, the outputs of image classifications were verified by conducting ground truth while recording x and y coordinates of the sample spatial features using GPS. Depending on the scope of the study and visual interpretation of the satellite imageries, five classes were identified in Koye-Feche and Qilinto periurban areas. These are forestland (including riverine bushes and shrubs), plantation (mostly eucalyptus woodlots), grassland, cropland and built-up area (including any sort of housing construction, road and other infrastructures) in the locality.