Impacts of forest cover change on carbon stock, carbon emission and land surface temperature in Sor watershed, Baro Akobo Basin, Western Ethiopia

A tree canopy covering more than 10%, a height of 2–5 m and an area coverage greater than 0.5 ha area is considered a forest (Brandt et al. 2020; Carrión-Klier et al. 2022; Fernández-Montes de Oca et al. 2022; Yan et al. 2022; He et al. 2023). According to a United Nations Report, forests cover about 31% (4.06 billion) of the total land surface with various forms of distribution and absorb 15.6 billion tons of carbon dioxide every year (Nakayama 2022; Schneidewind 2022; Wan et al. 2022; Wang et al. 2022c; Zhang et al. 2022). Russia, Brazil, Canada and the United States were the world's leading countries with forest coverage of 85, 497, 347 and 310 million hectares, respectively (Mamaeva & Isaeva 2021; Nakayama 2022; Xu et al. 2022a, 2022b; Zhang et al. 2023). Due to various anthropogenic activities about 10.34% of the global forest cover was lost between 1990 and 2020 (Lousada et al. 2022).

The Democratic Republic of Congo, Angola, Tanzania and Zambia were the major African countries with forest areas of 66.6, 45.7, 44.8 and 36.7 million ha (Li et al. 2021; Liu et al. 2021; Magos Brehm et al. 2022; Xiao et al. 2022). Substantial studies reported that the African forest coverage had declined to 3.9 million hectares from 2010 to 2020 due to uncontrolled anthropogenic activities (Khan et al. 2018; Zhang et al. 2021a; Han et al. 2023).

Ethiopia is Africa's largest landlocked country with an area of 1.13 km² and is ranked among the world's top 25 biodiversity-rich countries due to the existence of a wide range of topographical and physiographical features that contribute to the formation of various ecosystems and the existence of 6,500–7,000 species of higher plants with 12% species endemism (Tadese et al. 2021a). Ethiopia has five Biosphere Reserves, namely Kafa Coffee Biosphere Reserve, Yayo Coffee Biosphere Reserve, Sheka Forest Biosphere Reserve and Lake Tana Biosphere Reserve, which are home to many flora and fauna (Getahun & Keno 2019).

The Yayo Biosphere Reserve is found in the Illubabor and Buno Bedele zone of Oromia National Regional State in southwestern Ethiopia and covers an area of 167,021 ha. This Biosphere possesses a diversity of life including higher plants (450 species), mammals (50), birds (30), reptiles (10) and amphibians (20), and other cultural and historical significance due to the presence of archaeological sites, ritual sites, caves and waterfalls, as a result of which it is considered a regional forest priority area and forest coffee conservation site (Terfassa 2021). Besides this existing importance, most parts of this Biosphere Reserve face serious challenges due to uncontrolled human-induced activities like agricultural expansion, deforestation and forest degradation, LULC, firewood, charcoal production, investments, overgrazing and other unsustainable use of natural resources.

Different scholars have reported that more than 60% of the forest area in all biosphere reserves has been converted to agricultural land (Dagnachew et al. 2020; Zhang et al. 2021b; Hu et al. 2022; Dibs et al. 2023). Similarly, Tadesse & Aseffa (2019) reported that the reduction of forests in the Nono Sale district in southwestern Ethiopia is due to anthropogenic activities. A study by Negassa et al. (2020) reported that dense open forest in the Komto forest priority area of the Eastern Wollega zone is decreasing as forest is converted to agricultural land, grazing land and settlements.

A forest policy workshop was held in Addis Ababa on 18–19 September 2007 to increase Ethiopia's forest cover from 3.6 to 9% within 5 years, but forest cover is still shrinking in the region (Habtamu 2017). These phenomena can reduce the capability of the forest sector in carbon sequestration which can increase the concentration of greenhouse gas (GHG) in the atmosphere (Olorunfemi et al. 2022; Wang et al. 2023; Zheng et al. 2023; Zhu et al. 2023). The rising GHG emissions led to rising land surface temperature (LST) and climate change, which required urgent research to take appropriate management action (Krisnayanti et al. 2021; Wang et al. 2022a, 2022b; Yang et al. 2022a).

To address the challenges of forest loss, the United Nations organized a scientific group called the Intergovernmental Panel on Climate Change (IPCC) with the aim of decreasing carbon pollution by 45% by 2030 and reaching net-zero carbon emission by 2050 as the best management plan (He et al. 2021; Nieto 2022; Zheng et al. 2022). Forest cover has been playing a crucial role in decreasing carbon emissions and LST, but little research has been carried out to date in the Sor watershed. Assessing the deterioration of forest carbon stock and carbon emission initiate the main targets of climate change mitigation. Advanced technologies such as GIS and remote sensing were used to evaluate spatial and temporal forest cover change in relation to the degradation of carbon stock and carbon emission. This provides evidence-based information for decision makers and concerned stakeholders for sustainable forest resource management. Therefore, this study aimed to fill the existing research gap by analyzing forest carbon stocks, carbon emissions and LST by using geospatial techniques in the Sor watershed in Southwestern Ethiopia.

After image preprocessing such as layer staking of multispectral bands, and false colour combinations, LULC classification of 1992, 2003 and 2022 was applied. The LULC classes of the Sor watershed were classified into forest, grassland, agricultural land, settlement and bare land using supervised classification with a maximum likelihood algorithm.

For accuracy assessment of LULC classes, ground control points were collected using Google Earth for inaccessible areas and large areas. The overall accuracy was calculated by dividing all the pixels properly classified by a total number of pixels in the matrix (Equation (1)).

According to Zhang et al. (2021c), the thermal bands of Landsat 4-5 TM of 1992, Landsat 7 ETM+ of 2003 and Landsat 8 OLI/TIRS of 2022 were used to calculate the LST in the Sor watershed. Thermal bands are band 6 for Landsat 5 TM 1992 and Landsat ETM+ 2005, and band 10 for Landsat OLI/TIRS 2022. The Mono-window approach was used to calculate LST from Landsat TM of 1992, Landsat ETM+ of 2003 and Landsat OLI/TIRS of 2022 data (Chibuike et al. 2018).

⁠, ⁠;

⁠, ⁠, ⁠, ⁠, ⁠, ⁠.

Natural carbon and biomass reserves are found in forests. The world̀s forests are being rapidly destroyed by human activities. These anthropogenic activities, through deforestation and agricultural expansion, have led to increased concentrations of carbon dioxide (CO₂) and other GHGs in the atmosphere, and are a major contributor to global climate change. To calculate the carbon stock as biomass, the total biomass is multiplied by a conversion factor that represents the typical carbon content of the biomass. About 90% of the carbon was stored in plant biomass rather than in the atmosphere (Goheer et al. 2022; Bai et al. 2023; Yang et al. 2023). This study used Landsat TM of 1992, Landsat ETM + of 2003 and Landsat OLI/TIRS of 2022 to assess carbon stocks by using total forest cover (TFC) and IPCC conversion factors.

Area coverage of individual land cover types was calculated over the study period and expressed in km². Results revealed that agricultural lands, bare land, forest land grassland and settlement accounts for 681.9 km²(33.9%), 2.9 km²(0.1%), 1,172.5 km²(58.2%), 149.3 km²(7.4%) and 3.4 km²(0.2%), respectively in 1992. Similarly, in 2003 the total study area coverage consists of agricultural land 893.5 km²(44.5%), bare land 5.2 km²(0.3%), forest land 914.5 km²(45.5%), grassland 191.6 km² (9.5%) and settlement 5.2 km² (0.3%), respectively. In 2022, agricultural land and bare land account for about 956.9 km² (47.6%) and8.5 km² (0.4%) whereas forest, grass and settlement account for 855.9 km² (41.6%), 199.1 km² (9.9%) and 9.6 km² (0.5%) respectively out of the total study area (Table 2).

The accuracy of the classified LULC maps was evaluated in the present study to ensure their trustworthiness. The reference data were compared to the classified LULC classes. The overall accuracy of the LULC types for the study area in the years 1992, 2003 and 2022 was 87.8, 88.4, and 86.7%, respectively. As a result, for the study periods 1991, 2003 and 2021, the kappa coefficients were 0.87, 0.85, and 0.87, respectively (Table 3). The result of the accuracy assessment was more in line with Moisa & Gemeda (2021) which was done in Addis Ababa City.

Results revealed that forest land was reduced by 12.8% from 1992 to 2003. In contrast, agricultural land, bare land, grassland and settlement increased with an area of 10.5%, 0.1%, 2.1% and 0.1%, respectively from 1992 to 2003. Consequently, forest land declined by 16.7% from 1992 to 2022, whereas agricultural land, bare land, grassland and settlement increased by 13.7, 0.3, 2.5 and 0.3%, respectively, from 1992 to 2022 (Table 4). Mismanagement and unwise use of forest resources, timber and charcoal production are some of the major factors that accelerate deforestation in the study area. Our results are in line with the previous study (Negassa et al. 2020), forest land was rapidly decreased by 11.25% due to agricultural expansion, and timber and charcoal production in Komto protected forest priority area, East Wollega Zone, Ethiopia.

Declining forest area, increasing agricultural land and grassland areas indicated that the conservation of natural resources such as forests has declined due to rapid population growth. People living in and around the forest resources exploit timber and charcoal production and expand their agricultural land by deforestation. The finding of Moisa et al. (2022a) confirmed that the dense forest declined due to agricultural expansion in the Geba watershed from 1990 to 2020.

Carbon stock performance was determined based on the status of TFC in the study area. In the first decade, the total forest area accounts for about 1,172.5 km² which causes the occurrences of 675.1 m³ total wood volume, 290.3 ton DMB, 119,744.2 ton/km² carbon stock and 439,066.1 ton of carbon dioxide. In the second decade, the total forest area occupies 914.5 km² with a total wood volume (526.6 m³), DMB (226.4 ton), carbon stock (72,844.4 ton/km²) and carbon dioxide (26,098.6 ton). In the final decade of the study period, the forest area was about 835.9 km² with the extent of total wood volume (481.3 m³), DMB (207 tons), carbon stock (60,860.8 ton/km²), and carbon dioxide (223,158.2 tons) (Table 5). The results show the reduction of the forest area in line with the carbon stock assessment over the study period in the study area. Reduction of carbon stock implies the presence of higher carbon emissions from forest degradation due to various anthropogenic activities in and around the Yayo biosphere reserve. The continued depletion of forests and carbon stocks indicates that there is weak policy enforcement and little resource management in the study area.

The result revealed that the annual carbon stock was degraded to 4,263.62 tons due to 258 km² of forest area degradation which directly leads to 4,026,051.3 tons of total annual carbon emission from 1992 to 2003 in the study area. Similarly, annual carbon stock also degraded by 630.72 tons as a result of 78.6 km² of forest land degradation and causes about 181,443.4 tons of total annual carbon emission from the year of 2003 to 2022. In addition, from 1992 to 2022 annual carbon stock declined by 1,962.8 tons due to 336.6 km² of forest land degradation which may be responsible for 2,418,083.91 tons of annual carbon emission (Table 6). These final results realize that the continual rate of forest degradation is directly proportional to total annual carbon emission and inversely interrelated to the annual carbon stock of the study area. Annual carbon emission increased by 2,155.15 tons due to forest degradation in KPK, Pakistan from 1990 to 2020 (Goheer et al. 2023).

The correlation between LST and LULC was evaluated from 1992 to 2022. The results revealed that the mean of LST was increased in bare land, settlement and agricultural land with the values of 3.9, 3.6 and 3.1 °C, respectively from 1992 to 2022. However, the mean of LST values observed over forest and grassland was low with the value of 2.7 °C and 1.7 °C from 1992 to 2022 (Table 8). The reason for the fluctuating trends of mean LST in the study area is associated with variation in LULC over the study period. Degradation of forest land, expansion of agricultural land and bare land were the main causes of increasing LST. As a result, the conservation of natural resources like forest land was the best solution for climate change mitigation. The previous studies (Li et al. 2022; Merga et al. 2022; Moisa et al. 2022a; Yang et al. 2022c) highlight that high LST was observed in cultivated land and low LST was observed in forest land.

Forest degradation is one of the most recent global problems because it undermines the capacity of forest ecosystems to provide important goods and services to nature. Forest degradation is common in different regions of Ethiopia and is caused by rapid population growth, expansion of agricultural and grazing lands, urbanization and settlement, and increasing demand for forests and forest products for energy production and house furniture. Forest degradation leads to a decrease in forest carbon stocks and an increase in carbon emissions to the atmosphere. In this study, we used geospatial techniques to assess the impact of forest degradation on the carbon stock, carbon emission and LST in Sor watershed, Baro Akobo sub basin, western Ethiopia. Increments of carbon emission in the atmosphere cause a rise in the mean LST by 3.7 °C from 1992 to 2022. This result highlights the inverse relationship between forest coverage and carbon emissions and LST. Investing in afforestation and reforestation programmes and promoting a climate-resilient green economy is therefore essential, not optional, to mitigate the increasing trend of carbon emissions and LST. Moreover, all stakeholders should pay high attention to reducing forest degradation, carbon emission and LST.

The authors acknowledge Wollega University Shambu Campus, Wollega University Nekemte Campus, Raya University and Jimma University College of Agriculture and Veterinary Medicine for the existing facilities to carry out desktop analysis.

The authors agreed to publish this manuscript in the Journal of Water and Climate Change.

No funding was received for this study

M.B.M. participated in research design, document analysis, and manuscript writing. I.N.D. and K.T.D. participated in data collection, methodology, data analysis, and interpretation. M.M.G. and D.O.G. participated in research design, literature review, data analysis, and final draft edition. All authors read and approved the final manuscript for publication.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.