Multicriteria evaluation to identify potential areas for level Fanya juu terrace in Central Ethiopia

The costs of soil erosion are worsening in developing countries. In recent decades of Ethiopia, government sectors and different organizations are working on watershed management campaigns. Despite achievements, soil loss from cropland was not reversed to the productive limit. Therefore, this study attempts to provide information on appropriate sites for level Fanya juu terrace in croplands located in Sodo watershed, central Ethiopia. A spatial layer of soil loss, land use/land cover, slope, and topographic wetness index was managed using a multi-criteria decision analysis (MCDA) system. The suitability class of each layer was made based on the national guideline for soil and water conservation published in 2016. Then the analytical hierarchy process was used to obtain the relative weight of each layer. Finally, the suitability map was generated using the weighted combination method. The result shows that 5.5%, 3.2%, and 40.8% of the study area was highly, moderately, and less suitable for level Fanya juu terrace. The remaining is not appropriate at all. Among ground-truthing points collected on an existing terrace, 95% falls into a less suitable class. Assertively, the MCDA provided imperative information. According to pieces of evidence from level Fanya juu terrace, soil loss is estimated to be reduced by 77% in areas highly susceptible to soil erosion.


Abstract Background
The costs of soil erosion are worsening in developing countries. In recent decades of Ethiopia, government sectors and different organizations are working on watershed management campaigns.
Despite achievements, soil loss from cropland was not reversed to the productive limit. Therefore, this study attempts to provide information on appropriate sites for level Fanya juu terrace in croplands located in Sodo watershed, central Ethiopia.

Methods
A spatial layer of soil loss, land use/land cover, slope, and topographic wetness index was managed using a multi-criteria decision analysis (MCDA) system. The suitability class of each layer was made based on the national guideline for soil and water conservation published in 2016. Then the analytical hierarchy process was used to obtain the relative weight of each layer. Finally, the suitability map was generated using the weighted combination method.

Result
The result shows that 5.5%, 3.2%, and 40.8% of the study area was highly, moderately, and less suitable for level Fanya juu terrace. The remaining is not appropriate at all. Among ground-truthing points collected on an existing terrace, 95% falls into a less suitable class. Assertively, the MCDA provided imperative information. According to pieces of evidence from level Fanya juu terrace, soil loss is estimated to be reduced by 77% in areas highly susceptible to soil erosion.

Conclusion
Appropriate soil and water conservation measures unto applicable place is a central part of watershed management campaigns. In long run, this could be effective where farmers' acceptance, proper execution, and regular maintenance of the structure are kept sustainable. In general, procedure followed and results found in this paper can enable an informed decision on farmer's managed cropland to tackle soil erosion and boost productivity.
Background Soil erosion, worsen by land use/ land cover (LU/LC) change is the most common form of land degradation that covers 23% of Ethiopia (Gebreselassie et al. 2016). A gradual change in LU/LC features was observed sourced from demographic changes, poverty, topography, and land policy those are contributing to environmental changes such as biodiversity loss, soil erosion, local climate, and hydrological regime (Desta and Fetene 2020; Musie et al. 2020;Woldesenbet et al. 2020).
In East Africa, the estimated mean annual gross soil loss by water erosion amounts to 4 billion tons with a mean soil loss rate of 6.3 tons/ha/yr, of which 50% originates from Ethiopia. The study adds, cropland contributed 34.8 tons/ha/yr of mean soil loss in Ethiopia (Almaw et al. 2020).
The share of cropland over the total estimated soil loss accounted for 86% in the northwest (Alemu and Melesse 2020), 42.6% in Beshillo catchment, Blue Nile basin (Yesuph and Dagnew 2019), and 11% in the Meki river sub-basin, central rift valley lake basin (Woldesenbet et al. 2020). However, the rate of soil loss from cultivated land varies between 42 and 300 tons/ha/yr (Gebreselassie et al. 2016). Similarly, the upper catchments of the Meki River sub-basin are facing severe degradation irrespective of the land-use systems (Woldesenbet et al. 2020). The risk of soil erosion in the north-eastern watershed is reported very high to severe levels (Woldesenbet et al. 2020). Besides, 9% and 6.5% of the study area stated high to very high soil loss severity levels, respectively (Tadesse and Tefera 2020). In general, the annual cost of land degradation associated with LU/LC is estimated at $4.3 billion (Gebreselassie et al. 2016).
Ethiopia has 500-700 years of implementing traditional soil and water conservation (SWC) measures (Wolka et al. 2018) and ve decades of experience introducing physical SWC to degraded highlands (Herweg and Ludi 1999). Moreover, during the SWC campaign in the 1970s and 1980s, the transitional government (1980) has promoted farmer's level development and conservation of forest by planting trees within its locality (Federal Negarit Gazeta 1980). In 2010, the Ethiopian government launched a community-managed land restoration program through physical and biological SWC measures in more than 3,000 watersheds aiming to enhance agricultural productivity (Mekuria 2020). Also, the "Green Legacy" initiative has recently launched in June 2019 to rehabilitate degraded lands, increase forest cover and/or services, and tackle food insecurity (Getahun 2020).
Adopted SWC measures deposited 54-74% of soil loss and boosted soil quality in northern Ethiopian highland (Subhatu et al. 2017). The outcomes of the SWC measure can be explained as 1) short-term (e.g., runoff control and moisture retention) and 2) medium-term (e.g., vegetation cover and soil fertility improvement) biophysical outcomes ( implemented, soil loss on treated catchment was not as expected and yet exceeds the tolerable limit in Ethiopian highland (Subhatu et al. 2017(Subhatu et al. , 2018. Despite attentive farmers, the program was hardly effective by farmers who didn't obtain a short-term physical bene t (Assefa and Kessler 2018).
Level Fanya juu terrace is a contour terrace where the riser is built (throw up) at the shoulder of the ditch to block soil movement and enhance soil moisture content which further developed into a bench terrace Nevertheless, the construction of the SWC measure needs a considerable effort to not to below technical standards (Assefa and Kessler 2018) and applicability ranges (Hurni et al. 2016). In almost all parts of Ethiopia, dozens of literature are published on the extent and severity of soil erosion and the urgency of SWC measures. However, to support decision-makers and development agents (DAs), Information on the applicability of SWC measures is rare (Kibret et al. 2020). Therefore, this study has aimed (i) to identify areas suitable for level Fanya juu terrace based on national SWC guidelines using spatial-MCDA technique, and (ii) to estimate potential reduction in soil loss from Fanya juu terrace. This would be pertinent to ease the tasks of DAs (extension workers) and support initiatives working on tackling soil erosion and enhancing agricultural productivity. . The watershed is dominated by Vertic cambisol by 56% and the remaining takes chromic vertisol by 44% (Tadesse and Tefera 2020). Nineteen percent of the watershed has less than 5% slope, 30%, 34%, 10%, 5%, and 1% of the watershed area in a slope category of 5-10%, 10-20%, 20-30%, 30-50%, and > 50%, respectively (calculated from Digital Elevation Model (DEM)).

Study area description
Teff, maize, wheat, and enset are subsistence crops and the latter is mostly used by the growers (Tadesse and Tefera 2020; Woldesenbet et al. 2020). The main sources of domestic water in the area include handdug wells and ponds, seasonal streams and rarely availing groundwater wells (Fentaw and Manaye 2019b; Tadesse and Tefera 2020).

Data preparation
This study is a successor of the study that reported the level of risk and extent of soil loss using the Revised Universal Soil Loss Equation (RUSLE) in the same study area (Tadesse and Tefera 2020). To identify areas suitable to level Fanya juu terrace, data such as soil loss and LU/LC data that was mapped through hybrid (supervised and manual) technique from (Tadesse and Tefera 2020) and slope steepness (%) and topographic wetness index (TWI) derived from DEM were used (Table 1).
The quantity of soil loss not exceeding 18 tons/ha/yr is the upper limit where production is possible in the Ethiopian highlands (Hurni 1985  Merging of variables suitability and pairwise weight To map area suitability for level Fanya juu terrace, layers ranked with suitability level (X) and corresponding weight (W) were merged using a weighted sum (WS) approach (Eq. 1).

WS = ∑ X*W (1)
In the end, cropland suitability maps for level Fanya juu terrace were generated. And the result was reclassed; < 2.146 to less suitable, 2.146-2.528 to moderately suitable, and > 2.528 to highly suitable.

Validation of cropland suitability for level Fanya juu terrace
To check the consistency of modeled suitable site for level Fanya juu terrace, 42 ground-truthing points (GTPs) were collected on existing terrace structures that are maintained for a long time. Where the length of the existing terrace exceeds 30m, one or more GTPs were taken with a xed length of 30m difference. This has aimed to test whether areas less susceptible to soil erosion are modeled in need of level Fanya juu terrace or not? Vise-versa. Therefore, the study area has min-max ranges of elevation (1893-2513 m.a.s.l.) and rainfall (900-1400 mm) were considered and the mean percentage of soil loss reduced (77%) of Minchet catchment (Subhatu et al. 2018) and Maybar and Hunde Lafto (Herweg and Ludi 1999) were taken to estimate a potential reduction of soil loss using level Fanya juu terrace (Table 4). 1; Used to estimate an expected percentage of soil loss would be reduced from level Fanya juu. Therefore, the calculated mean of the three percent reduction is 77%.
2; not accepted because the rainfall (Afdeyu) and elevation (Gikuuri) are below the minimum range of the study area.
S: slope (%). Table 5 shows that the majority (78%) of estimated soil loss was fall in a normal level of soil erosion (≤ 18 tons/ha/yr) and set to less suitable for level Fanya juu terrace. Soil loss accounting for 15%, cropland (61%), slope gradient (64%), and TWI (39%) are supposed to estimate and map areas in need of intervention and highly suitable for level Fanya juu terrace (Table 5; Fig. 2). Due to the pairwise importance weight given, the extents of the suitability scale of the variables (Table 5; Fig. 2) would not be consistent for the nal map ( Fig. 3; Table 6).

Site suitability validation
Among the 42 GTPs collected on the existing terraces, 40 (95%) of them are fall in the range of less suitable class and the remaining 2 GTPs (5%) are into a moderately suitable class. Here, the status of the slope around the existing terraces in uenced the modeled suitability that causes 5% of GTPs are fall into a moderately suitable class.
Site suitable for level Fanya juu terrace The weighted overlay results in Table 6 and Fig. 3 shows that 5.5% and 3.2% of the study area is urgently recommended and highly to moderately suitable for level Fanya juu terrace. In line with Fig. 2, the nal suitability map (Fig. 3) attempts to subset croplands with a severe rate of soil erosion and then masked for promising relevance of level Fanya juu terrace (Table 6 and Fig. 3). According to reference proportion (Herweg and Ludi 1999;Subhatu et al. 2018), 77% of soil loss will be presumed to be reduced after implementation of level Fanya juu terrace (Table 7). Accordingly, in areas prominently in need of level Fanya juu terrace, the soil loss will be expected to reduce from 105,990 tons/ha/year to 4,378 tons/ha/year (Table 7).

Conclusion
This study has sought to present areas suitable for level Fanya juu terrace using multi-criteria evaluation of four contributing parameters as soil loss, LU/LC, slope, and TWI. In addition to that potential reduction of soil, the loss was estimated using an equivalent proportion sourced from literature. The result shows that 5.5%, 3.2%, and 40.8% of the area was highly, moderately, and less suitable for level Fanya juu terrace. The validation result indicated that 95% of GTPs collected from the existing terrace was fall into a less suitable class. And the remaining 5%, with the effect of slope, falls into the moderately suitable class. This implies that the MCDA tool, trending a decade of application, is e cient to determine site appropriate for level Fanya juu terrace.
Frequently reported issues allied to farmers' acceptance, structure age, regular maintenance, and proper implementation are important to de ne the effectiveness of SWC measures. Taking into consideration these issues, level Fanya juu terrace has a 77% reported mean potential to reduce soil loss. Accordingly, level Fanya juu terrace would have the potential of reducing soil loss from 105,990 tons/ha/yr to 24,378 tons/ha/yr in areas highly susceptible to soil erosion.
Providing local-level information on the appropriateness of SWC measure is pertinent to support development agents (DAs) and make an informed decision on farmers managed croplands. Therefore, this study has provided technical information on croplands suitable for the most effective physical SWC measure, level Fanya juu terrace. Indeed, for the further effectiveness of level Fanya juu terrace, incorporating with a biological approach is importantly recommended and also practitioners need to work accordingly.