The method presented in Section 2.2
was used to obtain projected rainfall data, which were then utilized to estimate annual average precipitation (H) for each scenario. Then, according to the procedure outlined, the mean annual temperature map was created by taking into account each potential future scenario (RCP 4.5, and RCP 8.5). These maps were used to solve Equation (2) and produce the temperature coefficient (T) maps depicted in Figure 7
. Then, we performed the used model to estimate soil losses in the future, with a static land cover to assess the impacts of climate change on soil erosion, irrespective of changes in land cover.
3.4.1. Projected Soil Erosion Rates Using All Models
depicts the predictions of erosion rates using the EPM model for each model (ECEARTH, IPSL, MPI, and HAD), scenario (RCP4.5 and RCP8.5), and time period (2030–2060 and 2060–2090). Comparing the models, it appears that models CNRM, ECEARTH, and MPI forecast extremely high rates, whereas model IPSL predicts the lowest rate. In fact, the estimated rates for the years 2030–2060 under the RCP4.5 scenario range from 0–37,014, 0–37,621, 0–29,120, and 0–56,952 m3
for models CNRM, ECEARTH, HAD, and MPI, respectively, while model IPSL predicts the lowest rates, which will not exceed 13,233 m3
. Nevertheless, under the pessimistic scenario (RCP8.5), the rates will rise to 59,514 m3
(model MPI) and 42,077 m3
(model CNRM). For the period 2060–2090, model CNRM predicts a significant increase in rates, with maximum values of 42,077 m3
under RCP4.5 and 51,615 m3
under RCP8.5. Also, models ECEARTH and MPI forecast continuously high rates, ranging between 0–40,723 (RCP4.5) and 0–48,302 (RCP8.5), 0–59,600 (RCP4.5), and 0–59,472 (RCP8.5) m3
, respectively. However, models HAD and IPSL predict an increase in soil losses under the RCP4.5 scenario, with maximum values reaching 30,212 and 20,000 m3
, respectively, whereas, under the pessimistic RCP8.5 scenario, the soil losses decrease significantly, with maximum values reaching only 28,794 and 10,837 m3
(model HAD and model IPSL, respectively) (model IPSL).
These results enabled us to calculate the change rates for the minimum and maximum values (Table 7
). Consequently, the table demonstrates an increase in soil loss rates across all time periods and scenarios. In fact, for the first period (2030–2060) under scenario RCP4.5, the average maximum values of the five models will increase by approximately 10,981.94 m3
(46.13%) compared to the initial value (baseline). The situation becomes more serious under the RCP8.5 scenario, where the maximum losses reach 61.81% (14,714.94 m3
). For the second period (2060–2090), and across the two scenarios, the change rates reach 56.59% (about 13,472 m3
) for the maximum losses under the RCP4.5 scenario. Moreover, the change is approximately 67.2% (15,997.94 m3
) for the maximum losses under the RCP8.5 scenario.
3.4.2. Averaged Projected Soil Erosion Rates
The annual soil loss (m3
/year) and catchment erosion rate (m3
/year) were calculated using the previously specified factors and the EPM technique for the baseline (1976–2021) and future periods (2030–2060/2061–2090), taking into consideration two distinct future scenarios (RCP4.5 and RCP8.5) and a static land cover (Figure 11
and Figure 12
). Lastly, the findings were compared in order to quantify the effects of climate change on soil erosion.
The findings (Table 8
) show that the total soil loss would be elevated by 137,738 m3
/year, 176,789 m3
/year, and 107,885 m3
/year for the periods RCP4.5/2030–2060, RCP8.5/2030–2060, and RCP4.5/2061–2090, respectively, and, as a result, the erosion rate would be elevated by 97.11 m3
/year, 124.64 m3
/year, and 76.06 m3
/year, in the same order. However, the RCP8.5/2061–2090 scenario suggests a 175,612 m3
/year reduction in total soil loss and 123.82 m3
/year in the erosion rate. Table 8
summarizes the detailed findings for the present and future climates.
To examine the influence on the spatial variation of erosion rates, Figure 12
illustrates the classes of the variation in the erosion rate from the present condition to the end of the 21st century, in view of the two climate change scenarios, RCP4.5 and RCP8.5. Clearly, the areas occupied by the classes with soil loss rates below 5000 m3
/year will decrease significantly in comparison to the classes with soil loss rates beyond this threshold. In fact, the soil losses between 500–5000 m3
/year will decrease from 53.40% of the total basin surface in the baseline period to 51.33% in the period 2030–2060 under the RCP4.5 scenario, to 50.34% under the RCP8.5 scenario and to 50.59% for the period 2060–2090 under the RCP4.5 scenario, and to 48.53% under the most pessimistic scenario, RCP8.5, by the year 2090. However, the loss rates in excess of 5000 m3
/year will increase significantly, particularly in the first period under the two scenarios.
In addition, in order to highlight the spatial variability of the change rates under different scenarios in the two time periods (Figure 13
), we have spatialized these changes by calculating the difference between the expected losses for each period and those of the baseline period. Consequentially, the modeling results indicated that, in high-altitude regions and notably upstream of the catchment, erosion rates might significantly increase under the considered climate change scenarios.
In terms of the Mediterranean region’s future climate conditions [36
], and especially Morocco, a warmer and dryer climate is anticipated. Furthermore, an increase in temperature of between +2 °C and +5 °C is anticipated in Morocco, despite the fact that there would be a −20% decrease in precipitation in the center of Morocco [34
]. In addition, according to a review of the literature, future soil erosion will be affected by climate change. However, all the previous research has shown that the projected soil erosion patterns in the studied areas could be extremely heterogeneous. Our research substantially supports these findings, confirming an overall rise in soil erosion in the catchments of the Atlas Mountains caused by climate change, modulated particularly by rainfall as a determining factor that can lead to regional patterns of increasing erosion. The definition of a general trend becomes very difficult, as a result. In comparison to other catchments in the Atlas Mountains, the rate of erosion in the research region was assessed to be rather high. The weak geological subsurface and steep slopes, in addition to the degraded forest cover, were shown to enhance the development of erosion phenomena. Concerning the estimated erosion rates for the baseline period (3129 m3
/year on average), they are quite equivalent to those determined by [40
] in the Moroccan context. In addition, it was revealed that, although precipitation will decrease, the soil loss rates will continue to rise [43
], particularly in the first period between 2030 and 2060. This can be explained by the fact that, even if the annual rainfall in the basin decreases in quantity, taking into account its spatial variability, it is clear that the maximum values expected will be localized in the upstream portion of the basin and coincide with the areas most susceptible to high loss rates, due to their high exposure resulting from their degraded vegetation cover with significant slopes. On the Mediterranean scale, [45
] the research confirms similar losses in an Italian river basin, as well as a tendency of change in loss between 6 and 10% by 2060, which is consistent with the results of this study. Forthmore [39
] suggested an increased soil loss in a mountainous catchment in Greece.
Comparing the two scenarios, it appears that scenario 4.5 predicts a rise in soil loss during both time periods, with a peak in the first. In contrast, the second scenario (8.5) forecasts a substantial increase in losses during the first period and a major drop during the second. This can be attributed to the fact that the first scenario, 4.5, is more optimistic regarding the rates of change of precipitation and temperature during the initial period. In contrast, scenario 8.5 is more pessimistic, predicting a large decrease in precipitation by 2090, which will reduce the closely associated loss rates. Comparing the periods, it is evident that, given the precipitation and temperature trends, soil losses in the first period (2030–2060) are going to be greatly susceptible to the effects of climate change. In addition, the results of the three RCP 4.5 (2060 and 2090) and RCP 8.5 (2060) scenarios indicate that precipitation will fall significantly while erosion rates go up (+3%). This can be explained by the spatial component; even if precipitation on the basin scale decreases, the rates upstream of the basin maintain their high values (Figure 9
a–d), and these areas coincide with the regions with degraded vegetation and steep slopes, potentially leading the EPM model to estimate high erosion rates.
These results can be a useful resource for decision-makers and long-term management. However, it is essential to recognize the limitations of the employed approaches. Certainly, there is a pressing need for accurate future climate estimates in order to plan catchment management and infrastructure projects. The regional climate model (RCM) is the most current instrument for simulating future climatic conditions. Using models, on the other hand, might increase uncertainty with respect to a variety of key temporal and geographical scaling concerns with the input data. Generally, RCMs overestimate temperature and underestimate monthly precipitation, whereas better simulation is possible in the spring and summer for temperature data and the winter and autumn for precipitation data [36
]. Another constraint is that the study only takes climate factors (rainfall and temperature) into account, assuming that all other factors (slope, soil erodibility, land use and land cover, and conservation practices) will remain static in the future. Soil erosion and its vulnerability may rise or fall throughout the twenty-first century depending on changes in these factors. However, the statistically based downscaling model and the EURO-CORDEX regional climate model predict that, in the future, there will be more or less rainfall, which will lead to an increase or decrease in vegetation cover, land use, and conservation practices. In addition, in general, the climate change scenarios exhibit considerable uncertainty in their projections. Finally, empirical models represent an alternative analysis tool in the absence of specific tangible data. Nonetheless, field observations, coupled with an estimation of the future evolution of additional factors influencing erosion rates, such as vegetation and human impact, could boost the accuracy and dependability of the entire process.