3.1.2. Future Rainfall Pattern

The results of break-point detection and the Mann-Kendall test performed on annual rainfall under RCP4.5 and RCP8.5 are summarized in Table 4.

The results of homogeneity tests and the Mann-Kendall test suggest that, all over the watershed, there is neither break-point nor a linear trend in annual rainfall time series, under emission scenarios RCP4.5 and RCP8.5. Statistically, rainfall time series are homogenous and present no trend. Nonetheless, some variabilities are observed. Specifically, in the northern part and for RCP8.5, rainfall might decrease by 2035 and increase thereafter. The increase in annual precipitation over recent decades in Mono river watershed seems not to be maintained in the future. Future pattern of rainfall may be marked by high variabilities. Such an absence of significant trend in rainfall is also reported by N'Tcha M'Po et al. [40] in Ouémé river basin by 2050, using a REMO model. Similarly, Lawin et al. [42] reported no trend in rainfall pattern in the Imbo north plain region in Burundi under central Africa climatology using an ensemble of eight regional climate models.


**Table 4.** Results of break-point detection and Mann-Kendall test on annual rainfall under RCP4.5 and RCP8.5.

The SPI computed for each region emphasizes this variability (Figure 5).

Under RCP4.5, the number of projected deficit years in the watershed increases slightly from south to north. The years 2020 and 2024 are projected to be extremely wet in the south, whereas 2031 is expected to be extremely dry in the north. However, under RCP8.5, the projected number of deficit years decreases from south to north. The year 2033 is projected to be extremely dry in the north, and again, years 2020 and 2024 for this scenario are expected to be extremely wet in the south. The agreement of both scenarios for years 2020 and 2024 shows that those years will potentially be characterized by extreme precipitations. Furthermore, from south to north, RCP8.5 projects more extremely wet years than RCP4.5.

Furthermore, the pattern of the seasonal cycle of rainfall is expected to undergo some modifications. In the southern part (Figure 6a), rainfall seasonal cycle is projected to keep its bimodal pattern under RCP4.5 and RCP8.5. In addition, both scenarios project almost the same pattern. As in the normal period, the first peak is recorded in June but with a slightly lower amount. The second peak, which normally occurs in September, is expected to extend to October with a higher value.

In the central and northern parts (Figure 6b,c), and under RCP4.5 and RCP8.5, the rainfall regime is projected to be characterized by late onsets and lower precipitation until June, compared to observations, and higher thereafter. Both scenarios converge on the fact that rainfall peak will probably occur in September. The northern part is expected to keep its unimodal pattern under both scenarios, whereas a shift from a transitional regime to a unimodal one is expected in the central region.

Projected future rainfall regimes under the two scenarios are quite similar, apart from in August, where RCP4.5 predicted a slightly larger amount than RCP8.5, and an inverse proportion was predicted in June

Moreover, on a monthly scale, Figure 7 depicts how the rainfall seasonal cycle is expected to change under RCP4.5 and RCP8.5 compared to the baseline period.

Under RCP4.5, the relative change in monthly rainfall varies from −5.5% to 8.4% in the southern part, −29.9% to 22.2% in the central part and −39% to 91.4% in the northern part. For RCP8.5, the expected change ranges from −3.5% to 5.8% in the southern part, −55% to 20% in the central part and −64% to 85.9% in the northern part. Therefore, the biggest changes (both increase and decrease) in monthly rainfall are expected in the northern part of the watershed, regardless of the scenario considered. According to RCP4.5 and with respect to the observation period, the highest rainfall decrease during the period 2018–2050 is expected to occur in February, whereas January and November will record the highest increases.

**Figure 5.** Standardized precipitation index of annual rainfall under RCP4.5 (**left panel**) and RCP8.5 (**right panel**).

**Figure 6.** Seasonal cycles of rainfall in the south (**a**), center (**b**) and north (**c**) under RCP4.5 and RCP8.5.

**Figure 7.** Expected change in rainfall seasonal cycles under RCP4.5 (**a**) and RCP8.5 (**b**).

Under RCP8.5, the highest decrease is projected to affect rainfall in February, whereas the highest increase is expected in November. Thus, the two scenarios project more than 70% increase of rainfall in the month of November by 2050, compared to the observation period. In addition, the highest increase is projected by RCP4.5 and the highest decrease by RCP8.5. Globally substantial changes are expected prior to and at the end of rainy seasons.

#### *3.2. Temperature Trend*

#### 3.2.1. Present Temperature Change

The two homogeneity tests performed on annual temperature revealed the presence of break-point in the time series (Figure 8).

**Figure 8.** Pettit test (**a**) and SNH test (**b**) performed on annual temperature over the period 1981–2010.

The Pettitt test indicates that since 1996 mean annual temperature has increased by 0.51 ◦C in Mono watershed compared to the period 1981–1995, whereas the results of the SNH test implies an increase of 0.52 ◦C from the period 1981–1996 to 1997–2010. In addition, the Mann-Kendall test suggested a significant increasing trend (*p*-value = 3.457 × <sup>10</sup>−<sup>6</sup> and <sup>τ</sup> = 0.6).

Anomalies computation revealed that, from 1981 to 1997, temperature was globally below and near normal, but since 1998 it has stayed above the normal (Figure 9).

**Figure 9.** Anomaly of mean annual temperature in Mono watershed.

It therefore corroborates the outputs from homogeneity tests. In addition, these results are in line with previous studies which noted similar increasing trend of temperature in West Africa (Badjana [43] in the Kara river basin of Togo, Kabo-Bah et al. [44] in the Ghana part of Volta river basin, Oguntunde et al. [45] at Ibadan and Collins [5] over the West-African region).
