2.3.2. Interannual Rainfall Variability Analysis

The standardized precipitation index (SPI) [25] is a tool recommended by the World Meteorological Organization (WMO) and widely used for quantifying the precipitation deficit over different timescales (3 to 48 months). For the selected timescale, rainfall records are fitted with a probability distribution which is then transformed into a normal distribution so that the mean SPI for the location and desired period is zero. Hence, this method improves the common anomaly method (Equation (2)), which does not take into account the fact that rainfall is typically not normally distributed for a cumulative period of 12 months or less.

$$I(i) = \frac{\mathfrak{x}\_i - \overline{\mathfrak{X}\_{\mathfrak{m}}}}{\sigma},\tag{2}$$

where *I*(*i*), *xi*, *xm* and σ are respectively the standardized index of year *i*, the value for the year *i*, the average and the standard deviation of the time series.

In the present study, the SPI 12 (for 12 month's timescale) is used to assess rainfall deficit or excess on a yearly basis. Moreover, SPI 12 is the one recommended for watershed analysis [26]. Table 2 presents the guidelines for analyzing SPI values [25,26].

**Table 2.** Standardized precipitation index (SPI) values and their meanings.


#### 2.3.3. Future Climate Analysis

Raw outputs from RCMs must be corrected prior to local impact studies because of the bias they encompass. There are several bias correction methods but in this study, the methods of delta, linear scaling and empirical quantile mapping (EQM) are used because they have produced satisfactory results in previous studies carried out in similar climatic regions [27–29]. The results of Ntcha M'po et al. [27], Essou and Brissette [30] and Speth et al. [31], who bias-corrected REMO data in the Ouémé watershed (Benin), guided the choice of correction methods in this study. Rainfall was corrected with delta method in the south, multiplicative scaling in the central part and EQM in the north. As for temperature data, they were corrected using only EQM.

2.3.4. Percentage of Relative Change in Rainfall Seasonal Cycle

It is computed using Equation (3).

$$P\_{c,m} = \frac{R\_{proj,m} - R\_{obs,m}}{R\_{obs,m}} \times 100\tag{3}$$

where, *Pc*,*m*, *Rproj*,*<sup>m</sup>* and *Robs*,*<sup>m</sup>* are respectively the percentage of change for *m*th month, average projected rainfall for month m, and the average rainfall of month m during observation period.
