Figure 1 presents the study area. The Republic of Guinea is divided into four (4) natural regions, each with distinct microclimatic characteristics.

1) Lower Guinea (Maritime Guinea)

Lower Guinea (LG) includes the synoptic stations of Boké, Conakry, and Kindia. This region is also known as Maritime Guinea due to its coastal location. It experiences a humid tropical climate, with peak rainfall occurring in August, reaching up to 4000 mm annually in Conakry [23] .

2) Middle Guinea (Fouta Djallon)

Middle Guinea (MG) comprises the synoptic stations of Koundara, Labé, and Mamou. This is the most mountainous region of the country, dominated by the Fouta Djallon massif, which covers approximately 80,000 km², with its highest peak being Mount Loura (1532 m). The rainy season lasts between 5 and 8 months, depending on the location, with annual rainfall below 1300 mm in the northern areas [23] .

3) Upper Guinea

Upper Guinea (UG) includes the synoptic stations of Faranah, Kankan, and Siguiri. It lies between Forest Guinea and Fouta Djallon, at the western edge of the vast Niger Basin. Climatically, this is the driest region in Guinea, characterized by a Sudanian climate. Annual precipitation ranges between 1200 and 1800 mm, with the longest dry season (7 - 8 months). Average temperatures remain relatively high throughout the year, reaching up to 40˚C between March and April [24] .

4) Forest Guinea

Forest Guinea (FG) encompasses the synoptic stations of Kissidougou, Macenta, and N’zérékoré. It features a sub-equatorial climate, with abundant and nearly continuous rainfall for 8 to 9 months each year. This region records a high number of rainy days [25] . The average annual rainfall ranges from 1800 to 2300 mm [23] . Temperatures are mild throughout the year, fluctuating around 25˚C.

The climatic data used in this study consist of time series of precipitation from twelve (12) synoptic stations covering the Guinean territory, where observations are conducted 24 hours a day. These data were obtained from the National Meteorology Directorate (DNM) of Guinea, which manages the observation stations represented in Figure 1 . The precipitation data correspond to daily averages over a 31-year period (1983-2013) for all twelve (12) synoptic stations. These stations are distributed across Guinea’s four (4) natural regions, with three (3) stations per region ( Table 1 ).

The original dataset covered the period from 1981 to 2021. After applying quality control procedures, years with missing data were detected at four stations. Consequently, a complete and continuous period from 1983 to 2013 was selected for the analysis [26] . The completeness rate is then calculated by dividing the number of valid values by the total number of expected observations [27] . Finally, the results are summarized in a comprehensive table ( Table 1 ), providing a clear overview of the quality for all data stations used. This procedure follows the recommendations of the National Weather Service (NWS) regarding quality control, which includes the detection of missing values and the reconstruction of absent data-essential elements for ensuring the reliability and consistency of meteorological time series.

The set of pluviometric stations used exhibits a data completeness rate of 99.929%, reflecting an almost full coverage of the observational series for the period 1983-2013 ( Table 1 ). Specifically, less than 0.1% of the expected observations are missing, indicating that the datasets are practically complete and homogeneous. This very high level of availability enhances the reliability of climatological and statistical analyses by minimizing biases associated with data gaps, providing a robust foundation for the study of rainfall variability and climatic trends in Guinea.

In this study, our approach is based on the calculation of precipitation indices, which are grouped in Table 2 . The precipitation indices are computed following the definitions provided by the Expert Team on Climate Change Detection and Indices (ETCCDI). A detailed description of these indices is available at https://rdrr.io/cran/climdex.pcic/man/climdex.prcptot.html .

The calculation of the total precipitation index (RPTOT) is based on the annual accumulation of rainfall. The frequency index (R1mm) represents the total number of rainy days (days with precipitation ≥ 1 mm) per year, while the dry day index (DD) refers to the number of non-rainy days (precipitation < 1 mm). The intensity index (SDII) is calculated as the mean precipitation on rainy days. The dry spell (CDD, consecutive dry days) and wet spell (CWD, consecutive wet days) indices are based on a threshold of seven (7) consecutive dry and wet days, respectively. The Max_CDD and Max_CWD indices represent the longest annual sequences of consecutive dry and wet days, respectively [28] [29] .

For extreme rainfall event indices (ERE_R99p), considering that in regions where the number of rainy days is very high (which may be the case in this study), the calculation is not solely based on rainy days (above 1 mm) but rather on all days, as recommended by [30] . In such cases, these authors used a percentile-based approach considering all days. Accordingly, we define the ERE_R99p index as the number of days with precipitation exceeding the 99th percentile threshold, computed over the entire period (1983-2013) and considering the rainy season from February to December.

The calculation of the onset and length of the rainy season indices is based on the method of [31] . In this study, the onset of the rainy season (ORS) is defined as the first date after January 1st when cumulative precipitation over three consecutive days reaches at least 20 mm, provided that no dry spell exceeding seven days occurs within the following 30 days to avoid false onsets. Similarly, the end of the rainy season is defined as the first date after September 1st when 20 consecutive dry days are recorded. The length of the rainy season (LRS) is then calculated as the difference between the end and onset dates.

The monsoon onset index (IoM) for each year is determined as the difference between the onset dates at different stations and the station with the earliest onset date, which is considered the reference station.

To quantify the trends associated with the temporal evolution of each index, we applied the Mann-Kendall statistical test with a significance level of 90%. A 90% significance level was selected for the Mann-Kendall test during this period in order to detect moderate trends while limiting the risk of Type I error (incorrect rejection of the null hypothesis). Considering the natural characteristics of precipitation over the 31-year study period, this threshold represents a reasonable compromise between sensitivity to trend detection and statistical reliability. It thus allows for a cautious quantification of both the magnitude and the uncertainty of the observed trends in the analyzed rainfall series.

In this study, the indices are classified into four categories based on their characteristics:

1) Hydrological Indices

Figure 2 illustrates the spatiotemporal evolution of hydrological indices (RPTOT, R1mm, and SDII) in the Republic of Guinea from 1983 to 2013. The total precipitation (RPTOT) is characterized by high rainfall amounts ranging from 1000 mm in Koundara to 4500 mm in Conakry, with a well-defined meridional precipitation gradient ( Figure 2 ), consistent with the findings of [37] .

Figure 2(b) shows a decrease in precipitation frequency towards the north, with a maximum observed in the southern forest region (Macenta), which may be attributed to the influence of vegetation cover [38] . This result aligns with the work of [25] on the spatiotemporal distribution of precipitation in Guinea. A similar spatial distribution is observed for precipitation intensity ( Figure 2(c) ), with the highest values recorded in the coastal (Conakry and Boké) and forest (Macenta) stations.

It is noteworthy that the high rainfall amounts recorded in Conakry are primarily driven by precipitation intensity rather than frequency. According to [37] , this spatial distribution of precipitation can be explained by moisture transport, precipitable water, and humidity divergence patterns. However, in other stations, total precipitation is influenced by both frequency and intensity. With hydrological indices, the observed decrease in precipitation frequency towards the north, with a maximum in the south in the forest region, aligns with the work of [25] .

2) Agro-hydrological Indices

The spatiotemporal evolution of agro-hydrological indices (ORS, LRS, and IoM) is illustrated in Figure 3 . This figure highlights the northward shift of the Intertropical Convergence Zone (ITCZ) during the rainy season. This meridional displacement of the ITCZ largely explains the variability of West African precipitation, particularly in Guinea.

Specifically, the rainy season typically begins around January 15 in N’zérékoré (south) and July 2 in Koundara (north), representing a six-month difference between the southernmost and northernmost stations in the country ( Figure 3(a) ). The stations of Kissidougou, N’zérékoré, and Macenta are characterized by a season length of up to 330 days, indicating that daily precipitation frequency is higher in the southern region, covering almost the entire year ( Figure 3(b) ). In contrast, northern stations (Siguiri and Koundara) experience shorter rainy seasons, ranging from 120 to 150 days. However, a decline in the length of the rainy season has been observed since 2003 in these regions, with durations varying between 110 and 150 days.

Similar to the onset of the rainy season, the monsoon onset index ( Figure 3(c) ) exhibits the same spatial distribution, with an earlier onset in the south and a progressive northward evolution.

3) Agronomic Indices

Figure 4 illustrates the spatiotemporal evolution of agronomic indices, specifically the number of dry days (DD), seven (7) consecutive dry days (CDD), and seven (7) consecutive wet days (CWD).

Consistent with the distribution of the number of wet days index (R1mm, Figure 2(b) ), Figure 4(a) shows that the number of dry days (DD) is higher in the north, particularly in the Fouta Djallon region (Koundara) and Upper Guinea (Siguiri), as well as in the central-eastern areas (Kankan and Faranah), which aligns with the findings of [39] . However, the CDD and CWD indices exhibit a highly heterogeneous spatial distribution across Guinea ( Figure 4(b) and Figure 4(c) ), with some stations recording up to 10 events of either CWD or CDD.

Figure 4(c) further reveals that northern and central-eastern stations are rarely associated with wet sequences lasting more than seven days. In contrast, coastal stations (Boké, Conakry, Kindia) and southern stations (Macenta) exhibit more frequent wet sequences.

4) Extreme Indices

Figure 5 illustrates the spatiotemporal evolution of extreme precipitation indices (ERE_R99p, Max_CWD, and Max_CDD) in the Republic of Guinea from 1983 to 2013.

Figure 5(a) shows that Conakry experiences the highest number of extreme rainfall events (ERE_R99p), with up to 14 events per year. In contrast, the rest of the country is less affected by these extreme rainfall events, with a maximum of only two events in certain years. This pattern can be attributed to the high precipitation intensity and totals recorded in Conakry ( Figure 2(c) ). Regarding the high rainfall amounts recorded in Conakry, explained by the intensity of precipitation, these are confirmed by [37] and [40] , where maxima of moisture flux and precipitable water are observed in coastal areas.

Figure 5(b) highlights that the stations of Conakry and Kindia record the highest number of consecutive wet days (Max_CWD) in the country over the study period. The Max_CWD index is generally higher in the southern regions and decreases progressively towards the north.

Conversely, the maximum number of consecutive dry days (Max_CDD) is recorded in the northern part of the country, with peaks observed in the Boké and Labé stations. This Max_CDD index gradually decreases toward the south ( Figure 5(c) ).

The evolution of precipitation indices in Guinea indicates an increasing trend in hydrological indices ( Figures 6(a)-(c) ). Figure 6(a) shows a significant upward trend in RPTOT, with an increase of 3 mm/year. This increase is associated with a rise in both precipitation intensity (SDII, Figure 6(b) ), which increases by 0.011 mm/day per year, and rainfall frequency (R1mm, Figure 6(c) ), which increases by 0.6 days/year across the entire country. This result aligns with the findings of [21] and [41] .

Similarly, agro-hydrological indices also exhibit increasing trends for ORS and IoM, while LRS shows a decreasing trend ( Figures 6(d)-(f) ). The declining trend in LRS ( Figure 6(e) ) can be explained by the increasing trends in ORS (0.2 days/year, Figure 6(d) ) and IoM (0.2 days/year, Figure 6(f) ), indicating a tendency toward a delayed onset of the rainy season in Guinea.

Regarding agronomic indices, we observe a decreasing trend in DD and CWD, in contrast to an increasing trend in CDD ( Figures 6(g)-(i) ). The most significant change is observed in the DD index, which shows a decreasing trend of 0.65 days/year ( Figure 6(g) ), while CWD exhibits a decline of 0.06 events/year ( Figure 6(i) ). In response to the decrease in CWD, the CDD index shows an increasing trend of 0.015 events/year ( Figure 6(h) ).

Finally, extreme precipitation indices (ERE_R99p, Max_CDD, and Max_CWD, Figures 6(j)-(l) ) reveal a strong decreasing trend of 0.06 events/year for ERE_R99p ( Figure 6(j) ) and an increasing trend of 1.4 days/year for Max_CWD ( Figure 6(l) ). However, Max_CDD shows a very slight decreasing trend of 0.01 days/year ( Figure 6(k) ).