Hanang district is located between latitudes 3.56˚S and 5.87˚S and longitudes 34.16˚S and 39.25˚E, at an altitude of about 1700 m above the sea level. The district is found in Manyara region within north-eastern highlands of Tanzania and has tropical type of climate modulated by the warm tropical Indian Ocean [27] . Manyara region experience two distinct seasonal rainfall-the short rainfall season (Vuli) that starts in October and continues through December (OND) and the long rain season (Masika) that starts in March and continues through May (MAM) [28] - [30] . The seasonal rainfall over Manyara varies in space and time. The variation is higher during short rains (Vuli) than the long rains (Masika) seasons. The area was purposely selected due to floods and landslide disaster which happened on 3^rd December, 2023.

Both primary and secondary data were collected, secondary data were collected from relevant written sources. Primary data were collected using survey and participatory rural appraisal (PRA). A total of 182 respondents were randomly selected in Hanang-Katesh. Also, the study involved the collection and analysis of rainfall, temperature and wind data of Hanang district spanning the period from 1981 to 2023. The Climate Hazards Group Infrared Precipitation with Station (CHIRPS) version 2.0 with global coverage of 50˚S - 50˚N latitude and spatial resolution at 0.05˚ - 0.05˚ latitude–longitude grid from 1981 to 2023 was used. CHIRPS is blended gauge‐satellite rainfall estimates covering most land regions and has high resolution, moderately low latency, low bias, and a long period of record. The low bias of CHIRPS is in reference to the Global Precipitation Climatology Centre dataset [14] [31] .

Daily rainfall data from seven meteorological stations were obtained from the Tanzania Meteorological Authority. These data were aggregated to compute annual rainfall totals and to calculate relevant statistical measures, including mean (P*), maximum, minimum, and standard deviation ( ${\sigma }_p$) [32] for the period of 1981-2023. The equations for these statistical measures are:

$P^{*}=\frac{1}{N}{\displaystyle \underset{i=1}{\overset{N}{\sum }}{P}_i}$ (1)

${\sigma }_p=\sqrt{\frac{1}{N-1}{\displaystyle \underset{i=1}{\overset{N}{\sum }}{\left(P_i-P^{*}\right)}^2}}$ (2)

where:

$P_i$ = individual precipitation values.

N = total number of observations.

Additionally, seasonal rainfall anomalies were determined by comparing observed rainfall (P) with long-term averages for specific time periods [33] , such as March to May (MAM) and October to December (OND). Rain days, defined as days with rainfall exceeding 1 mm, were analyzed alongside rainfall anomalies to assess variations in precipitation frequency. The standardized precipitation anomaly (SPA) was computed using the following equation:

$\text{SPA}=\frac{P-P^{*}}{{\sigma }_p}$ (3)

where:

P = observed precipitation.

P* = mean precipitation.

${\sigma }_p$ = standard deviation of precipitation [34] .

To analyze trends in both rainfall and temperature, Sen’s slope estimator was applied to determine the magnitude of trends, calculated as follows:

$\text{Sen}’\text{s Slope}=\frac{P_j-P_i}{j-i}$ (4)

where:

$P_j$ and $P_i$ are the data points at times j and i respectively [35] .

The Mann-Kendall test was applied to assess the significance of trends, with the test statistic (S) defined as:

$S={\displaystyle \underset{i=1}{\overset{n-1}{\sum }}{\displaystyle \underset{j=i+1}{\overset{n}{\sum }}\mathrm{sgn}\left(x_j-x_i\right)}}$ (5)

where:

sgn is the sign function, returning +1, 0, or −1 [36] [37] .

The coefficient of determination (R-squared) was calculated to evaluate the goodness of fit of the linear regression model:

$R^2=1-\frac{{\displaystyle {\sum }_{i=1}^N\left(y_i-{\hat{y}}_i\right)}}{{\displaystyle {\sum }_{i=1}^N\left(y_i-\bar{y}\right)}}$(6)

where:

$y_i$ = observed values.

${\hat{y}}_i$ = predicted values.

$\bar{y}$ = mean of observed values [38] .

Graphical representations, including plots and figures, were utilized to visually depict the temporal patterns of rainfall variability. This methodology enabled a comprehensive assessment of rainfall dynamics in the study area, providing insights into long-term trends, seasonal variations, and potential impacts on local hydrological systems and hazards, such as landslides [39] . The trend analysis (1981-2023) was conducted using average rainfall from the seven meteorological stations within the Hanang district in the Manyara region, where annual and seasonal rainfall as well as the number of rainy days were analyzed to identify whether heavy rainfall was a source of landslides in that area. Additionally, temperature and wind trends were analyzed using linear regression to determine their impact on rainfall patterns.

Figure 1 designates the locally based indicators of climate change in Hanang district. The study observed that, majority (92%) of the respondents appealed to have experienced changes in the climatic elements. Such changes included rainfall and temperature variations, floods, drought and changes in the onset and cessation of rainfall. Respondent’s perceptions on climate change based on their experiences with the climatic situation of their areas over several years. The analysis of locally based perceptions formed a substantial basis in associating with meteorological data. Similar observation has also been reported by other scholars confirming that local communities have been reporting to have observed changes in the climatic elements [40] - [43] .

The study analyzed the rainfall data for the period of 43 years (1981-2023). The annual total rainfall plot ( Figure 2(a) ) exhibits a variability ranging from a minimum of 421.1 mm to a maximum of 1200.5 mm, with a mean value of 728.6 mm and a standard deviation of 151.4 mm. The positive trend line, characterized by a slope of 0.8252 and an R-square value of 0.0047, indicates a slight upward trend in rainfall over the observed period. Despite the low R-square value suggesting a weak explanatory power of the trend line, the positive slope implies a gradual increase in annual rainfall over time. This suggests a potential shift in precipitation patterns, warranting further investigation into underlying climatic factors driving this trend. Increased rainfall, particularly during intense or prolonged precipitation events, can saturate soil and increase the likelihood of geo-hydrological disaster. If the upward trend in annual rainfall leads to more frequent or intense rainfall events surpassing critical thresholds for soil saturation, it could contribute to an increased risk of disasters. The additional results provide valuable insights into the variability of annual rainfall and its potential implications for geo-hydrological disaster events.

The fact that 26 out of 43 years had total rainfall greater than the long-term mean suggests a considerable proportion of years experiencing above-average precipitation. This could indicate periods of heightened geo-hydrological disaster risk during years of elevated rainfall, particularly if accompanied by intense or prolonged precipitation events. Many other scholars have reported that, intense rainfall events can trigger multiple hazards [44] - [46] . Recently, there has been notable rankings of rainfall totals, with 2020 being the highest and 2019 ranking fifth highest. This suggests a recent period of increased rainfall intensity or frequency. These years may have experienced conditions conducive to elevated landslide and floods risks, highlighting the importance of understanding recent trends in precipitation for landslide hazard assessment and mitigation. The variability in annual rainfall totals underscores the dynamic nature of precipitation

patterns and their influence on landslide occurrence. Understanding the temporal variability in rainfall, including both long-term trends and short-term fluctuations, is essential for assessing landslide risk and implementing effective mitigation measures. Based on the observed rainfall anomalies shown in Figures 2(b)-(d) both present the relationship between the number of rainy days and seasonal rainfall anomalies, highlighting distinct patterns. During 2019, 2020, and 2023, both metrics exhibited positive anomalies, indicating intensified and prolonged rainfall episodes. This convergence implies frequent with much intense rainfall events, which could have led to persistent soil saturation, reducing slope stability and triggering the landslide. The event in December 2023, characterized by significantly lower total rainfall compared to historical records and ranking as the fourth lowest in history (1981-2023), is noteworthy. While this event may represent a deviation from the recent trend of elevated rainfall ( Figure 2(b) ), it does not necessarily diminish the overall risk of landslides, as other factors such as antecedent soil moisture conditions and terrain characteristics also influence landslide susceptibility in Hanag district.

The range of rain days observed over the 43-year period, from a minimum of 10 to a maximum of 33 days, highlights a considerable variability in the frequency of rainfall events during the October to December season. The average number of rain days during this period, calculated at 19 days, serves as a baseline for understanding deviations from typical rainfall frequencies. The year 2023 rain days ( Table 1 ) as the fourth-ranked highest year for rain days, with 30 rain days recorded during the October to December period, suggests a notable increase in precipitation frequency compared to the long-term average. This elevated number

Source: Tanzania meteorological authority 2024.

of rain days may indicate a period of enhanced rainfall activity during the season, potentially contributing to heightened landslide risk due to increased soil saturation and runoff. The occurrence of a maximum daily rainfall of 78.6mm on November 2^nd, 2023, underscores the intensity of precipitation events experienced during the season. Such high-intensity rainfall events can contribute to rapid soil erosion, runoff, and increased geo-hydrological disaster susceptibility, particularly if they coincide with periods of already elevated soil moisture levels.

Overall, November experienced the highest rainfall among the three months with frequent and intense rainfall events contributing to a substantial cumulative rainfall for the month. December also witnessed significant rainfall, although slightly lower than November, while October had comparatively lower rainfall with sporadic rain events and fewer rainy days. By integrating information on rainfall frequency, intensity, and anomalies ( Figure 2(d) ), over the October to December period, alongside historical rainfall data and geo-hydrological disaster events, a more comprehensive understanding of the relationship between precipitation patterns and geo-hydrological disaster risk can be achieved. This holistic approach facilitates the identification of periods of elevated landslide susceptibility and informs proactive measures to mitigate the associated hazards.

The analysis of the annual mean temperature plot ( Figure 3(a) and Figure 3(b) ) reveals several key findings regarding temperature trends over the observed period. The mean annual temperature for the study area is calculated to be 20.3˚C, with fluctuations ranging from a minimum of 19.6˚C to a maximum of 20.9˚C. The positive trend line fitted to the data indicates a gradual increase in mean annual temperatures over time, supported by a moderate R-squared value of 0.4561. This suggests that approximately 45.61% of the variation in annual mean temperatures can be explained by the linear trend observed. The slope of the trend line is calculated to be 0.0178, indicating a steady upward trend in temperatures. While the slope value may seem small, its significance lies in its consistency over the observed period, signifying a persistent warming trend. However, it’s essential to note that local factors, such as urbanization, land use changes, and proximity to water bodies, may influence temperature readings and should be considered when interpreting the trend. Overall, the positive trend in annual mean temperatures highlights the importance of continued monitoring and adaptation efforts to address the impacts of climate change in the study area. Further analysis reveals that 23 out of the 43 years in the historical dataset (1981-2023) exhibited mean temperatures exceeding the long-term average of 20.3˚C. Notably, the highest recorded temperature in this dataset, reaching 20.9˚C, occurred in 2023, representing an outlier compared to the historical temperature records. This anomaly underscores the significance of recent temperature variations and their potential implications for climate trends in the study area. The frequency of years with temperatures surpassing the long-term mean suggests a consistent pattern of above-average temperatures, which aligns with broader climate change trends observed globally. These findings highlight the need for ongoing monitoring and assessment of temperature trends, as well as proactive measures to mitigate the impacts of climate variability and change on local ecosystems, communities, and infrastructure. Several studies on climate change have reported that changes in temperature may have imperative hydrological consequences, both indirect and direct, through prompting potential evaporation, total evaporation and soil moisture [6] [7] .

The trend line plot for the MAM period ( Figure 3(c) ) exhibits a positive slope of 0.0276, indicating a slight upward trend in temperatures during this season. The R-squared value of 0.1199 suggests that approximately 11.99% of the variability in MAM temperatures can be explained by the linear trend observed. Temperature anomalies during the MAM period range from a minimum of −2.2˚C to a maximum of 2.4˚C, indicating significant variability in temperature deviations from the long-term average.

The OND trend line plot ( Figure 3(d) ) also shows a positive slope, with a steeper rate of increase compared to the MAM period, reflected in a slope value of 0.0499. The R-squared value of 0.3924 indicates that approximately 39.24% of the variability in OND temperatures can be attributed to the observed trend. Temperature anomalies during the OND period vary from a minimum of −2.4˚C to a maximum of 2.0˚C, indicating notable deviations from the long-term average, with temperatures tending to be warmer overall. These findings suggest distinct seasonal temperature trends, with both the MAM and OND periods exhibiting positive anomalies and upward temperature trends. While the explanatory power of the trend lines is relatively modest, the consistent positive slopes and variability in temperature anomalies underscore the importance of considering seasonal variations in temperature when assessing long-term climate trends. Scholars have suggested that, given the complexity of temperature anomalies, further research may focus on understanding the underlying drivers of these seasonal temperature patterns and their implications for regional climate dynamics and ecosystem functioning [47] .

Table 2 and Figure 4 indicates the wind rose statistics for annual, March to May (MAM) and October to December (OND) for climatology period of 1991-2020. The observed directions were SSE (South South East), EEN (East East North), SE (South East) and E (East) wind direction. Based on the wind rose statistics provided for the annual, March to May (MAM), and October to December (OND) periods from 1991 to 2020, we can observe that the predominant wind direction (SSE - EEN) accounts for the majority of occurrences, with varying wind speed ranges. Wind speeds between 2.1 m/s and 5.7 m/s constitute the largest percentage of occurrences (36.8% to 39.2%). For MAM wind distribution, the predominant wind direction is SE - E. The wind speeds between 2.1 m/s and 3.6 m/s have the highest percentage of occurrences (40.1%). For OND, the predominant wind direction is SSE - EEN, consistent with the other periods. Wind speeds between 3.6 m/s and 5.7 m/s have the highest percentage of occurrences (44.7%). Wind patterns can have significant impacts on local weather patterns, including rainfall distribution, wind speed and direction, and temperature variations. The prevailing wind directions across all periods are generally consistent, suggesting a certain regularity in wind patterns. The moderate wind speeds in the range of 2.1 m/s to 5.7 m/s are quite common, especially during the months of November and December. Wind blowing from the SSE - EEN direction, which is predominant, could potentially influence soil erosion particularly if coupled with heavy rainfall. Numerous researchers have emphasized the influence of high wind speed on evaporation processes which destabilizes the boundary layer and causes deep convection, and hereafter, intensifies rainfall [48] . These rainfall patterns, especially the intense rainfall in November, could have significant implications for the occurrence of the landslides and floods in the study area.