The Influence of Weather and Climate Variability on Groundwater Quality in Zanzibar

Climate change and variability have been inducing a broad spectrum of impacts on the environment and natural resources including groundwater resources. The study aimed at assessing the influence of weather, climate variability, and changes on the quality of groundwater resources in Zanzibar. The study used the climate datasets including rainfall (RF), Maximum and Minimum Temperature (T max and T min ), the records acquired from Tanzania Meteorological Authority (TMA) Zanzibar office for 30 (1989-2019) and 10 (2010-2019) years periods. Also, the Zanzibar Water Authority (ZAWA) monthly records of Total Dissolved Solids (TDS), Electrical Conductivity (EC), and Ground Water Temperature (GWT) were used. Interpolation tech-niques were used for controlling outliers and missing datasets. Indeed, correlation, trend, and time series analyses were used to show the relationship between climate and water quality parameters. However, simple statistical analyses including mean, percentage changes, and contributions to the annual and seasonal mean were calculated. Moreover, t and paired t- tests were used to show the significant changes in the mean of the variables for two defined periods of 2011-2015 and 2016-2020 at p ≤ 0.05. Results revealed that seasonal variability of groundwater quality from March to May (MAM) has shown a significant change in trends ranging from 0.1 to 2.8 mm/L/yr, between climate and water quality parameters in MAM and OND. Besides, the paired correlation has shown significant changes in all parameters except the rainfall. Conclusively, the study has shown a strong influence of climate variability on the quality of groundwater in Zanzibar, and calls for more studies to extrapolate these results throughout Tanzania.


Introduction
The quality of groundwater sources is essential as long as chemical, physical, and biological characteristics play a vital role in providing suitable conditions for domestic, agricultural, and industrial uses amongst others [1] [2]. Apart from biological and chemical contaminants water quality is also affected by extreme weather events [3] including drought and floods, which may affect the quality and quantity of both underground and surface water [4]. For example, during severe to moderate droughts groundwater recharge is highly affected, and aquifers are declined, and this decline, may result in either increase or decrease of electrical conductivity (EC) and dry dissolved solids (TDS) [5].
Moreover, natural and anthropogenic climate variability and changes may degrade water quality by deposing contaminants on freshwater sources and hence exploit both pure and freshwater availability [6] [7] [8]. Indeed, reference [9] has shown how the changes and variability in weather and climate can result in the variability of physical characteristics of freshwater for example elevation of water temperature, water discharge, water level as well as retention time, also groundwater temperatures are highly variable due to increased air temperature [10].
The continual abstraction of the groundwater and its unregulated drilling in Zanzibar have been challenging the sea and freshwater balance in aquifers leading to seawater intrusion, along with the events of tourism expansion and population increase in Zanzibar are likely to lead to the total aquifer exploitation if not monitored [11]. Climate change is also well-thought-out to speeding up intrusions and there is a likely increase in sea level and intrusion in low-lying coastal areas of Tanzania. For instance, reference [12] has noted that the Jozani groundwater forest has been affected by the saltwater intrusion and the salinization level is probable to increase under enhanced sea-level rise.
Furthermore, the amount of TDS seems to be increasing due to the introduction of different contaminants including salts that later tend to dissolve into positively and negatively ions, inducing the transfer of electrons to the groundwater source. However, TDS varies concerning time and space, therefore, changes into space and time influence the TDS as the geology and amount of seasonal  [13]. Therefore, it is important for determining groundwater salinity in parts per thousand, or gram per liter. Though the previous research results including [14] have found that there is a correlation between TDS and EC but not always linear, and the associations between EC and TDS are given by In natural waters, the relationship between EC and temperature tends to be nonlinear [15], although, the degree of nonlinearity is not significant in ambient environmental temperature range of (0˚C -30˚C), and a linear equation expressed by [16] is given by Equation (2) ( ) where: EC t = electrical conductivity at a given temperature, t = temperature (˚C), EC 50 = is electrical conductivity at 25˚C, and a = (˚C −1 ) = (0.0187) is a temperature compensation factor [16].
Additionally, the root-mean-squared ( where: e = RMS percentage error, m = the number of data points, EC mes = measured EC, and EC eqn = the predicted EC (Lynne, 2014).
Globally, groundwater is estimated to supply some 36% of all potable water supply, whereby 43% of the water is used for irrigated agriculture, and 24% of direct industrial water supply [17]. However, one of the impacts of the most noticeable climate change could be variation in both surface and groundwater levels and quality, the greatest concern of stakeholders is the decline of quantity and quality of groundwater supplies as the main potable water supply source for environmental and human consumption [18]. However, recently the volume and quality of water have changed resulting in decreasing in groundwater reserves due to extensive overconsumption of water sources and ongoing reduction of precipitation trends and intensity, especially through the past decade [14]. As for recharging the groundwater sources including aquifers, two factors are crucially responsible for the groundwater recharging system. These include precipitation, controlling evapotranspiration as well as controlling irrigation water, or other artificial recharge [19].
Though Zanzibar lies in a tropical climate with two rainfall regimes and significant seasonal and annual rainfall amounts (1500 -1700 mm/yr), Zanzibar people living in both urban and rural areas to a large extent rely on groundwater as a  [11]. Though studies including [11] have shown that 97% of boreholes in Zanzibar municipality have been observed to have a positive trend in EC, salinity, TDS, and chloride levels, indicating the increased salinity due to saltwater intrusion, as well as the change in chloride which is estimated to range from 110 mg/L in 1993 to 284 mg/L in 2004 (i.e. 60% increase).
Apart from extraction and distribution of groundwater in Zanzibar, ZAWA plays a role to supervise water management systems and water quality and standards under the World Health Organization (WHO) guidelines. For instance, based on WHO guidelines the recommended maximum temperature limit at the tap, and for drinking water is 25˚C and below 25˚C [20], respectively. As for TDS, EC, and pH the desired WHO limits for drinking water are 500 -100 mg/l, less or equal to 400 μS/cm and 6.5 to 8.5 [21], respectively. Studies to examine water quality levels based on different environmental and oceanographic parameters had been conducted in Zanzibar. For example, the studies by [11] [12] [22] have been tried to examine the influence of seawater intrusion on water quality in various areas of Zanzibar, but either limited or no study has been conducted to assess the impacts of weather and climate variability on water quality in Zanzibar,

Geological Structure of Rocks and Soil in Unguja
Being the best-documented island in the region in terms of geology, contemporary flora and fauna, this makes Zanzibar (Unguja and Pemba) to be an excellent case study for exploring the effects of island formation [25]. Unguja Island is underlain by Miocene sandy clay marl, alluvial deposits and laterites which are found on the northwest part of the Ubguja up to 130 m above sea level [26]. This area supports a small number of perennial rivers and numerous seasonally active streams, which tend to divert into the ground once they intercept the porous [26]. The rest of the Island is dominated by quaternary coralline limestone reef terraces [26]. Over these lime stones, the landscape is typical of karst (limestone) environments with the development of sink holes, caves, and doline features.
These subterranean features which are not controlled by surface topography supports water flow, especially in the dry season. This is due to the high permeability of limestone rock enabling water to infiltrate into the bedrock to form cave systems, rather than converging in topographic depressions or river channels, resulting in a variable relationship between terrain and water table depth. Through dissolution widening of fractures, preferential flow paths and conduits develop in a positive feedback loop eventually leading to the development of sinkholes and cave systems.

Study Sites, Data Sets and Data Processing
The study took place on purposively sampled ZAWA groundwater wells located of water quality parameters involving TDS, EC and GWT over the purposively selected ground water stations. The data quality control was conducted using the interpolation methods where the missed data during a specific month on annual records were handled by taking the mean of the nearby stations or mean of the long term data of that station for the specific month. This process was conducted under the assumption that there was a small spatial variability of the water quality parameters over nearby stations as well as the mean long term will not highly differ from the specific month [37]. The scarce data values were first identified through a sorting datasheet and graphed so that the extreme value (outlier) was spotted. The fate of the outlier was decided based on its statistical significance.

Correlations, t and Paired t Tests and Simple Statistics
The Pearson double moment correlation analysis was used to show the extent and direction to which climate/weather is likely to relate with water quality parameters.
Hence, the strength between the associations of variables has been calculated using Equation (4) where, xy r is a correlation coefficient of the linear relationship between the variable x and y, Atmospheric and Climate Sciences rameters using long term time series plots and trends of rainfall and temperature. Also, simple statistical analysis included the mean percentage change was used for calculating seasonal means and percentage contributions to those changes over time. However, T-test (t) was used to show the significance of the association between the climate and water parameters at the most significant level (p ≤ 0.05).

Palmers Drought Severity Index
Palmers Drought Severity Index has been used to calculate or estimate the wetness and dryness thresholds where its standardized index spans from ≤−1 (for dry conditions) to ≥+1 (for wet conditions) [38]. The Precipitation Index (PI) equation presented in Equation (5) was used to mark the wet and dry seasons or years, where X is the monthly rainfall total, X is the long term mean rainfall and σ is the standard deviation. The wet and dry conditions are defined as 1) If PI ≥ 1, it was defined as a wet season or year.
PI ranges from 1 ≤ PI ≤ 1.49 or moderate wet; while 1.5 ≤ PI ≤ 1.99 defines very wet; and PI ≥ 2 is defined as extremely wet. Similar negative PI ranges hold for the dryness conditions. Also, the Paired T-test (t) was used for mark the significant change between the two means of climate and water parameters for the two five years' periods of 2011-2015 and 2016-2020, and its significance was decided by the p value of ≤ 0.05 as an influence as noted by [39]. Also, the paired T test was conducted under the assumption (Null hypothesis) that there was no significant change between the mean of climate and water quality parameters for the two defined periods, and its calculation was based on Equation (6) m t s n = (6) where m = mean differences, n = Sample size, s = Standard deviation [40].

Long Term Seasonal Variability Rainfall and Temperature
The results of the long term rainfall variability for the two seasons of MAM Madden Julian Oscillation (MJO) as agreed by [29].
As for the Maximum (T max ) and Minimum (T min ) Temperature, the results of the seasonal variability of T max and T min for the last ten years presented in  for about ≥1.5˚C/yr; indicating strong cold days due to heavy downpours.

Seasonal Variability of Water Quality Parameters
The results of the long term variability of TDS, EC, and GWT during MAM and OND seasons are presented in Figure 3 to Figure 5. The variability of TDS, As for OND season Figure 3    that higher RF increases the number of soluble ions in the water sources our results are in agreement with [42] who noted that rainwater has an aggressive behavior with a high ability to dissolve soil salts, and that the amounts of TDS increase with infiltration process i.e. rainwater increases in salinity during infiltration with increasing depth. Thus the periods with more rainfall (wet seasons) have more influence of TDS and dry seasons have less influence. A similar situation holds for air temperature seems to behave like RF variability with TDS since T min has shown an increasing trend by 0.01˚C/yr Figure 2 air temperature changes as noted by [43]. Indeed, [44] noted groundwater natural quality is more affected by climatic variations than anthropogenic contamination. This indicates that conductivity of ions in water depends on the water temperature i.e. ions move faster when the water is warm. Hence, the apparent conductivity is increased when the water has a higher temperature Indeed, simi- The results of the variability of the EC during OND presented in Figure 4(b) shows that EC at Hali ya Hewa and Makunduchi groundwater stations were decreasing with time (i.e. shows a negative trend) of about −0.18 and −0.08 μS/cm/yr, respectively. Indeed, the increasing trend of about 0.13 μS/cm/yr was noticed at Bumbwini Makoba. Further results revealed that during OND 2012 all three groundwater stations had high EC of up to 3.0 μS/cm/yr, while during MAM RF and T max has shown to be increased. Moreover, the OND 2019 which was characterized by higher RF (Figure 1(b)) has shown to have negative EC for Hali ya Hewa and Makunduchi groundwater station; indicating that EC amount can be manipulated by a factor more than climates such as environmental and geological composition.
As for the variability of GWT results showed a negative trend of >−0.1˚C/yr below the normal range for Hali ya Hewa station during MAM, and a weak positive increasing trend of greater than 0.01 and 0.02˚C/yr for Bumbwini Makoba and Makunduchi, respectively. This variability in GWT trend (positive to negative) could be explained by the fact that RF activities are likely contributing to the cooling effects of groundwater as shown in Figure 1(a) and Figure 2(a) and as in the year 2016. This study finding is well supported by [45] who noted that shifting in precipitation from warmer to colder months could be supported with an increased groundwater recharge during cool seasons which then cool groundwater sources.
The GWT during OND has shown a positive trend of 0.03˚C/yr within the groundwater stations of Bumbwini Makoba and Makunduchi, while negative trend of about −0.12˚C/yr was found at Hali ya Hewa station, the results which can be traced as due to the possible reduction in T max as shown on Figure 5(b) and Figure 2(b). These finding strongly agrees with [10] who noted that both minimum and maximum air temperature are showing positive correlation with the corresponding groundwater temperatures.

Extent to Which Climate Parameters Affect Groundwater Quality
The percentage changes in TDS for the five years (2011-2015) presented in Table 1 shows that TDS in the Bumbwini Makoba has changed by 33.2% while its inter annual variability is in decreasing trend at a rate of −0.  (Table 1). TDS percentage changes at Makunduchi was 32.8%, 29.7% and 37.5%, with a decreasing trend rated at −3.1%/yr (during 2010-2015) and the sharp increasing trend of 7.7%/yr (during 2015-2020). EC had percentage changes of 32.8%, with decreasing trend rated at −1.5%/yr for 2011-2015 at Bumbwini Makoba, and no significant differences in EC percentage changes for 2016-2020 periods. Similar results of 33.4% and 33.4%, holds for changes in EC at Hali ya Hewa (Table 1). Also, results show a 0.1% as a difference in percentage changes for five and ten years' intervals indicating a positive increasing trend over time. EC Makunduchi revealed a percentage change of 28.8%, and 36.9% (Table 1)  As for the percentage changes in climate parameters (RF, T max and T min ) during MAM and OND for the aforementioned periods presented in Table 2   and decline of about 10.8%, and −6.5% respectively. However, T max changed by 35.5% and 34.1% (Table 4)

Correlation between Weather and Water Quality Parameters during MAM and OND
The results of the correlation between weather and groundwater quality parameters at Hali ya Hewa during MAM show that the two were significantly correlated. For instance, TDS has shown a strong negative correlation with T min (r = −0.72, at p ≤ 0.02) while no significant association with T max . Also, TDS and RF had high negative correlation but not significant, indicating that decreased Note that in Table 1, the numbers 11 -15 and 16 -20 refer to the years 2011-2015 and 2016 to 2020, respectively. Also, % C refers to percentage changes RF, T max and T min , while D%RF, D%T max and D%T min refer to the differences in the percentage changes in TDS (mm/L/yr), EC (μS/cm/yr) and WT (˚C/yr), respectively. groundwater recharge, reduces the organization of underground salt as noted by [46]. EC has shown a strong positive correlation with T max (at r = 0.67 and p ≤ 0.05) indicating that like the TDS also EC is affected by higher temperature during MAM season, while GWT has shown a strong negative relationship with the minimum temperature of r = −0.7 (p ≤ 0.05), indicating that lower minimum temperatures affects the groundwater temperatures.
As for OND results revealed a negative correlation −0.5 (not significant) between TDS and T min . However, EC has shown a high negative relationship r = As for the OND season, all correlations were either weak positive or negative but not significant, except for RF and GWT which had a strong significant negative correlation (i.e. r = −0.7 at (p ≤ 0.02).

Paired t-Test
Results of the paired t-test between the means of the groups of weather and water quality parameters at all station under investigation, under the null hypothesis that; "there are no significant changes in means of weather/climate and water quality parameters for periods of 2011-2015 and 2016-2020" presented in Table 3   Note that ** indicates significant correlation resulting to accept the null hypothesis.

Influence of Dry MAM and OND and Seasons on Water Quality Parameters
The

Influence Wet MAM and OND Seasons on Groundwater Quality
The   Table 5(a).
The results of the correlation between groundwater and climate parameters during wet years presented in Table 5(b) shows a strong negative correlation between T min and GWT at r = −0.9 (p ≤ 0.01) otherwise all the results were either weak positive or negative and not significant. Indeed, Table 5(b) shows a strong relationship between parameters of the same type.

Discussion
The mm/L/yr indicating that decreased TDS due to increased rainfall had ejected higher level of impurities to groundwater and affected its quality as supported by [45] and [42] who noted that that higher rainfall with increase the amount of indicating that rainfall variability has an impact on GWT as noted by [45] who noted that shifting in precipitation from warmer to colder months will increase ground research and results in cooling of groundwater sources. Also, the presented results have shown that the percentage changes in seasonal variability in climate parameters were 14.3%, 18.8% and 11.1% for RF, T max , and T min , respectively. While for the groundwater parameters the changes were 5.6% for TDS, 1.75% for EC and 2.4% in GWT. These percentage changes indicate that the variability among weather and climate parameters is noticeable between seasons and how they impact the water quality parameters as noted by [47] that irregularities in-season rainfall and warming can influence annual precipitation and daily temperature over time, and this can influence the variability of groundwater parameters though they show gradual changes as agreed by [9] who noted that weather variability also influences physical variability of water.
Indeed, the presented result in pared t-test has shown that the mean changes in climate and water quality parameters in two times periods (2011-2015 and 2016-2020) were significant except for the RF (Table 3) indicating that the increasing climate variability has posed the changes in ground ware quality. Furthermore, the presented correlation results for the climate parameters and water quality parameters have been varying from the station and with time as well as in direction. For instance, RF had a positive and linear relationship with TDS and EC as agreed by [45] who noted that higher rainfall results in higher TDS and EC amounts since water recharge is responsible for the dissolving of salts into ions.

Conclusions
Based on the presented results and a foregoing discussion the study concludes the following: 1) The variability and changes of climate/weather parameters had to a great extent affected groundwater quality parameters.
2) Seasonal variability of the weather parameters has been shown to affect both direction and strengths of the seasonal groundwater parameters.
3) The variability of the wet and dry seasons has both positively and negatively affects the groundwater parameters.

Recommendations
Based on the foregone discussion and presented conclusion the study recommends the following: 1) Well recording of groundwater data to facilitate further research work.
2) Updating the results with detailed data from Ungula and Pemba ground water stations wells.
3) Extensive monitoring of groundwater data should be taken into account for having a continuous flow of data records.
4) Each groundwater station should be equipped with thermometers.
me with a full-time Norhed Scholarship. Tanzania Meteorological Authority (TMA) and Zanzibar Water Authority (ZAWA) should be acknowledged for their data provision in this study. Last but not least, I would like to send my special and sincere thanks to all people who in one way or another contributed to the fulfillment of this paper.