Analysis of spatio-temporal climate variability of a shallow lake catchment in Tanzania

This study analyzed the trends and spatio-temporal variability in rainfall and temperature, and the length of the rainy season (LRS) in the Lake Manyara catchment, Tanzania, covering a period between 1988 and 2018 using stations and satellite climate product. The Mann-Kendall statistical test, Sen ’ s slope estimator, and inverse distance weighting interpolation techniques were used to detect the trends, magnitude of trends and spatial distribution of rainfall and temperature. A modi ﬁ ed Stern ’ s method and water balance concept were used for rainfall onset, cessation and LRS analysis, while standardized precipitation index (SPI) was used to investigate the wetness or dryness of the area. The results showed high variability and decreasing trend (4 mm/y) in annual rainfall, and non-signi ﬁ cant increasing trend for minimum and maximum temperature. Rainfall increased from the Western to the Northern part of the catchment whereas reversal pattern was noticed for temperature. The SPI shows a signal of normal condition (about 65%) for all stations – with few years showing evidence of wetter and drier conditions. The LRS showed a decreasing trend indicating a potential negative in ﬂ uence on rain-dependent activities. There is a need, therefore, for adaptation measures such as improving water productivity and irrigation at the farm and catchment level.


INTRODUCTION
Climate variability and changes have been projected to increase by various climate modeling systems. These changes have been associated with increase in anthropogenic greenhouse gases (GHGs) emissions leading to global warming. Other anthropogenic factors such as industrial activities, economic and population growth are also known to contribute to climatic variabilities (Pachauri et al. ). Anthropogenic activities such as industrial and agricultural activities have contributed significantly to the increased in atmospheric concentrations of greenhouse gases since the 20th century (Solomon et al. ). Undeniable evidence suggest that the impact of greenhouse gases on the earth's atmosphere is significant and inevitable (Hartmann et al. ). The GHGs emissions impacts are associated with changes in various climatic parameters such as temperature and rainfall that may lead to catastrophic environmental and socioeconomic events including water scarcity, health problems, energy deficiencies, poor livelihoods, food insecurity, human insecurity, poor forestry practices, poor agricultural yields and low economic growth (Holman ; Chang'a et al. ).
Precipitation and temperature trends have a direct influence on the management of water resources. Various studies indicate that the changing patterns of precipitation have a direct impact on the catchments ( Jones et al. ). Various studies report a considerable attention on rainfall and temperature variability related to the water resources. Trend detection in climate time series is one of the interesting research areas in climatology in recent years (Zarenistanak et al. ). Spatial-and temporal-scale rainfall trend analysis has been of concern since the past century because of the increased attention given to the global climate change by the scientific community (Haigh ). Trend analysis of precipitation and temperature series research has also gained substantial attention lately due to improved capabilities to explore the climatic variability and extreme climatic events (Nenwiini & Kabanda ). In Tanzania, for example, heterogeneous climate condition is experienced due to the complicated topographical patterns, numerous inland water bodies, variation in vegetation types and land-ocean contrasts (Kijazi & Reason ). This complexity has led to experiences of the spatial climatic variationa relatively small change in distances provides substantial variation (Karim et al. 
This study was conducted in the Lake Manyara catchment, which is at the upper part of Lake Manyara sub-basin with an area coverage of about 7,920 km 2 (Attarzadeh et al.

Climate data
Daily rainfall and temperature data for 32 rainfall weather stations (Table 1) (5)) calculates the slope of the line formed by plotting the variable of interest against time, but only considers the sign and not the magnitude of this slope. The MK statistic S, is computed as follows: where x j and x k are sequential data values for the time series data of length n. The sgn series is defined as: ( 2) whenever there is an identical and independent dataset distribution, the mean of S is zero whereas the variance of S is given by Equation (3).
( 3) where t i is the extent of any given tie. Σt i denotes the summation over all ties and is only used if the data series contain tied values. The standard normal variate Z is calculated as indicated in Equation (4): The trend is decreasing if Z is negative and increasing if Z is positive. H 0 , the null hypothesis of no trend, is rejected if the absolute value of Z is greater than Z1-α/2, where Z1α/2 is obtained from the standard normal cumulative distribution tables.
The Sen's slope estimator (Equations (5) and (6)) was used to determine the magnitude of the trends after obtaining the direction of the trend with the Mann-Kendall test.
The method uses a linear model to calculate the change of slope, and the variance of the residuals should be constant in time (Sen ).
for all k < j and i ¼ 1, ::N (5) where Q i is the slope between data points X j and X k , Q med is median slope estimator which reflects the direction of the trend in the data.

Spatial distribution analysis of rainfall and temperature
The inverse distance weighting (IDW) interpolation technique was used to spatially interpolate the rainfall and IDW method is given by Equation (7) as follows: where p is the number of points, d is the distance between points, and w is the weighting function.

Rainfall anomaly and LRS analysis
An investigation of dry, normal and wet years in the Lake Manyara catchment were performed using the Standardized

RESULTS AND DISCUSSION
Monthly mean rainfall and temperature The results for the mean monthly rainfall, maximum and minimum temperature for Babati, Monduli and Mbulu

Spatial-temporal variability of rainfall and temperature
The results for the annual series plots of rainfall for the Lake Manyara catchment (Figure 3), and for the Eastern, Western and the Southern part of the catchment (see Figure S1 in

Trends of rainfall and temperature
The results for the MK trend analysis (Table 2a)  4.3 mm/y and 5.0 mm/y was noticed for the Babati and Monduli stations respectively, while a slightly decreasing trend was noticed for the Mbulu station. The entire catchment also showed a decreasing trend in annual rainfall at a rate of 5 mm/y. The MK test also computed for the maximum temperature reveals a non-significant increasing trend in two out of the three stations (Table 2b).
Unlike maximum temperature, the mean annual minimum temperature showed a significant positive trend for the three stations (Table 2c). This means that during the night, temperature increase significantly due to the release of longwave radiation. The minimum temperature showed an increasing trend at the rate of 0.024 C/y com-

Standardized precipitation index results
The analysis was focused on understanding the sensitivity of SPI to rainfall deviation from the annual mean and discovered the wet, normal and dry condition in the area. The SPI analysis results (Table 3 and

Rainfall onset, cessation and length of the rainy season
Rainfall in Lake Manyara catchment exhibit variability for both time and space. The start and end of rainfall seasons for the period from 1988 to 2018 are given in Table 4. The results show inconsistency and high variations in the start A similar pattern is followed during the cessation period.
The results of the length of seasonal rainfall (LRS) in days are presented in Figure 9 and

CONCLUSIONS
This study provides a clear picture of the spatial and temporal variability of rainfall and temperature, trends and length of rainy season over the past 30 years for the Lake Manyara catchment in Tanzania. From this study, climate status of Lake Manyara catchment shows high variability in both spatial and temporal rainfall and temperature distribution.
Decreasing trends in annual rainfall was found in most stations while the maximum and minimum temperature showed an increasing trend in the area. Locally, drought conditions showed a more enhanced signal compared to wet conditions. The periods of onset and end of the rainy season were inconsistent in most of the meteorological stations leading to high variability in the length of the rainy season.
The present study gives valuable information to authorities responsible for planning of water resources within the catchment and in the country at large. Therefore, climate change adaptation policies in the studied catchment may include measures such as changing the crop type and improving water productivity and irrigation practices at the farm and basin level.