This study used quality-controlled mean monthly, daily maximum and minimum temperature data from Arusha and Kilimanjaro meteorological stations, for the period 1961-2019 and 1972-2019 respectively. The temperature data was obtained from Tanzania Meteorological Authority (TMA). The description of

the data used included the geographical coordinates and altitude of the respective stations as provided in Table 1. The use of quality controlled data is a necessary prerequisite in the analysis of climate Indices [17] [18]. It provides an opportunity to identify suspicious data and any erroneous outlier that might influence the analysis, including daily values outside the user-predefined threshold. It helps to verify whether the reported data value is representative of what was intended to be measured, and has not been contaminated by unrelated factors. RClimdex and National Climate Monitoring Products (NCMP) software used in this study for quality control of the data, are often used to examine the data and identify those that fail pre-defined tests [7] [19].

RClimdex and NMCP softwares were used for the computation of the climate extreme Indices, adopting a similar approach as used by Expert Team on Climate Change Detection Indices (ETCCDI) and Expert Team on National Climate Monitoring Products (NCMPs) [18].

Detailed description of the methods and approach used in the computation of the climate indices using Rclimdex and NCMP software is shown in [7] [14] [16] and can also be accessed from [11] [20] [21]. The linear trends presented in this study (Figure 2 to Figure 7) are computed from the indices series using a Kendall’s tau based slope estimator [22]. This estimator is robust enough to detect of outliers in the series. It has been widely used to compute trends in hydro-meteorological series, as seen, for example, in [23] [24] [25] [26]. The significance of the trend was determined using the Mann-Kendall test and was estimated at a 5% significance level [27]. The trend was considered to be statistically significant if it was less than or equal to a level of 5% [28] [29] [27] [30].

Daily maximum and minimum temperature data are analyzed for the estimation of warm and cold day/night trends, and extreme warm and extreme cold day/night trends using eight (8) extreme temperature indices provided in Table 2. Moreover, the computation of the top five extreme temperatures that occurred during the period 1961-2019 and 1972-2019 for Arusha and Kilimanjaro respectively is carried out; and finally analysis of mean monthly temperature data for temperature anomalies on seasonal and annual timescales is implemented.

It has been observed that, for both Arusha and Kilimanjaro regions, the percentage of warm days and warm nights exhibit an increasing trend; while the percentage numbers of cold days and cold nights show a decreasing trend. Figure 2 and Figure 3 depict a trend of the percentage of number of warm and cold days/nights at Arusha airport and KIA station respectively. In both regions, it is observed that, the trend of cold days (TX10p) and nights (TN10p) is decreasing fast. The sharp and fast decrease in the number of cold days and cold nights may be translated into a warming of the minimum temperature across the region, which is consistent with national and regional temperature patterns [7]. Time series analysis of the number of cold days and cold nights over Arusha and Kilimanjaro stations indicates that the percentage number of both days and nights exhibits a decreasing trend, and is characterized by stronger inter-annual variability (Figure 2 and Figure 3). Comparatively, the percentage number of warm nights and warm days has been increasing, and the trend is more pronounced in the percentage number of warm nights, particularly in Arusha, indicating that the nights are warming faster (Figure 2). These findings are consistent with other nation wide analyses [7], and also compliant with the observed temperature extreme trends in different parts of the world as depicted in the IPCC Assessment Reports [1] [31] [4] and [7], which indicates that the number of cold days and nights have been decreasing, while the number of warm days and nights are increasing.

The trends of extreme warm days (TXx) and extreme warm nights (TNx) are observed to increase at both Arusha and Kilimanjaro meteorological stations (Figure 4 and Figure 5). Similar trend pattern is observed for extreme cold days and extreme cold nights, whereby warming of the extreme cold nights is more pronounced than warming of extreme cold days. This implies that the nights are getting warmer faster than the days.

The analysis of trends in seasonal and annual temperature anomalies is done for four seasons; January-February (JF), March to May (MAM), June to September (JJAS) and October to December (OND). It is well known that, during March to May (MAM) and October–December (OND) the study domains experience “long rainy” and “short rainy” seasons respectively. During January-February (JF), the regions experience a transition period which is relatively dry, while during June to September (JJAS) both regions experience relatively dry and cold weather conditions respectively.

It has been observed that in all seasons (JF, MAM, JJAS and OND) for both Arusha and Kilimanjaro, mean seasonal temperature anomalies are characterized by stronger inter-annual variability, and exhibits a well pronounced increase in the trend (Figures 6(a)-(e) to Figures 7(a)-(e)). For JF and JJAS seasonal temperature anomalies in both Arusha and Kilimanjaro, earlier decades

(1961-1981) have been characterized by negative temperature anomalies, while recent decades (1989-2019) have been characterized by positive temperature anomalies (Figure 6(a) and Figure 6(c)). However, MAM and OND seasons have been dominated by positive temperature anomalies (Figure 6(b) and Figure 6(d)). Mean annual temperature anomalies exhibit stronger inter-annual variability, whereby the earlier decades (1961-1981) have been dominated by negative temperature anomalies, while the last two decades have been dominated by positive temperature anomalies (Figure 6(e)).

For Kilimanjaro region, the JF, MAM, JJAS and OND have been dominated by positive temperature anomalies especially in the last two decades (2000-2020), while earlier decades (1971-1990) have been largely dominated by negative temperature anomalies. These results indicate the accelerated warming over the last two decades.

Monitoring and characterization of extreme temperature indices is very important and useful for climate change detection and for effective planning and designing of various infrastructures related development activities. The result indicates that, for both Arusha and Kilimanjaro regions, the top five highest daily

maximum temperatures have been observed in the last two decades, that is between 2000 and 2020 (Table 4 and Table 6). The highest ever-recorded daily maximum temperatures for Arusha region is 36.3˚C observed on 16^th February 2011, and for Kilimanjaro it is 38.6˚C observed on 22^nd February 2005. Table 3 and Table 5 present the lowest five daily minimum temperatures for Arusha

and Kilimanjaro respectively. Most of the lowest recorded minimum temperatures were observed before 2000 (Table 3 and Table 5). These results are consistent with global warming patterns as depicted in IPCC and WMO reports indicating that the last three decades have been successively warmer than any preceding decades since 1850.