Analysis of long-term climatic changes at Hodeidah-Yemen during the period between 1984 and 2019

Yemen is one of the Arab country that is vulnerable to climate changes, and this is clear from the indicators of impact on water resources, coastal zone environments, etc. This work focuses on studying the climatic variability at Hodeidah city-Yemen during the period between 1984 and 2019. This study aimed to characterize trends in mean monthly, seasonal and annual temperature. To attain these objectives the collected data were analyzed using both parametric (linear regression) and non-parametric (Mann–Kendall, Spearman and Sen's slope estimator tests) methods to detect the trend and the magnitudes of rates of changes of temperature over time. Analysis of data indicates clear climatic uctuations of temperature. The annual means of temperature during the period of study were varied between 26.9°C and 30.1°C. The warmest years were observed during the more recent years of the study period ( 2005 to 2018). The increasing rate of annual temperature is about + 0.075°C /year, + 0.37°C/5year, + 0.75°C/decade ,+2.53°C, over the whole period of study(1985 to 2019), + 3.7°C/50 year and increase to + 4.85°C in 2050. On a monthly timescale, there are similar magnitudes of rates of change from December to September with highest rates in October and November. The results also showed that most months and seasons have signicant positive trends in temperature and (Z-α/2) values of the MK Test > 1.96 and positive value of Sen’s slope estimator indicates signicant an increasing trend towards warmer years. Anomalies of temperature conrm signicant increasing trends towards warmer years (2000s to 2019).

All previous reports and studies referred to the impact of climate change in Yemen, while the long-term analysis and study of weather elements as an indicator to prove the climate changes are almost rare due to unavailability of historical data. Therefore, the major objective of this study is to investigate the variability of temperature, detection of signi cant trends and evaluate the increase of it an addition to showing the anomalies in the monthly, seasonal and annual climatic conditions in the Hodeidah city over the last 35 years. Studies uctuation of temperature is as an indicator of climate changes in one governorate of Yemen during the period between 1984 and 2019, to helping decision makers, interested in human services and development to presenting projects that limited the problems with it.

Data And Methods Of Analysis
Monthly temperature data used in this study was provided by Meteoblue AG -Switzerland as hourly time-series for 35 years from 1985 to 2019, this historical data are simulation for Hodeidah Airport Station, with high precision and more precise than the observational data from the airport station, which is located at more than 10 km away from the Hodeidah coast [www.meteoblue.com]. Weather data and model simulations for more than 30 years (from 1985 onwards) offers via Meteoblue-Data.org for climate information on it's website. This data gives a good view of the expected weather and climate patterns (temperature, rainfall, sunshine, wind speed and direction). A spatial resolution of the simulated weather data has about 30 km. In temperature analysis, a 5-year running average was used. The moving average or running mean is used in looking at noisy sequence data of the series (Sneyers 1992). Moving averages used to smooth out short-term uctuations in a time series and still preserving the slowly varying trend. This technique applied to long-term time series to remove the ne-grained change between time steps. To characterize climate oscillation, we prepare air temperature data done by calculating monthly mean, seasons mean and annual mean of daily historical weather data. The seasons were determined according (Douabul and Haddad 1999), as winter (December-March; spring (April-May); summer (June-September); and autumn (October-November).
A temperature anomaly is the difference from an average, or baseline temperature. The baseline temperature is typically computed by averaging 30 or more years of temperature data (the base period of data study 1985 to 2019). When the observed temperature is higher than the baseline temperature, the positive anomaly will be occur, the opposite of that shows negative anomalies. Climate oscillation can be described by linear trend analysis of that. In time series analysis, linear regression (parametric method) to identifying linear trend to obtain the slope of hydro-meteorological variables on time. Based on linear regression the positive/negative values of the slope show increasing/decreasing trend. The advantage of the linear regression method provides a measure of signi cance and gives the magnitude of the rate of change based on the hypothesis test on the slope (Hirsch et al. 1991).
In this study, both parametric (linear regression) and non-parametric (Mann-Kendall ,Spearman and Sen's slope estimator tests) methods were used to detect the temperature trends, so the simplicity of the parametric method is distinguished from others (Mosmann et al. 2004).
The Mann-Kendall test is a non-parametric test for identifying trends in time series data, various researchers considered this test (Burn 1994;Douglas et al. 2002;Yue and Hashino 2003;Burn et al. 2004;Lindström and Bergström 2004;Jain and Kumar 2012;Panda and Sahu 2019) one of the best methods and used for analysis and ascertains statistical signi cance by hypothesis test of hydrological variables. The Mk trend test is used to perceive statistically signi cant decreasing or increasing trend in long term temporal data. According to this test based on two hypotheses; one is null (H 0 ) assumes that there is no trend (the data are independent and randomly ordered) and the other is the alternate (H 1 ) hypothesis elucidate signi cant rising or declining trend in temperature data. Positive/negative values of Z MK indicate increasing/decreasing trends in the time series. The null hypothesis is rejected when |Z MK |>Z 1−α/2 , and a signi cant trend exists in the time series. Z 1−α/2 is the critical value of Z from the standard normal table, for 5% signi cant level the value of Z −α/2 is 1.96 (Shadmani et al. 2012). On the basis of 5% signi cance level, if p Value is ≤ α = 0.05, then the alternative hypothesis is accepted, which signi es the presence of trend in the data, the absence of trend in the data showed if the p Value is ≥ α = 0.05 then H 0 will be accepted that denotes. Sen's slope estimator (Sen 1968) using a nonparametric method can be used to determine the magnitude of a trend of temperature in a time series. The true slope of an existing trend estimated to an amount of change per year. An upward or increasing trend indicated by positive value of Sen's slope and a negative value gives a downward or decreasing trend in the time series (Sen 1968;Da Silva et al. 2015;Dawood 2017;Tabari et al. 2011)". Sen's slope is considered better to detect the linear relationship as it is not affected by outlier in the data (Gilbert 1987). For this test have some advantage, the data need not conform to any particular distribution, but due to inhomogeneous time series the other advantage of the Sen's slope test is its low sensitivity to abrupt breaks (Jaagus 2006). (Mann 1945) & (Kendall 1956) used this test and subsequently derived the test statistic distribution.

Results And Discussions
To study climate oscillation during long-term period of this case study, we are calculating monthly mean, seasonal mean, annual mean and anomaly mean temperature of daily historical weather data. (Fig. 2:4) shows time series of monthly, seasonal and annual mean of temperature for Hodeidah city during the period 1985-2019 and trend analysis using a linear regression model. The slope of the regression line in gures shows the rate of temperature change, also a 5-year running average was added to the time series to remove the ne-grained variation between steps. Descriptive statistics of daily temperature are calculated and given in (Table.1) including: mean, standard error, median, standard deviation (SD), sample variance, kurtosis, skewness, range, minimum and maximum.
An estimated trend was shown in (Table.2) also presented via a curve line in (Fig. 5) and show the variability of trend values during the months, seasons and annual of the study period. The results of parametric (Linear model) and non-parametric models are shown in (Table.3) were used to assess and test the temperature oscillation of monthly, seasonal and annual means. Parametric model is done using the least squares method, the magnitude of the trends was derived from the slope of the regression trend line and Sen's Slope Estimator, while the statistical signi cance was determined by the Mann-Kendall and Spearman correlation coe cient test Sneyers (1990). Temperature anomalies relative to the base period of data study during 1985 to 2019 were estimated and used trend analysis of monthly, seasonal and annual mean anomaly shown in ( Fig. 6:8) to determine warm and cold period. 3.2 Statistical trend analysis using the parametric method

Statistical tests using Non-parametric method
The non-parametric method using (Mann-Kendall, Spearman and Sen's slope estimator) tests, these statistical tests have been widely used to demonstrate the signi cance of trends; it shows either increasing or decreasing trends. Moreover, estimate the magnitude of temperature can be estimated using a linear regression model and Sen's slope that mean, how much increase or decrease per year and the correlation coe cients and equations were calculated and presented in (Table.3). On the basis of 5% signi cance level for this test result p (2-tailed) with Spearman test and p value in linear regression model are less than the signi cance level α = 0.05 (Burn 1994), then the alternative hypothesis is accepted which signi es the presence of trend in the time series, H o , (there is no trend), hence, the hypothesis is not accepted and rejecting H o .
The results revealed the statistically signi cant increasing trends for all months, seasons and annual (at the 5% signi cant level the value of Z −α/2 is 1.96 (Shadmani et al. 2012), the most values of Z −α/2 values of the MK Test > 1.96, revealed an increasing trend in temperature. The Mann-Kendall and Spearman tests con rmed that signi cant positive trend with 95% con dence limit using a linear regression model except April, June and October, which occurred no trend with Mann-Kendall and positive trend with other test. The magnitude of rates of annual temperature was calculated using Sen's slope. A positive value of (Sen's slope = 0.076) of the mean annual indicates an upward or increasing trend towards warmer years, these values con rmed that estimated by linear regression model. These results of non-parametric method has shown in (Table.2 & 3). The statistically signi cant increasing trends during spring, summer and autumn led to increase in mean temperature, positive trend slope and positive values of anomalies starting from April to November, therefor the warm period appears to be longest time period. These results can be considered to be in agreement with those found in the WMO report (WMO 2011) on the status of the global climate in 2010, global average temperature has varied in different time scales ranging from a few years to several decades. According to the reports, Yemen is vulnerable to climate change, based on that uctuation: water resources, agriculture, and coastal zones will vulnerable to impacts of climate change caused an increased water scarcity, reduced water quality, increase the frequency and magnitude of disasters and an increase of the environmental problems. This study investigated monthly, seasonal and annual climatic uctuation in Hodeidah city based on daily mean air temperature during the period between 1984 and 2019.
The results show the monthly mean of temperature changes between highest value is 32.5°C in July and August and lowest value is 23.7°C in January. year and increase + 4.85°C in 2050. On a monthly timescale, the mean monthly temperature has a similar an increase rate from December to September with the greatest rate in October and November, an increase in the length of warm period is attributed to the positive trend in summer and greatest an increase rate in autumn. A similar pattern of anomalies air temperature from January to December and four seasons, an accelerated warming trend occurred after 2010. Moreover, the values of anomalies were larger in autumn than other three seasons. Based on the anomalies of temperature a cool period was evident from in the time series, and the results revealed a statistically signi cant increasing trend for all months, seasons and annual. This results of this study are consistent with other researches that con rm a general warming trend and an increase in the occurrence of events (Al-jibly 2018), which referred to a scenarios of climate change for future of Yemen for 2050 which get warmer, more so in the cold winter months with annual mean temperature will rise to an average of 2°C in 2050, either to Arab region extent (RICCAR 2013) clari ed that in mid-century temperatures increase to 1.7°C-2.6°C, while at the end of the 21st century, it will increase to 3.2°C-4.8°C, and a global extent (WMO 2011), it's report referred to global average temperature have varied on different time scales ranging from a few years to several decades.
The most values of Z-α/2 values of the MK Test > 1.96, revealed an increasing trend in temperature and con rmed that signi cant positive trend with 95% con dence limit using a linear regression model. A positive value of (Sen's slope = 0.076) of the mean annual indicates an increasing trend towards warmer years con rmed with an estimated by linear regression model. On a monthly time scale an increase of warm period starts from April to November noticed based on an increase in mean temperature, positive trend slope and positive values of anomalies during spring, summer and autumn. Therefore, signi cant increase trend of air temperature for this case study may be due to increased concentration of CO2 that leads to global warming caused via vehicles and thousands of motorcycles in city, the spread a lot of factories around the city, passive activity for human, and the uprooting of large numbers of trees to be used for fuel during the war, and the scarcity of green spaces around the city due to the lack of rain. Finally an increasing trend of temperature indicates the region is heading towards a warmer climate and vulnerable to climate change impact.

Recommendations
The results of the analysis indicate that in the study area, as in many other countries, the effects of climate change will be signi cant. Therefore, it is recommended to those concerned and decision-makers to take many actions in reducing exposure to the risks of climate uctuations, such as building capacities, knowledge and relevant institutions to face crises from the consequences of the future climate, most important recommendations are provide climate data and the data about the impact of climate change on water resources; Agriculture; coastal areas etc.