Table 1 summarizes the heavyrain and the dry periods since the Christ^{[6]}. Immense efforts have been spent to detect current climate changes in North American, Europe, Russia and Australia, but only few such studies have been conducted in Jordan and neighboring countries. Cohen and Stunhill^{[11]} analyzed daily records of temperature, rainfall and global irradiance at three locations in the Jordan Valley. Their findings revealed significant decreasing trend in annual maximum mean temperatures, no significant trends in annual rainfall and an increasing (but inconsistent) trend in the annual mean minimum temperatures. Hasanean^{[12]} investigated the trends in air temperature series from eight meteorological stations in the East Mediterranean including Amman. He found a negative trend at a 95% confidence level. His findings showed a 2°C increasing trend in minimum temperatures and a 1°C decreasing trend in maximum temperatures.
Table 1: 
Dry and wet periods since the Christ^{[6]} 

Smadi and Zghoul^{[13]} examined changes, trends and fluctuations in the total rainfall and number of rainy days at Amman Airport Meteorological station in Jordan during the period 19222003. Their statistical analysis detected a sudden change and shift in the average total rainfall and annual number of rain days that occurred in 1957. AbuTaleb et al.^{[14]} looked at changes in annual and seasonal relative humidity variations in five stations around Jordan. Their analysis indicates an increasing trend in relative humidity at all different stations. They also noticed a major change point in the annual relative humidity in 1979. Above studies have come up with important findings, they pointed out that Jordan is indeed being affected by the global warming, although they have either focused on only one meteorological parameter at a time, or they were restricted to certain geographical locations, particularly Amman Airport station. In this study more meteorological parameters are considered and several trend tests are applied in order to have a comprehensive image about what is going on in real world in Jordan. This study aimed at studying climate parameters including precipitation, relative humidity and air temperature in order to find out if global climate changes and how global warming are impacting the climate in Jordan. We also examined some of the meteorological variables concerning climate change in Jordan other than exploring choices involved in the methodology in detecting temporal trends, changes and randomness. Types of trend hypothesis were discussed, i.e., step trend versus monotonic trend as well as the statistical category to employ: parametric versus nonparametric. Also, light was shed on data analysis so that best and legitimate results were achieved through choosing and using of a proper trend technique for that analysis from a wide variety of trend detection methods. MATERIALS AND METHODS
Data collection: In studying climate change we have needed data about
many climate elements like relative humidity, rainfall averages, minimum and
maximum air temperature, studying historic climate of the area of Middle East
and Jordan and reviewing previous studies that predicted so many changes on
climate in Jordan. Available selected meteorological data were retrieved from
six different observation weather stations with medium to long term annual records.
Records were ranging between 3083 years. The stations are located in different
geographical and biomes in Jordan. These stations comply by the requirement
of the World Metrological Organization regulations. We assume that the data
sets used in this study were collected and reported in a consistent and reliable
manner. These data were obtained from Jordanian Department of Meteorology.
Analysis of databases was carried out using different statistical methods depending on the distribution pattern of characteristics and other specific features. Statistical analysis: Statistics can be a useful tool for analyzing and drawing meaningful conclusions about characteristics associated with the meteorological data set used in this study. We think that the strict standardization of methods selected in this study is required for proper application of statistical techniques and to proper interpretation of these data for trends and change. In this manuscript we examine several simple statistical approaches both parametric and nonparametric to detect significant, providing evidence, not proof of trend, changes, randomness and normality. Trend is a monotonic or step change in the level of the time series, i.e., a general increase or decrease in observed values of the studied meteorological random variables over time. It is preferable to think of trend tests as a quantitative basis for deciding whether apparent trends are real^{[15]}. Keeping in mind the advantages and drawbacks of both parametric and nonparametric tests, an adequate approach to analyze data is to use the appropriate tests, regardless if they are parametric or nonparametric. Nonparametric tests (distribution free) make no prior assumption(s) for probability distribution which does not depend upon the form of the underlying distribution of the null hypothesis that meant for testing certain hypothesis. These tests tend to ignore the magnitude of the observations in favor of the relative values or ranks of the data. They are preferable tests when implicit assumptions about the data are not met which is sometimes the case in real applications. In addition, these tests tend to be simple in design, easy to understand, used to model both liner and nonlinear relations and used for evenly and unevenly spaced observations. Most nonparametric methods can work with short series (5 or more observations) and handle unequally spaced, ties and missing data. These methods are convenient to use in investigations of multiple data sets because exhaustive checking of distributional assumptions is not required. The tests are designed for use with data expressed in a nominal or ordinal scale. Furthermore, many difficulties which appear to make the data unusable with a parametric technique can in fact, be overcome using nonparametric ones. Therefore, fewer assumptions have to be made about the data. On the other hand, parametric tests assume that the time series data and the errors (deviations from the trend) follow a normal distribution with homogeneous variance. In general, these tests are more powerful for given n when the variable is normally distributed, but much less powerful when it is not, compared with the nonparametric tests^{[16]}. Testing data to ensure that these assumptions are met can become tedious especially when many trends need to be examined. These tests are useful as they also quantify the change in mean or gradient of trend in the data. The dividing line between parametric and nonparametric tests is not a sharp one^{[17]}. All tests used in this research are considered confirmatory data analysis tools which designed for specifically testing certain hypotheses. Appropriate parametric tests can overcome some difficulties of series instability that encountered with the used data. Although meteorological time series are occasionally not normally distributed and contain properties that are undesirable for use with parametric tests^{[18]}, SkewnessKurtosis Normality test was conducted in this study. Unlike other normality tests, this specific test is not affected by ties. In this case, we resort to use distribution free methods rather than parametric ones. Skewnesskurtosis normality test was implemented in this study to test all used data for normality. Null hypothesis of normality test states that the used data are normally distributed. Passing the normality test only allows stating no significant departure from normality is found. It is widely recognized that nonparametric procedures can have significantly higher power and efficiency than parametric procedures in cases where there is a substantial departure from normality and the sample size is large which is commonly encountered for many climate change variables^{[19]}. Even for small departures from normality, the performance of the nonparametric procedures is similar to or better than that for the parametric procedures. Correlating time series afflicted with a trend, or other serial dependencies, may lead to false correlations that do not represent real mechanistic relationships, but occur only owing to the presence of instationarities^{[20]}. In addition, serial correlation affects the test’s ability to assess the significance of a trend which might lead to or increase the possibility of rejection of the null hypothesis of no trend while it is true^{[21]}. Therefore, significance of serial correlation of the time series was evaluated prior to conducting trend analysis. For detecting trend, both monotonic and step change in the mean or median of the tested data, the critical test statistic value, at α/2 is used (twosided tail) along with 90, 95 and 99 percent confidence level for evaluating the test hypothesis. Therefore, three parametric and three nonparametric tests were used to determine the possible existence of statistically significant trends, step change in mean/median and difference in mean/median in two different data periods of the tested meteorological parameters assuming a 10, 5 and 1% probability levels. Although there are other statistical tests, such as Spearman rank statistic and the test by Cramer, the MannKendall rank statistic is considered the most appropriate^{[22]} for the analysis of climatic changes in meteorological time series especially detection of a climatic discontinuity. The following are brief description of the statistical tests used in this study. Mannkendall test: This popular nonparametric test is used to test if the values of the studied time series are going up or down in a manner not due to chance alone. No assumption of normality is required but there must be no serial correlation in the Y values for the resulting pvalues to be valid. The test is expected to be less affected by the outliers since its statistic is not directly based on the values of the random variable but on the sign of divergences^{[23,24]}. The test is widely used in the analysis of climate changes. The method is described briefly by^{[25]}. This test is directly analogous to regression, where the test for significance of the correlation coefficient r is also the significance test for a simple linear regression. Moreover, MannKendall test is an appealing and more costeffective alternative to the parametric regression method. We are giving time series the symbol x_{i}, a sequence of measurement over time as x_{1}, x_{2},….., x_{n}. Each data point x_{i} is used as a reference point and compared with other data points such that:
A null hypothesis, H_{o} (no trend in the data), was tested such that the data come from a population where the random variables are independent and identically distributed^{[24]}. Under the H_{o}, the MannKendall’s test statistic (S) is:
For situation where there may be ties in the series, S is asymptotically normally distributed with zero mean and variance equaling^{[26]}:
where, p is the number of tie values in the data set. t_{j }is the
number of data points in the j_{th} tied values. Consequently, one can
check whether or not an upward or downward trend is significantly different
from zero. If Sis significantly different than zero, the alternative hypothesis,
H_{1}, is accepted. The alternative hypothesis, H_{1}, states
that the data follow a monotonic trend with time taking into account that no
reversals in direction^{[27]}. The zstatistic is therefore (critical
test statistic values for various significance levels can be obtained from normal
probability tables):
A positive value of the test statistic S indicates that there is an increasing trend and vice versa.
Linear regression test: Regression statistics provide an estimate of
degree of association based on leastsquares regression analysis of potentially
false significance. Simple linear regression of Y(meteorological variable in
this study) on time (x in year) is essentially a trend test. Linear trend means
that rate (magnitude of change over time) of increase or decrease is constant.
The null hypothesis is that there is no significant trend or increase/decrease
of the studied parameters with time, i.e., the gradient is zero. The test statistic
S follows a studentt distribution with n2 degrees of freedom under the null
hypothesis. If the gradient (trend) is positive or negative, then null hypothesis
will be rejected.
This parametric test assumes that the data are normally distributed and that the errors (deviations from the trend) are independent and follows the same normal distribution with zero mean. Also, the test can be used with both evenly or unevenly spaced observations. Moreover, the test is not very sensitive to small deviations but very sensitive to outliers, abnormally high or low values at the start or end of a series will influence on the estimate of gradient^{[28]}. The regression gradient is estimated by:
and
The test statistic S is:
Where
CUSUM test: This is a distribution free technique tests whether the means in two parts of a time series are different (for an unknown change in time). The CUmulative SUM test is based on the cumulative sums charts to detect systematic changes over time in one or more measured variables^{[29]}. The null hypothesis, H_{0}, for this test is no step jump or change in the mean/median. This means that the data collected before a specific time is from a clearly different population than the data collected after that time. The test statistic of the studied time series is:
where, k = 1, 2,….., n
and
x_{median} is the median value of xi time series data set. The distribution of test statistic, V_{k}, follows the KolmogorovSmirnov twosample statistic. A negative value of V_{k} indicates that the latter part of the record has a higher mean than the earlier part and vice versa. Note that the separation period can be determined in this test^{[27]}. Worsley likelihood ratio test: This is a parametric test assumes that the data are normally distributed. The likelihood ratio Technique detects whether the means in two parts of a record are different (for an unknown time of change). The deviation from the means are calculated as^{[30]}:
where, k = 1, 2,….n Depending on their position in the time series the test weights the values of S_{k} such that:
The test statistic W is:
where
A negative value of W indicates that the latter part of the record has a higher mean than the earlier part and vice versa^{[27]}. The null hypothesis H_{o} here is no step change in the mean between two data periods. Both CUSUM and this test are used in this study when the meteorological records being analyzed are naturally broken into two distinct periods. Ranksum test: This nonparametric technique tests whether the medians in two different periods are different. In this test the data ranked from 1 (smallest) to N (largest) using an average of ranks when ties exist in the data set. S statistic is the sum of ranks of observations in the n group (smaller group) and m, the larger group. The null hypothesis H_{o} of this test is no difference in the median between two data periods. The theoretical mean and standard deviation of S under the null hypothesis is:
The test statistic, Z_{rs} is approximately normally distributed and is equal to^{[31]}:
The rank sum for related samples represents nonparametric alternatives to the ttest. The main limit of rank sum test is that it was originally designed for detecting single pointchanges. Differently, the MannKendall and CUSUM tests as examples are nonparametric tests particularly suitable in sequential analysis^{[32]}. Considerations of power and efficiency were taken when decided to choose these tests. Student’s ttest: The test is one of the simplest parametric techniques for testing a signal change of the mean value on the basis of a difference between sample means^{[33]}. The test assumes that the data are normally distributed which tests whether the means in two different periods of the studied time series are different. The null hypothesis H_{o} states that no difference in the mean between two data periods:
are the means of the first and second period, respectively. S is the standard
deviation of the sample and m and n are the number of observations in the first
and second periods respectively.
Rank difference test: This test uses the ranks of the data rather than their raw values to calculate the statistic. This is done by replacing the n time series values by their relative ranks starting at 1 for the lowest up to n. The statistic U is the sum of the absolute rank differences between successive ranks as shown in the following formula^{[34]}:
The null hypothesis H_{o} for this test is that the data come from
a random process. For large n, the statistic U is normally distributed with
the following mean and variance:
μ = (n+1) (n1)/3
σ = (n2) (n+1)(4n7)/90 
To find the z statistic, the following formula was used^{[31]}:
Autocorrelation test: Autocorrelation or serial correlation can be defined as a measurement which at one time period reflects the level of the tested variable at a previous time period. In other words, the extra data does not provide any new information. Under realistic stochastic processes (exhibiting seasonality, skewness and serial correlation), it is robust in comparison to parametric alternatives. The null hypothesis H_{o} for this test is that the data come from a random process. In general, most, but not all, timeseries data with time steps shorter than the annual time step are serially correlated. This parametric test is used to test if the time series data used in this study is serially correlated. An autocorrelation function at lag one is suggested by many researchers as a parametric test for finding trends which is more powerful than those provided by the MannKendall statistic for detecting trends especially when used for discovering purely stochastic trends^{[31]}. The lag one autocorrelation coefficient is calculated as:
The mean and the variance given that the data come from a random process are:
Then the z statistic is:
RESULTS
Outputs of the eight statistical tests are summarized in Table 26. As could be inferred from Table 2,
there is no sound evidence that there are increasing or decreasing trends in
the relative humidity records at Baqoura, Amman Airport and Queen Aliaa' Airport.
At Irbid, Ma'an and Dir Alla, the situation is different. Table
2 shows that the relative humidity has increased in those stations. Precipitation
records do not show clear trends at all stations (Table 3).
This is supported by the findings of the study conducted by^{[35]}.
In fact by referring to Table 1^{[6]}, we learn that
precipitation has varied for long periods of time.
Table 46 summarize the outputs of the
statistical tests applied on temperature records including maximum annual temperature,
minimum annual temperature and annual temperature range. Maximum temperature
does show a clear increasing trend (Table 4) at some examined
areas, Dir Alla, Baqoura and Queen Aliaa’ Airport. Minimum temperature shows
at five of those areas (Table 5). Significant downward trend
in maximum air temperature was shown at Baqoura, Dir Alla and Queen Aliaa' Airport
areas. This finding comes in a accordance of^{[27]} study. Trend was
detected in minimum air temperature at all areas except Baqoura. At the Irbid
Station, the minimum mean annual temperature time series shows an increasing
trend.
Table 2: 
Trends in mean annual relative humidity along with percent
confidence level at all stations 

^{a}: Data show a statistically significant step jump
at specified confidence level (data in later years are higher than earlier
years); ^{b}: Mean of 19771994 is less than mean 19942005 at α<0.01;
^{c}: Median of 19771990 is smaller than 19912005 at α<0.1;
NS: Nonsignificant; ^{d}: Mean of 19771987 is smaller than 19872005
at α<0.01; ^{e}: Median of 19932005 is greater than 19771992
at α<0.01; ^{f}: Mean of 19771997 is smaller than 19972005
at α<0.01; ^{+: }Increasing trend at specified significance
level 
Table 3: 
Trends in mean precipitation along with percent confidence
level at all stations 

NS: Nonsignificant at least α = 0.1; ^{a}: Data
show no statistically significant step jump at α = 0.1; ^{b}:
Data show a statistically significant step jump at α(pvalue or sign
level)<0.05; : Decreasing trend at specified confidence level 
Table 4: 
Trends in mean maximum temperature along with percent confidence
level at all stations 

NS: Nonsignificant at least α = 0.1; ^{+}: Increasing
trend at specified confidence level 
Table 5: 
Trends in mean minimum temperature along with percent confidence
level at all stations 

NS: Nonsignificant at least α = 0.1; ^{+}: Increasing
trend at specified confidence level 
Table 6: 
Trends in annual temperature range along with percent confidence
level at all stations 

NS: Nonsignificant at least α = 0.1; ^{+}: Increasing
trend at specified confidence level; : Decreasing trend at specified confidence
level 
In Amman Airport station where the longest meteorological records exist, a clear positive (increasing) trend in the minimum mean annual temperature is observed using both the Mannkendall and simple linear regression tests at 99% significance level. Moreover, Annual temperature range is showing a clear decreasing trend (Table 6). DISCUSSION Table 2 shows that the relative humidity has increased in those stations, however, this increase is not sufficient to conclude a climate change. There are other factors that contribute to the increasing moisture content in the atmosphere including water bodies and green lands. The last decades have witnessed extensive efforts to increase areas of irrigated cultivated lands in the Jordan valley through constructing a long canal to carry water from Yarmouk River to southern parts of the valley. New cultivated areas have flourished in several parts of the valley including Dir Alla, which used to be a barren dry land. Last decades have also witnessed constructing several dams, artificial lakes and wastewater treatment plants. Irbid governorate is hosting several water projects including Wadi Araba Dam, Alwihda Dam and a wetland for wastewater treatment at Rumtha, 15 km to the northeast of Irbid City. Furthermore, the people of Irbid have abandoned cultivating traditional crops, basically grains. Irbid meadows hosted flourished wheat and barley agriculture in the past. Grains were usually harvested around the end of spring (April and May) leaving behind thousands acres of dry soil. Currently, Irbidies are active in cultivating Olive trees on their lands including areas that were never cultivated in the past. Olive trees are evergreen trees, therefore they act as natural humidifiers through their transpiration and exhaling processes. Ma'an has also witnessed an increase in agricultural activities in its surrounding due to the availability of fresh ground water. Thousands of wells have been drilled in the past few years to provide huge areas with water necessary for cultivating potato, watermelon and fruit orchards in both sides of Wadi Araba, which is located upwind from Ma'an. Other mega projects aiming to flourish agriculture are also available in the eastern and southern parts of Ma'an governorate including the ones in the northern parts of Saudi Arabia. Table 6 shows that precipitation has varied for long periods of time. Since detecting increasing or decreasing trends in precipitation requires data from hundreds of years and our data was collected in less than a 100 year span, it might be difficult to detect trends in such a short time interval. We anticipate that in the next 50 years it might be possible to have strong evidences of decreasing precipitation trends. Annual temperature range is showing a clear decreasing trend (Table 6), which means that the Earth Atmosphere has become more efficient in preserving its energy for longer durations. Dir Alla is one of the stations located in the Jordan Valley, an area characterized with relatively high air temperature: very hot in summer and warm in winter. Observed mean daily minimum and maximum temperature ranges between 9 and 39°C. The Jordan Valley, a vegetable basket of Europe and winter resort is considered a natural greenhouse which makes it one of the climatically unique spots in the region. The longest meteorological records at Amman station, shows a clear positive (increasing) trend in the minimum mean annual temperature. This might confirm the fact of greenhouse effect gases that rises minimum temperatures by warming gases. At the Irbid Station, the minimum mean annual temperature time series shows an increasing trend. This trend in temperatures will increase the demand for agricultural and urban domestic water use and encourage eutrophication in streams and lakes that affect freshwater ecosystem. Moreover, in arid and semiarid regions such as Jordan, changes in temperature will increase the evapotranspiration, reducing optimality of plants, water stress which leads to a significant reduction of food production. This increases risk of desertification in those areas not to mention many biophysical and social impacts. Furthermore, rising temperatures increases energy consumption for cooling. Mortality rates would also rise during the winter months and hot summers. Individuals in developed countries will be vulnerable to heat which increases by virtue of circulatory problems related to vascular and heart disease^{[36]} However, longterm time series, which are not available in most developing countries, are required to define a definite climatic trend. CONCLUSION In this study we attempted to detect and quantify evidences of climatic change in Jordan. Data from six meteorological stations at Irbid City, Baqoura, Dir Alla, Ma'an, Amman Airport and Queen Aliaa' Airport were examined using several tests in order to detect any changes in air temperature, relative humidity and precipitation over last decades. our findings revealed that minimum air temperature has increased since the seventies of the last century. This increase indicates a slight change in regional climate. Rainfall records have revealed that precipitation has been fluctuating at all stations and no statistical trends of increase or decrease in the annual precipitations indicating climatic change were detected. Relative humidity records do not indicate clear trends at Baqoura, Amman Airport and Queen Aliaa' Airport. However, it increased at the remaining three stations. The relative humidity increases at Irbid, Dir Alla and Ma'an may not be considered a consequence of global warming or an indication of climate change because it was accompanied by rapid growth in cultivated areas and mega hydroprojects. Annual maximum air temperature records do not show clear trends, but annual minimum temperatures have increased while the annual range of temperature have decreased. Decreasing temperature range proves that the earth is becoming more efficient in trapping terrestrial infrared radiation, which is responsible of the global warming. " target="_blank">View Fulltext
 Related
Articles  Back
