The difficulty of monitoring hydrometeorological trends remains relevant because of the importance of climate change. In this study, innovative polygon trend analysis (IPTA), new graphical methods, have recently been presented, Mann-Kendall and innovative trend analysis (ITA) with significance tests are employed at monthly and annual hydrometeorological data for Ankara province in Turkey. Finally, sequential Mann-Kendall (SQ-MK), CUSUM, and standard normal homogeneity test (SNHT) are applied to detect any abrupt changes in annual time series. The results indicate that ITA with significance test, IPTA, and MK capture precipitation trends in 83, 94, and 0.2% of all months (3 stations × 12 months), temperature trends in 86, 75, and 22% of all months, relative humidity trends in 80, 80, and 30% of all months, and evapotranspiration series in 91, 80, and 47% of all months. These findings suggest that the ITA with significance test and IPTA are more sensitive than the MK test, with the precipitation series being the most sensitive and the evapotranspiration series being the least sensitive. SQ-MK, CUSUM, and SNHT tests on station 17664 were successful in detecting a change in annual evapotranspiration in 2005. The annual total evaporation series have been rising since 2005, according to the SQ-MK test.

  • Graphical and traditional trend tests were utilized at monthly and annual hydrometeorological data in the context of climate change.

  • Graphical tests are most sensitive than the traditional test with the precipitation series being the most sensitive.

  • Evaporation in February, July, and September indicate an increasing trend, and maximum temperature in February and July observed an increasing trend.

Graphical Abstract

Graphical Abstract
Graphical Abstract

Climate change, rapid urbanization and industrialization, land degradation, and growing populations are all factors that have a substantial impact on numerous hydrometeorological variables. Variability in rainfall, relative humidity, evapotranspiration, and temperature can have a negative impact on the occurrences of natural hazards, particularly floods and droughts (Lehner et al. 2018; Arab Amiri & Gocić 2021; Esit et al. 2021). By 2100, climate change is expected to increase a 0.5–1.0 m rises in mean sea level and a 1.0–5.7 °C increase in mean temperature as well as higher rainfall, wind speed, and more frequent extreme events in some areas, according to the Intergovernmental Panel on Climate Change (Allan 2021). Climate change has a variety of substantial effects in humid, semi-arid, and dry regions, according to IPCC studies (IPCC 2013). Climate change influences the hydrological cycle, water availability, quality, and services, and observations and records from all continents and oceans indicate unexpected behaviors. These unexpected hydrological cycle behaviors may be seen as either trends or sudden increases in average and variation values. They also have an impact on a variety of human activities, including groundwater recharge, water supply, agricultural activities, hydroelectric power generation, and irrigation methods (Şen 2012, 2014; Şan et al. 2021). As a result, interested institutions and organizations should investigate changes in hydrometeorological information to obtain a preliminary understanding of climate change.

One of the most useful methods for observing the effects of climate change on hydrometeorological variables is trend analysis. Most of the hydrometeorological variables are stochastic and complex (Hırca et al. 2022). While monitoring their trends can provide a useful perspective on water resources and meteorological sciences, the intrinsic characteristics of hydroclimatic variables require new technologies and methods for trend assessment (Şen 2020). The use of non-parametric tests in statistical analysis is crucial. They have been commonly employed for the analysis of hydrometeorological time series for the past two decades. Mann-Kendall (MK) test, Spearman's Rho (SR) test, regression analysis, and wavelet analysis are all frequently suggested as classic mathematical methods (Malik et al. 2020; Ivanovski et al. 2021; Yuce & Esit 2021; Shokrian et al. 2022).

The approaches for trend analysis described above, on the other hand, are pure statistical methods that do not allow for the detection of trends in low, medium, and high values in a single computation procedure. The majority of non-parametric tests can be worthless in some cases without graphical and exploratory data analyses (Cengiz et al. 2020). Onyutha (2016) proved that using a combination of graphical and statistical trend testing methods provides more useful and influential information than using only statistical methods. The majority of classical trend analysis methods show monotone trends across the whole time series (Yue et al. 2002; Shadmani et al. 2012; Eris et al. 2019; Isensee et al. 2022). However, it does not demonstrate any periodic variations (such as monthly). Hence, trend characteristics are critical when illustrating seasonal trend behavior. Because of anthropogenic effects in the atmospheric and hydrologic environments, as well as the resulting climate change, the stationarity principle is no longer applicable (Mohorji et al. 2017).

Hydrometeorological parameters (precipitation, temperature, relative humidity, and evapotranspiration) variation have a direct effect on agriculture, irrigation, and water resources system (Abdi et al. 2016; Cetin et al. 2018; Yuce et al. 2019). In particular, the change in precipitation is important in the hydrological cycle, as it has effects on both the occurrence of drought and floods. To overcome the above restrictions, innovative trend analysis (ITA) introduced by Şen (2012), was recently used to detect trends in hydrological and meteorological data in various parts of the world (Mallick et al. 2021; Singh et al. 2021; Akçay et al. 2022; Gumus et al. 2022). Although studies have demonstrated that sample size, distribution shape, and serial correlation influence the ITA approach, its benefits can be implemented in a variety of aspects (Serinaldi et al. 2020). In comparison to the MK and SR tests, Kisi (2015) discovered that the graphs created by the ITA approach can better demonstrate hidden trends in pan evaporation. In comparison to other statistical tests, the strength of ITA is its high performance of graphical visualization in identifying hidden trends in hydrometeorological time series (Cengiz et al. 2020). The ITA method is not significant if the data do not fit a normal distribution, the data length is small, or the data are serially dependent (Şen 2012, 2017). The method has been widely utilized to detect trends (Yilmaz 2019; Caloiero 2020). A novel method also proposed by Şen et al. (2019), the Innovative Polygonal Trend Analysis (IPTA), which can identify small-scale time series (daily, weekly, monthly, etc.) as well as determine trend transitions between successive sections of two equal segments of the original hydrometeorological time series, resulting in the polygon trend. A few studies were conducted to determine the applicability of IPTA in practice (Ceribasi & Ceyhunlu 2020; Achite et al. 2021; Harka et al. 2021; Ahmed et al. 2022). However, more research is needed to see if IPTA can be used to analyze hydrometeorological data and if it may be a suitable alternative trend analysis approach, such as the MK test.

The degree and slope of trend transitions between consecutive segments (e.g., months) can be evaluated using the IPTA method, and information on the magnitude and slope of the trend transitions between consecutive segments (e.g., months) can be obtained. This approach can also be used to observe the one-year behavior of time series. Internal variability that is time-dependent might be assessed in this way (Akçay et al. 2022). This approach was applied to monthly precipitation and flow data of selected stations in various parts of the world by Şen et al. (2019). The findings revealed many increasing/decreasing trends, and the slopes and lengths of these trends were determined. Furthermore, the monthly precipitation is studied using the IPTA graphical approach, which provides a clear and intuitive knowledge of precipitation fluctuations throughout the year, allowing for active and efficient response to soil erosion and vegetation degradation. Yenice & Yaqub (2022) evaluated trend analysis of temperature data by applying the IPTA method at six stations in Turkey. Şan et al. (2021) compared ITA, IPTA, and MK trend test for rainfall data in Vietnam. Their results showed that trends could be seen in almost 90% of all months in IPTA and ITA, while with the MK test, this rate was only 23%. They also noted that ITA and IPTA are more sensitive than MK in detecting a trend. Ahmed et al. (2022) observed the trend analysis of monthly precipitation in the Hindukush–Karakoram–Himalaya River basins of Pakistan using MK, Sen's slope, ITA, and IPTA methods. According to their findings, the MK method was less sensitive than the IPTA and ITA methods at determining trends. Even if 90% of the innovative methods are compatible with one another, the IPTA provides more details on trend transitions between subsequent time series segments. Wu et al. (2022) analyzed the precipitation variations in the Tai Lake Basin from 1971 to 2018 based on innovative trend approaches. Their results indicated that the detection results from the ITA method and traditional trend analysis methods (linear regression analysis, MK, and modified MK) were consistent. As an addition to ITA, IPTA can systematically identify consecutive seasons and monthly transition characteristics. The period and transition properties of a time series cannot be seen using the general trend detection method. The innovative polygon trend analysis (IPTA), which is based on ITA, can reflect the periodic properties of hydrometeorological variables and is frequently used for monthly and seasonal scales as a way to discover the transition trends between months and seasons (Ceribasi & Ceyhunlu 2020; Adelodun et al. 2022; Akçay et al. 2022; Wu et al. 2022).

Since ITA and IPTA methods are newly published, a few studies have been done. The previous studies generally focused on precipitation and temperature data. But other meteorological parameters that are important in the hydrology cycle need to be examined. Therefore, all parameters should be evaluated together. Unlike other studies, evapotranspiration and relative humidity data were examined for the first time. Also, no research employing innovative techniques on temperature or precipitation data has yet been done for the Ankara province. Therefore, it provides the opportunity to look at the region to be analyzed from a broad perspective. The aim of this study is to detect the trend of hydroclimate data (precipitation, temperature, relative humidity, and evapotranspiration) monthly and annually by comparing it with both the classical MK method and the ITA and IPTA methods. Ankara region was chosen as the study area because it is one of the most important provinces as the capital city of Turkey. To identify the trend slope, Sen's slope was used in this study. CUSUM and SQ-MK tests were employed to detect abrupt changes in hydrometeorological variables.

Ankara is located in the northern part of the Central Anatolian Region, which constitutes the interior of Turkey. It is neighbors with Çankırı in the northeast, Kırıkkale in the east, Kırşehir and Aksaray in the southeast, Konya in the south, Eskişehir in the west, and Bolu in the northwest. In addition, the city of Ankara is located between 38°40′ and 40°45′ north latitudes and 30°50′ and 33°53′ east longitudes. There are climatic differences from place to place in the wide area of the province. In the south, there is the steppe climate, which is the distinctive feature of the Central Anatolian climate, while in the north, the temperate and rainy states of the Black Sea climate can be seen. In this region where the continental climate prevails, winter temperatures are low and summer is hot. The hottest month is July–August, and the coldest month is January. The amount of precipitation falling in the region differs in the northern and southern parts. Kızılcahamam and Çubuk districts located in the northern direction are characterized by the precipitation regime of the Black Sea Region. In the south direction, it shows the climate characteristic of the Central Anatolia Region (Çi̇çek 2003).

The average temperature in the province is 11.7 °C, and the annual average precipitation is 389.1 mm. The highest temperature value was determined as 40.8 °C and the lowest temperature as −24.9 °C. The number of days with frost is 60–117, and the number of snowy days is 30.5 days a year. The highest snow thickness was determined as 30 cm. When we look at the wind conditions of the city center and stations in general. It is seen that the prevailing wind changes depending on the terrain structure. According to this, prevailing winds are northeast in Ankara (center), Esenboğa, Çubuk, Ayaş, and Yenimahalle districts, west in Haymana (İkizce), Sincan, Dikmen district and Nallıhan districts, north in Polatlı and Şerefikoçhisar districts, southwest in Etimesgut and Elmadağ districts, southeast in Kızılcahamam district, and northeast in Beypazarı district work in the direction. The months with strong winds are March and April. The highest wind speed detected in Ankara is 29.2 m/s. The only exception to these measurements is the tornado disaster that caused great damage in the Akyurt district in 2007. According to the values measured for many years, Ankara's average pressure value is 913.1 Mb, the highest-pressure value detected is 935.0 Mb, and the lowest pressure value is 891.0 Mb (Afshar et al. 2020; Danandeh Mehr et al. 2020).

The average number of snowy days does not exceed 1 month. Steppe plants are seen in most of the province. The land, which is green in spring, is covered with yellowed and dried grass in summer. The influence of the maritime climate is seen in the north and northwest of the province and this region is covered with large forests. Forests and mountains in the north attract rain clouds, preventing them from descending to the south. Most of the forests are groves and coppice forests. 10% of its area is forested. 15% of the land is meadows and pasture. Grain is a vegetation that covers the largest land (Cakmak & Şimşek 2010). Figure 1 indicates information about the province of Ankara and the precipitation stations used. In this study, the most important stations 17128 (Ankara Airport), 17130 (Ankara Center), and 17664 (Kızılcahamam) are considered. Statistical characteristics determined from the time series of each gauging station are also presented in Table 1. The range of the following characteristics are evaluated from the observed hydrometeorological time series: the mean, standard deviation (SD), coefficient of variation (Cv), coefficient of skewness (Cs), and lag-one autocorrelation coefficient (r1). According to Table 1, the highest annual total precipitation is seen at station 17664 as 579.602 mm, while the lowest is observed at station 17130 as 391.051 mm. In addition, the highest and lowest annual mean maximum temperature is recorded at stations 17130 (25.05 °C) and 17664 (23.804 °C), respectively. It is seen that the temperature is low where the precipitation is high. The annual total evapotranspiration data matches with the stations indicated by the annual mean temperature data, whereas the highest and lowest values of the annual mean relative humidity are similar to the stations determined by the annual total precipitation series. The highest SD value is detected at station 17664 in precipitation data, whereas the lowest SD value is captured at station 17128 in temperature data. The Cv and Cs are determined in the temperature data with the lowest SD value as 0.039 and −0.061, respectively. The r1 value as observed at the precipitation station where the SD value is higher (station 17664).
Table 1

Statistical parameters of observed meteorological stations

Station codeMeteorological parametersLatitudeLongitudeEarliest record yearLatest record yearMeanSDCvCsr1
17128 ATP 40.124 32.999 1956 2021 410.503 84.345 0.205 −0.032 0.244 
AMT 24.091 0.936 0.039 −0.061 0.315 
AMRH 65.206 3.359 0.052 −0.317 0.584 
ATE 649.695 51.305 0.079 0.453 0.417 
17664 ATP 40.472 32.644 1959 2021 579.602 113.560 0.196 0.350 0.012 
AMT 23.804 1.031 0.043 −0.123 0.416 
AMRH 66.366 3.121 0.047 0.091 0.422 
ATE 635.094 43.885 0.069 0.515 0.367 
17130 ATP 39.972 32.863 1926 2021 391.051 79.802 0.204 0.373 0.214 
AMT 25.054 0.995 0.040 −0.562 0.148 
AMRH 60.112 2.880 0.048 −0.385 0.483 
ATE 712.428 52.285 0.073 0.653 0.400 
Station codeMeteorological parametersLatitudeLongitudeEarliest record yearLatest record yearMeanSDCvCsr1
17128 ATP 40.124 32.999 1956 2021 410.503 84.345 0.205 −0.032 0.244 
AMT 24.091 0.936 0.039 −0.061 0.315 
AMRH 65.206 3.359 0.052 −0.317 0.584 
ATE 649.695 51.305 0.079 0.453 0.417 
17664 ATP 40.472 32.644 1959 2021 579.602 113.560 0.196 0.350 0.012 
AMT 23.804 1.031 0.043 −0.123 0.416 
AMRH 66.366 3.121 0.047 0.091 0.422 
ATE 635.094 43.885 0.069 0.515 0.367 
17130 ATP 39.972 32.863 1926 2021 391.051 79.802 0.204 0.373 0.214 
AMT 25.054 0.995 0.040 −0.562 0.148 
AMRH 60.112 2.880 0.048 −0.385 0.483 
ATE 712.428 52.285 0.073 0.653 0.400 

ATP, annual total precipitation (mm); AMT, annual maximum temperature (°C); AMRH, annual mean relative humidity (%); ATE, annual total evapotranspiration (mm).

Figure 1

The location of the meteorological stations in the Ankara province.

Figure 1

The location of the meteorological stations in the Ankara province.

Close modal

Mann-Kendall (MK) test

The MK test (Mann 1945), also known as Kendall's tau statistic, is a widely used method for determining trends in hydrometeorological time series and is recommended by the World Meteorological Organization (WMO). The existence of a trend is investigated with the hypothesis test applied in the method. H0 hypothesis: Observation values ordered according to time (x1, x2, … , xn) are random variables with similar distribution and independent of time. Hypothesis H1: The distribution of xk and xj values in the series is not similar for (k, jn) including (kj) (Mann 1945; Kendal 1975). The formula for calculating the MK test statistic Z is as follows:
(1)
where n is the number of the data, xj and xk are the data point in years j and k (j > k), and ti is the length of tied rank group.
(2)
(3)
(4)

A positive Z number shows an upward trend, while a negative value shows a downward trend. Critical test statistical values at 90, 95, and 99% probability levels are 1.645, 1.96, and 2.58, respectively (Yuce & Esit 2021).

ITA with significance test

The ITA method developed by Şen (2012) is a technical analysis method that shows a possible general increase or decrease in a given time series. In this method, the available data are arranged according to the order, and then they are divided into two equal series. These series, which are divided into two, are ordered from smallest to largest. The first part of the series (Xi) is placed on the X-axis of the Cartesian coordinate system, and the second part (Xj) is placed on the Y-axis. The data are first examined for its position on the 1:1 line. Data above the 1:1 line indicate an increasing trend in the data, while data below the 1:1 line indicate a decreasing trend in the data. Secondly, it is examined in which cluster the data are among the low, medium, or high clusters. Finally, if the data are clustered around the 1:1 line, it is concluded that there is no trend (Caloiero et al. 2018; Serinaldi et al. 2020).

Şen (2017) proposed the statistical significance test. This method divides the hydrometeorological time series into two equal parts and calculates their arithmetic averages (y1 and y2). The following formula is used to calculate the trend slope (s):
(5)
(6)
(7)
(8)
(9)
E(s) is the first-order moment of the slope, n is the data length, is the cross-correlation coefficient between two parts, is the trend slope variance, and is the SD of the trend slope in these equations. The trend slope confidence interval is determined as follows:
(10)

The value z produced from the standard normal distribution at a given confidence level is referred to as . It is considered as an increasing (decreasing) trend if the trend slope exceeds the upper (lower) confidence level. There is no statistically significant trend at a specific confidence level if these conditions are not met. The level of confidence in this study is 95%.

IPTA method

The IPTA graphic approach was newly presented by Şen et al. (2019) and showed that it has several advantages over other classical trend methods: It can be used on a variety of time scales (i.e., daily, monthly, seasonal, annual, etc.). IPTA constitutes an approach in inter-period transitions, but other trend analysis approaches cannot form an approach in seasonal transitions and so display deficiencies in this regard. It is unaffected by serial analysis as well as non-parametric. IPTA's step-by-step computation methodology is demonstrated in Figure 2 as a flowchart. Trend slope and length are evaluated as follows:
(11)
(12)
where s represents the trend slope, |AB| is the trend length, x1 and x2 are two consecutive points in the first part in horizontal, and y1 and y2 are two consecutive points in the second part. Figure 3 shows a schematic representation of the monthly IPTA. The points on the IPTA template represent a decreasing (increasing) trend when positioned below (above) the 1:1 trend line in the Cartesian coordinate system (Şen 2012). The straight lines that connect the points provide information about the variation throughout the month. The contribution of changes between months to the average change in the hydrometeorological series is substantial if the slopes of the lines between consecutive months are far from each other, and vice versa.
Figure 2

IPTA flowchart procedure step by step.

Figure 2

IPTA flowchart procedure step by step.

Close modal
Figure 3

IPTA application (Şen et al. 2019).

Sen's slope estimator

The Sen's slope test introduced by Sen (1968) is a non-parametric test that calculates the slope of the trend in a data set. It is applied for equiponderant time series. For each data point, the slope difference is evaluated per changing time. The slope of the trend can be predicted by the median of all slopes between data pairs in the same season (Helsel & Hirsch 2002). All slope pairs are ranked from smallest to largest and if the calculated number of slopes (n) is odd, the median slope gives the slope S. If n is even, then the two median slopes are averaged. Where Q is the data, n is the length of the data, and T is the time. The slope of n pair of data is predicted by Sen's estimator:
(13)
(14)

Standard normal homogeneity test (SNHT), CUSUM test, and sequential Mann-Kendall test (SQ-MK)

The standard normal homogeneity test (SNHT) was introduced by Alexandersson (1986). The test statistic (Tk) is performed to compare the average of the first n year with the average of the last (nk) year with n data points (Vezzoli et al. 2012; Jaiswal et al. 2015). The Tk equation is written as follows:
(15)
Z1 and Z2 can be calculated as follows:
(16)
(17)

Here, and are mean and SD, respectively. The year in which reaches the maximum value is considered as the point of change.

Distribution free CUSUM (trendchange package in R) test is employed to determine the abrupt change point in a series of data. The change point in the time series will be the point at which the cumulative sum reaches its maximum value. The significance of the change point is indicated if the highest value is equal to or greater than the critical value ((Patakamuri & Das 2019).

It is well known that using the MK method to detect trends at the end of any time period does not provide a complete trend picture (structure of the trend) for the entire series. There may be changes in the trend across the investigation period, which can be noticed by performing the test for each period separately (Sneyers 1990). According to Makokha & Shisanya (2010), negative or positive trends are not always significant. However, it can be discovered using SQ-MK graphs.

The SQ-MK test is computed using rank values. The magnitudes of yi (i = 1, 2, 3, …, n) are compared with yj (j = 1, 2, 3, …, i − 1) of the original values in the series (x1, x2, x3, …, xn). The cases when yi > yj are counted and designated by ni for each comparison. As a result, the statistic ti is calculated as follows:
(18)
The distribution of test statistic ti has a mean given by:
(19)
Variance is evaluated as:
(20)
The following equation is performed to calculate the forward sequential values of the statistic (Sneyers 1990).
(21)

The backward sequential statistic u () are calculated in the same way, but starting from the end of the series. Many scientists have used this approach to determine the starting point of trends (Rahman et al. 2017; Salehi et al. 2020; Alhathloul et al. 2021) rather than identifying the whole trend. In this study, the SQ-MK method is applied to detect an abrupt change in the time series of hydrometeorological data.

Correlation analysis between meteorological variables

Figure 4 indicates correlation analysis using the Spearman test between meteorological variables including annual total precipitation, annual mean maximum temperature, annual total evapotranspiration, and annual mean relative humidity. According to Figure 4, there is no strong correlation between precipitation and temperature considering all stations. For example, a negative correlation is observed at station 17128 as −0.14, while 0.12 and 0.09 are detected at stations 17130 and 17664, respectively. However, a strong correlation is noted between temperature and evapotranspiration series at all stations. Correlation results between these two variables are calculated as 0.78, 0.61, and 0.55 at stations 17128, 17130, and 17664, respectively. Considering the relative humidity series, a weak correlation is captured with temperature and evapotranspiration series at all stations. Precipitation data is observed a good correlation with temperature and evapotranspiration data as 0.44 and 0.46 at station 17128, respectively.
Figure 4

Correlation results of meteorological variables, (a) station 17128, (b) station 17130, and (c) station 17664.

Figure 4

Correlation results of meteorological variables, (a) station 17128, (b) station 17130, and (c) station 17664.

Close modal

MK test results

Table 2 shows the results of the MK test for annual hydrometeorological variables (precipitation, temperature, relative humidity, and evapotranspiration). Confidence levels are selected to be 10, 5, and 1%. Annual total precipitation has no significant trend at stations 17128 and 17664, while an increasing trend has been found at station 17128. Stations 17128 and 17130 indicate a significant increasing trend for annual maximum Temperature, whereas there is no significant trend at station 17664. Additionally, all stations indicate no increasing/decreasing trend at annual mean relative humidity. But all stations show an increasing trend (10 and 5% confidence level) at annual total evapotranspiration.

Table 2

Results of Mann-Kendall test for annual hydrometeorological variables

 
 

MK and Sen's slope tests results for monthly hydroclimate variables are presented in Table 3. As shown in Table 3, the months in bold demonstrate significant trends at the 90, 95, and 99% confidence levels. According to monthly precipitation data, an increasing trend is indicated for station 17130 in January (90% confidence level), March (95% confidence level), and October (90% confidence level), while a decreasing trend is detected for station 17128 in February (90% confidence level). An increasing trend is obtained from stations 17128 and 17130 in January, February, June, July, and September for temperature data, whereas no trends are observed at station 17664.

Table 3

Results of the Mann-Kendall for monthly hydrometeorological variables

Station NoMeteorological VariablesTestsJanFebMarAprMayJunJulAugSepOctNovDec
17128 ATP MK 0.094 − 1.8260.465 0.149 1.107 1.218 0.011 1.484 1.218 1.533 1.173 1.256 
SS 0.025 0.263 0.066 0.029 0.220 0.196 0.000 0.100 0.076 0.176 0.140 0.195 
AMT MK 1.822.16** 1.617 0.255 1.9322.375** 2.226** 1.639 2.857*** 1.035 0.736 0.537 
SS 0.041 0.060 0.025 0.005 0.027 0.033 0.029 0.020 0.035 0.018 0.011 0.007 
AMRH MK 0.083 − 2.640 − 2.873** 0.974 1.472 1.273 0.482 0.487 1.893 0.315 0.310 0.011 
SS 0.000 0.092 0.096 0.031 0.056 0.050 0.012 0.023 0.100 0.019 0.012 0.000 
ATE MK 1.370 2.34** 1.938** 0.554 2.210** 2.155** 3.345*** 3.244*** 3.877*** 2.696*** 0.709 0.124 
SS 0.000 0.015 0.104 0.033 0.125 0.143 0.230 0.261 0.216 0.157 0.024 0.000 
17130 ATP MK 1.640.332 2.415** 1.317 0.237 1.456 0.256 0.329 0.196 1.880* 0.282 0.101 
SS 0.144 0.033 0.189 0.119 0.027 0.127 0.009 0.006 0.007 0.125 0.018 0.010 
AMT MK 2.43** 3.036*** 0.779 0.003 0.855 3.119*** 2.014** 1.257 1.7791.605 1.150 0.516 
SS 0.029 0.039 0.009 0.000 0.007 0.023 0.015 0.009 0.015 0.014 0.008 0.005 
AMRH MK − 2.504** − 4.110*** − 2.384** 0.108 0.247 2.219** 2.532** 3.44*** 1.599 2.206* 1.019 − 2.289** 
SS 0.031 0.080 0.050 0.003 0.006 0.055 0.053 0.083 0.046 0.067 0.024 0.030 
ATE MK 2.16** 2.51** 2 93*** 0.874 0.618 0.805 2.265** 1.568 2.29** 1.600 0.931 0.279 
SS 0.000 0.028 0.110 0.036 0.023 0.031 0.102 0.072 0.090 0.061 0.022 0.000 
17664 ATP MK 0.415 1.257 0.973 0.504 0.896 0.243 0.249 0.474 0.581 0.712 1.163 1.341 
SS 0.157 0.367 0.163 0.112 0.204 0.065 0.032 0.043 0.070 0.144 0.191 0.475 
AMT MK 0.457 0.932 1.323 1.033 0.237 0.160 0.861 0.463 0.481 1.478 0.487 0.154 
SS 0.009 0.025 0.022 0.024 0.003 0.000 0.013 0.005 0.007 0.029 0.009 0.003 
AMRH MK 3.30*** 0.285 0.570 1.459 1.263 0.540 − 2.047** − 1.9340.807 1.436 0.042 1.317 
SS 0.130 0.015 0.019 0.060 0.074 0.029 0.083 0.089 0.035 0.067 0.000 0.050 
ATE MK 0.117 2.07** 0.777 0.154 1.550 2.26** 3.35*** 4.09*** 3.01*** 1.507 0.196 0.670 
SS 0.000 0.000 0.050 0.000 0.100 0.150 0.230 0.336 0.172 0.083 0.008 0.000 
Station NoMeteorological VariablesTestsJanFebMarAprMayJunJulAugSepOctNovDec
17128 ATP MK 0.094 − 1.8260.465 0.149 1.107 1.218 0.011 1.484 1.218 1.533 1.173 1.256 
SS 0.025 0.263 0.066 0.029 0.220 0.196 0.000 0.100 0.076 0.176 0.140 0.195 
AMT MK 1.822.16** 1.617 0.255 1.9322.375** 2.226** 1.639 2.857*** 1.035 0.736 0.537 
SS 0.041 0.060 0.025 0.005 0.027 0.033 0.029 0.020 0.035 0.018 0.011 0.007 
AMRH MK 0.083 − 2.640 − 2.873** 0.974 1.472 1.273 0.482 0.487 1.893 0.315 0.310 0.011 
SS 0.000 0.092 0.096 0.031 0.056 0.050 0.012 0.023 0.100 0.019 0.012 0.000 
ATE MK 1.370 2.34** 1.938** 0.554 2.210** 2.155** 3.345*** 3.244*** 3.877*** 2.696*** 0.709 0.124 
SS 0.000 0.015 0.104 0.033 0.125 0.143 0.230 0.261 0.216 0.157 0.024 0.000 
17130 ATP MK 1.640.332 2.415** 1.317 0.237 1.456 0.256 0.329 0.196 1.880* 0.282 0.101 
SS 0.144 0.033 0.189 0.119 0.027 0.127 0.009 0.006 0.007 0.125 0.018 0.010 
AMT MK 2.43** 3.036*** 0.779 0.003 0.855 3.119*** 2.014** 1.257 1.7791.605 1.150 0.516 
SS 0.029 0.039 0.009 0.000 0.007 0.023 0.015 0.009 0.015 0.014 0.008 0.005 
AMRH MK − 2.504** − 4.110*** − 2.384** 0.108 0.247 2.219** 2.532** 3.44*** 1.599 2.206* 1.019 − 2.289** 
SS 0.031 0.080 0.050 0.003 0.006 0.055 0.053 0.083 0.046 0.067 0.024 0.030 
ATE MK 2.16** 2.51** 2 93*** 0.874 0.618 0.805 2.265** 1.568 2.29** 1.600 0.931 0.279 
SS 0.000 0.028 0.110 0.036 0.023 0.031 0.102 0.072 0.090 0.061 0.022 0.000 
17664 ATP MK 0.415 1.257 0.973 0.504 0.896 0.243 0.249 0.474 0.581 0.712 1.163 1.341 
SS 0.157 0.367 0.163 0.112 0.204 0.065 0.032 0.043 0.070 0.144 0.191 0.475 
AMT MK 0.457 0.932 1.323 1.033 0.237 0.160 0.861 0.463 0.481 1.478 0.487 0.154 
SS 0.009 0.025 0.022 0.024 0.003 0.000 0.013 0.005 0.007 0.029 0.009 0.003 
AMRH MK 3.30*** 0.285 0.570 1.459 1.263 0.540 − 2.047** − 1.9340.807 1.436 0.042 1.317 
SS 0.130 0.015 0.019 0.060 0.074 0.029 0.083 0.089 0.035 0.067 0.000 0.050 
ATE MK 0.117 2.07** 0.777 0.154 1.550 2.26** 3.35*** 4.09*** 3.01*** 1.507 0.196 0.670 
SS 0.000 0.000 0.050 0.000 0.100 0.150 0.230 0.336 0.172 0.083 0.008 0.000 

*, **, *** represent the significance levels at 90, 95, 99%, respectively. See

There is a decreasing trend in February (α = %1), March (α = %1), and September (α = %10) for station 17128 considering relative humidity data. For station 17130, a decreasing trend in January–March, and December, and an increasing trend are observed in June–August and October. According to station 17664, an increasing trend is detected in January (α = %1), while there is a decreasing trend in July (α = %5) and August (α = %10). For evapotranspiration data, an increasing trend in February–March and May–October periods at station 17128, an increasing trend in the January–March, July, and September periods for station 17130, and an increasing trend are observed in February and June–September periods. No trends are captured in April and November and December (except relative humidity for station 17130) significantly for all hydrometeorological variables by the MK test.

The highest decreasing Sen's slope is indicated in February (−0.263) for station 17128 (precipitation data), whereas the highest increasing SS is observed in August (0.336) for station 17664 (evapotranspiration data).

ITA with significance test results

ITA results are presented in Figure 5 for all hydrometeorological variables at a 95% confidence level. According to ITA methodology, if the slope value is between the lower and upper limits, there is no trend. Additionally, there is an increasing trend if the slope value is above the upper limit and a decreasing trend if the slope value is below the lower limit. All variables indicated a significance decreasing or increasing trend. Hence, the ITA approach is more sensitive than the MK trend test in detecting hidden trends. Trend slope, upper, and lower limits of the data at a 95% significance level are also given in Figure 5.
Figure 5

Innovative trend significance trend test results at a 95% significance level for all hydrometeorological variables.

Figure 5

Innovative trend significance trend test results at a 95% significance level for all hydrometeorological variables.

Close modal

At station 17128, trend slopes are observed higher than the lower limits at the winter season (December–January–February) for precipitation data. It means that there is a decreasing trend for the winter season. Decreasing trends are captured in May and September, while increasing trends are detected in the summer season (June–July–August) and October and November considering precipitation data. According to temperature and evapotranspiration data, the trend slope is much higher than the upper limits for all months (except December as a decreasing trend) at significance levels. Thus, increasing trends are observed for all seasons. Relative humidity shows non-similarity compared with temperature data. The decreasing trend appears in the spring season (March–April–May), August, September, and November, while the increasing trend is just observed in October. No trend is detected in June, July, and December.

At station 17130, decreasing trends are detected in January, March, April, June, August, October, and November, while decreasing trends are observed in February, September, and December for precipitation data. Trend slopes indicate lower values than upper/lower limits in May and July. For temperature data, increasing trends in the winter (DJF) season, March, June, July, September, and October and decreasing trends are detected just in November. Evaporation in the summer season (JJA) is in the increasing trend with the January–April period and September and October, while November and December are observed a decreasing trend. According to relative humidity data, decreasing trends occur in the winter season, and notable increasing trends are observed in the summer season. There are no trends in May and November.

At station 17664, precipitation in the winter (DJF) season, as well as May and November, indicate a decreasing trend, whereas an increasing trend appears in March, April, July, September, and October. The temperature in the summer season (JJA) and January, February, and November are observed an increasing trend, notable decreasing trends are shown in October and December. Relative humidity in the spring (MAM) and summer (JJA) seasons appears a decreasing trend with September, while December and January are detected in an increasing trend. According to evapotranspiration data, the summer season, as well as February, March, May, September, and October, demonstrate an increasing trend, while a decreasing trend is captured in January and April.

IPTA results

The IPTA graphs evaluated the trend slope and trend length to allow for a better understanding of hydrometeorological regimes in the following months. In addition, when the length and slope of the trend between two consecutive months are upward, the difference between the hydrometeorological variables including, rainfall, temperature, relative humidity, and evapotranspiration averages of the relevant months upward as well and these variables regimes perform more unpredictably. IPTA graphs for each variable are presented in Figure 6. The trend slope and the trend length for each mean hydrometeorological variable are given in Table 4.
Table 4

Results of the IPTA for monthly hydrometeorological variables

Station NoVariablesTestsJan–FebFeb–MarMar–AprApr–MayMay–JunJun–JulJul–AugAug–SepSep–OctOct–NovNov–DecDec–Jan
17128 – Ankara Airport P (mm) Trend length 15.33 8.40 4.99 11.11 20.30 31.74 2.99 7.11 19.53 4.95 29.62 7.79 
Trend slope 0.95 − 4.41 0.50 0.16 0.23 1.19 − 0.47 − 0.01 2.87 − 0.04 0.56 1.74 
T (°C) Trend length 4.23 9.57 6.51 4.74 5.03 4.71 0.24 4.50 7.02 10.44 8.48 4.27 
Trend slope 1.21 0.88 0.90 1.31 0.99 1.15 − 3.80 0.98 1.11 0.96 1.23 0.68 
RH (%) Trend length 6.35 9.89 6.93 0.81 6.55 12.96 1.26 6.26 13.92 13.52 10.77 0.90 
Trend slope 2.07 1.09 0.59 − 2.28 0.51 1.02 1.49 0.62 1.39 0.86 1.11 4.79 
ET (mm) Trend length 3.61 18.27 38.84 48.23 36.25 40.87 11.19 60.86 50.77 43.20 16.66 3.39 
Trend slope 1.21 1.24 0.94 1.09 1.03 1.13 0.84 1.09 1.02 1.19 1.04 0.94 
17130 − Ankara Center P (mm) Trend length 9.50 8.19 4.53 14.63 23.75 29.45 3.08 8.00 16.12 6.00 19.82 8.41 
Trend slope 8.98 43.69 6.06 0.37 0.74 1.31 − 0.20 0.29 4.27 0.23 0.52 − 0.03 
T (°C) Trend length 3.58 9.87 6.66 5.32 4.94 4.19 0.46 4.90 7.34 9.74 9.28 3.40 
Trend slope 1.48 0.86 0.84 1.03 1.32 0.90 − 0.61 0.85 0.91 1.30 0.86 0.80 
RH (%) Trend length 7.03 11.72 8.86 2.56 7.06 10.82 2.06 6.85 15.22 15.65 11.72 0.24 
Trend slope 1.89 0.77 0.61 1.49 0.61 0.96 0.19 0.58 1.11 0.73 0.85 − 0.19 
ET (mm) Trend length 3.80 18.92 40.69 53.15 40.06 41.64 12.31 64.46 53.73 42.50 20.48 4.70 
Trend slope 1.91 1.23 0.88 0.97 1.02 1.10 1.17 0.96 1.02 1.20 0.87 0.73 
17664 − Ankara − Kizilcahamam P (mm) Trend length 23.46 10.75 4.98 10.77 24.71 31.54 3.89 3.00 19.23 15.54 51.81 12.74 
Trend slope 0.63 − 1.36 1.06 − 0.33 0.80 0.76 − 8.45 − 1.78 1.64 − 0.06 0.94 2.20 
T (°C) Trend length 3.56 10.03 6.91 4.99 4.72 4.97 0.25 4.28 7.36 10.28 9.37 3.68 
Trend slope 0.87 0.82 1.01 1.36 1.01 1.23 − 38.75 1.20 1.05 0.91 1.11 0.68 
RH (%) Trend length 4.85 9.18 5.21 0.38 3.26 9.65 1.06 4.01 11.46 9.40 9.31 2.02 
Trend slope 3.06 1.19 1.14 −2.96 0.46 1.65 0.31 1.64 1.33 0.91 1.27 0.20 
ET (mm) Trend length 3.14 18.02 8.12 47.33 35.71 38.04 13.02 57.04 48.00 41.03 17.25 4.03 
Trend slope 1.46 0.98 0.95 1.22 1.00 1.15 0.78 1.18 1.03 1.15 0.99 1.02 
Station NoVariablesTestsJan–FebFeb–MarMar–AprApr–MayMay–JunJun–JulJul–AugAug–SepSep–OctOct–NovNov–DecDec–Jan
17128 – Ankara Airport P (mm) Trend length 15.33 8.40 4.99 11.11 20.30 31.74 2.99 7.11 19.53 4.95 29.62 7.79 
Trend slope 0.95 − 4.41 0.50 0.16 0.23 1.19 − 0.47 − 0.01 2.87 − 0.04 0.56 1.74 
T (°C) Trend length 4.23 9.57 6.51 4.74 5.03 4.71 0.24 4.50 7.02 10.44 8.48 4.27 
Trend slope 1.21 0.88 0.90 1.31 0.99 1.15 − 3.80 0.98 1.11 0.96 1.23 0.68 
RH (%) Trend length 6.35 9.89 6.93 0.81 6.55 12.96 1.26 6.26 13.92 13.52 10.77 0.90 
Trend slope 2.07 1.09 0.59 − 2.28 0.51 1.02 1.49 0.62 1.39 0.86 1.11 4.79 
ET (mm) Trend length 3.61 18.27 38.84 48.23 36.25 40.87 11.19 60.86 50.77 43.20 16.66 3.39 
Trend slope 1.21 1.24 0.94 1.09 1.03 1.13 0.84 1.09 1.02 1.19 1.04 0.94 
17130 − Ankara Center P (mm) Trend length 9.50 8.19 4.53 14.63 23.75 29.45 3.08 8.00 16.12 6.00 19.82 8.41 
Trend slope 8.98 43.69 6.06 0.37 0.74 1.31 − 0.20 0.29 4.27 0.23 0.52 − 0.03 
T (°C) Trend length 3.58 9.87 6.66 5.32 4.94 4.19 0.46 4.90 7.34 9.74 9.28 3.40 
Trend slope 1.48 0.86 0.84 1.03 1.32 0.90 − 0.61 0.85 0.91 1.30 0.86 0.80 
RH (%) Trend length 7.03 11.72 8.86 2.56 7.06 10.82 2.06 6.85 15.22 15.65 11.72 0.24 
Trend slope 1.89 0.77 0.61 1.49 0.61 0.96 0.19 0.58 1.11 0.73 0.85 − 0.19 
ET (mm) Trend length 3.80 18.92 40.69 53.15 40.06 41.64 12.31 64.46 53.73 42.50 20.48 4.70 
Trend slope 1.91 1.23 0.88 0.97 1.02 1.10 1.17 0.96 1.02 1.20 0.87 0.73 
17664 − Ankara − Kizilcahamam P (mm) Trend length 23.46 10.75 4.98 10.77 24.71 31.54 3.89 3.00 19.23 15.54 51.81 12.74 
Trend slope 0.63 − 1.36 1.06 − 0.33 0.80 0.76 − 8.45 − 1.78 1.64 − 0.06 0.94 2.20 
T (°C) Trend length 3.56 10.03 6.91 4.99 4.72 4.97 0.25 4.28 7.36 10.28 9.37 3.68 
Trend slope 0.87 0.82 1.01 1.36 1.01 1.23 − 38.75 1.20 1.05 0.91 1.11 0.68 
RH (%) Trend length 4.85 9.18 5.21 0.38 3.26 9.65 1.06 4.01 11.46 9.40 9.31 2.02 
Trend slope 3.06 1.19 1.14 −2.96 0.46 1.65 0.31 1.64 1.33 0.91 1.27 0.20 
ET (mm) Trend length 3.14 18.02 8.12 47.33 35.71 38.04 13.02 57.04 48.00 41.03 17.25 4.03 
Trend slope 1.46 0.98 0.95 1.22 1.00 1.15 0.78 1.18 1.03 1.15 0.99 1.02 
Figure 6

Innovative polygon trend analysis graph for hydrometeorological variables.

Figure 6

Innovative polygon trend analysis graph for hydrometeorological variables.

Close modal

At station 17128, a sharp transition decreasing trend area to an increasing area is observed from February to March, while an increasing trend area to the decreasing area is seen in the transition from August to September for mean precipitation data. The winter (DJF) season is in the decreasing area far away from 45° (1:1 line). October appears to be a strong increasing trend, while April, May, and September showed a significant decreasing trend. The maximum trend length for precipitation data is evaluated as 31.74 mm in the June–July transition zone, whereas the lowest trend length is calculated as 4.95 mm within the October–November transition zone. The maximum trend slope is seen in the February–March transition zone as −4.41 and the lowest trend slope is evaluated as −0.01 within the August–September transition zone. According to temperature data, all months are in an increasing transition zone except December. In other words, the fours season is almost in an upward trend with far away from 45° (1:1 line). The trend length varies from 0.23 to 10.43 (°C) and the trend slope ranges from −3.8 to 1.308. The maximum trend length is seen in the October–November transition zone. The mean value of evapotranspiration data is similar to the temperature data as a significantly increasing trend zone. The maximum trend length is observed as 50.77 in the September–October transition zone, while the lowest trend length is seen in January–February as 3.61. The trend slope changes from 0.83 (July–August) to 1.24 (February–March). A sharp transition decreasing trend area is observed for relative humidity data in the spring (MAM) season, January–February, and August–September periods, a notable increasing trend appears in October. The trend slope varies between −2.27 (April–May) and 4.78 (December–January). The trend length ranges from 0.81 (April–May) to 13.91 (September–October).

At station 17130, rainfall has a transition from September to October in the decreasing trend zone to the increasing trend zone, while January–February, August–September, and November–December are in the significantly increasing trend zone. Rainfall in May and June are clustered around 45° (1:1 line) which means no trend. October–November has seen a strong increasing trend according to Figure 6. The maximum trend length is calculated as 23.75 in May–June transition zone, while the lowest value is seen as 4.53 from March to June periods. The trend slope for rainfall data varies from −0.02 (December–January) to 43.68 (February–March). According to temperature data, the winter (DJF) season is observed in the increasing area with far away from 45° (1:1 line), a notable decreasing trend is just seen in the November–December transition zone. The temperature in April, May, and August is clustered around no trend line. According to Table 4, the maximum trend length occurred as 9.86 from February to March, while the minimum trend length is calculated as 0.46 from July to August. The highest trend slope appears as 1.48 (January–February), notable the lowest trend slope is seen as −0.61 (July–August). Evaporation data for the same station, there is a distinct transition from October to November from the significantly increasing to decreasing trend area, while no trend is observed in May, June, and December. The trend slope ranges from 1.90 (January–February) to 0.73 (December–January). The maximum trend length varies from 53.72 (September–October) to 3.80 (January–February). In addition, the minimum trend slope and trend length are calculated in January–February transition zone. According to relative humidity data, the winter (DJF) season is evaluated in the decreasing area with far away from 45° (1:1 line), whereas the summer (JJA) season shows in the increasing area with the September–October periods. The maximum trend length is seen as 15.64 from October to November, while the minimum trend length is calculated as 0.23 from December to January. Trend slope for relative humidity data changes from −0.18 (December–January) to 1.89 (January–February).

At station 17664, the winter (DJF) season is in the decreasing transition area with May–June, August, and November periods, whereas the March–April, July, and September–October periods are in the significantly increasing for precipitation data. A sharp increasing trend from an upward to downward trend zone is observed in the transition from April to May, July to August, and October to November. A larger trend is clearly seen in October–November than in another month in the upward trend zone but in a decreasing direction. The maximum trend length for precipitation is calculated as 51.81 from November to December, notable the minimum decreasing trend length is seen as 2.99 from August to September. The highest trend slope is shown as −8.45 (July–August), while the lowest trend slope is seen as −0.06 (October–November). According to temperature data, there is one transition (February–March) that appears from an increasing trend area to a decreasing trend area. In general, the remaining months are clustered at an upward trend area except for May–June, September–October, and December which are near to no trend lines. Temperature from July to August has the minimum trend length and the maximum trend slope as 0.25 and −38.75, respectively. In addition, the maximum trend length and the minimum trend slope are observed as 10.08 (October–November) and 0.6 (December–January). According to evaporation data for the same station, from April to May is also one observed transition from a downward trend area to an upward area. The summer (JJA) season is shown in a significantly increasing trend, while the winter (DJF) season are observed as no trend due to clustering around 45° (1:1 line). The maximum trend length for ET data is calculated as 57.03 from August to September, whereas the minimum trend length is seen as 3.14 from January to February. From July to August is evaluated as the minimum trend slope with 0.78, notable that the maximum trend slope is observed from January to February as 1.46. According to relative humidity data, spring (MAM) and summer (JJA) seasons are shown to a decrease trend zone significantly. The increasing trend is observed in January, October, and December at the upward trend area. The trend length ranges from 0.38 (April–May) to 11.46 (September–October). The trend slope varies from 0.195 (December–January) to 3.05 (January–February).

A general assessment of trend tests (MK, ITA, and IPTA)

The results of MK, ITA, and IPTA tests are presented in Tables 5,678 for each hydrometeorological variable. When the tables are examined in general, there are differences between the MK test and the innovative tests. According to Table 5, only 1 of 36 months has a significant trend (decreasing/increasing) when using rainfall data, according to the MK test. However, IPTA and ITA capture a significant trend in 30 and 34 of 36 months, respectively. Innovative approaches caught the same trends as MK captured in the months when MK captured a trend. The ITA and IPTA, on the other hand, exhibit trends for the large majority of the 35 months where MK did not find any significant trend for precipitation data. The explanation for the difference in results between MK and innovative methods is that innovative methods are more sensitive in detecting trends (Achite et al. 2021; Şan et al. 2021; Akçay et al. 2022). The hidden trend in meteorological variables is detected effectively using graphical methods rather than the classical trend method.

Table 5

Comparison of MK, ITA, and IPTA for monthly precipitation data

 
 
Table 6

Comparison of MK, ITA, and IPTA for monthly temperature data

 
 
Table 7

Comparison of MK, ITA, and IPTA for monthly evapotranspiration data

 
 
Table 8

Comparison of MK, ITA, and IPTA for monthly relative humidity data

 
 

According to three approaches, it is obtained that there is an increasing trend in February, Jun, July, and September in temperatures and a decreasing trend in February and March in relative humidity for station 17228. Evaporation for the same station, the summer (JJA) season, February, May, September, and October indicate a significantly increasing trend considering all three methods. At station 17130, precipitation in March shows an increasing trend, while no trends are detected in May and July. Increasing trends are detected in January, February, June, and July, whereas April, May, and August show no significant trend for temperature. For relative humidity, the decreasing trend that begins in the winter season continues until the end of March. However, the summer season shows an increasing trend that continues until the end of October in terms of MK, ITA, and IPTA methods. Furthermore, there is no trend found in May and November. For evapotranspiration, an increasing trend is observed in January, February, March, July, and September, while no trend just is detected in May. According to station 17664, precipitation and temperature indicate the difference between MK and innovative approaches. In other words, the innovative method shows a significant trend, while the MK test does not capture in the same manner. Relative humidity in January is seen as an increasing trend by considering three methods. For evapotranspiration, an increasing trend is detected in the summer season, February and September. In addition, February, July, and September are observed as increasing for all three methods as well as for all stations.

Change point detection results

SNHT, CUSUM, and SQ-MK tests are applied to annual hydrometeorological variables to find any abrupt changes. Figure 7 indicates SQ-MK and CUSUM tests results for each station. Abrupt changes for both SNHT and CUSUM tests are presented in Table 9. The results show that SNHT and CUSUM tests fail to detect any abrupt changes at station 17128 in the annual maximum temperature and annual evapotranspiration series. Similarly, no changes are captured by the SNHT test at station 17664, while the CUSUM test is detected for annual precipitation and annual maximum temperature in 2008 and 2007 at a 95% confidence level, respectively. Three tests successfully capture a change point in 2005 at annual evapotranspiration for station 17664. According to station 17130, any changes in annual total precipitation, annual maximum and annual total evaporation series are detected successfully in 1960, 1997, and 1997 by both SNHT and CUSUM tests. However, interestingly, the SQ-MK test is unable to capture the same year in precipitation and evapotranspiration except for temperature. Upward trends are observed in annual precipitation around 1950, the annual maximum temperature in 1997, and the annual evapotranspiration series after 2015 by the SQ-MK test.
Table 9

SNHT and CUSUM tests results for abrupt changes in annual hydrometeorological variables

StationMeteorological variablesSNHT
CUSUM test
TkYearVkα = %5Year
17128 Annual total precipitation (mm) 5.0448 – 11.04 – 
Annual maximum temperature (°C) 22.331 1997 20 11.04 1993 
Annual mean relative humidity (%) 29.215 – 10 11.04 – 
Annual total evapotranspiration (mm) 14.881 2015 20 11.04 2011 
17130 Annual total precipitation (mm) 12.696 1960 15 13.35 1960 
Annual maximum temperature (°C) 17.596 1997 18 13.35 1997 
Annual mean relative humidity (%) 12.584 – 12 13.35 – 
Annual total evapotranspiration (mm) 39.276 1997 18 13.35 1997 
17664 Annual total precipitation (mm) 4.8957 – 16 10.79 2008 
Annual maximum temperature (°C) 12.701 – 48 10.79 2005 
Annual mean relative humidity (%) 4.1061 – 10.79 – 
Annual total evapotranspiration (mm) 23.283 2005 16 10.79 2005 
StationMeteorological variablesSNHT
CUSUM test
TkYearVkα = %5Year
17128 Annual total precipitation (mm) 5.0448 – 11.04 – 
Annual maximum temperature (°C) 22.331 1997 20 11.04 1993 
Annual mean relative humidity (%) 29.215 – 10 11.04 – 
Annual total evapotranspiration (mm) 14.881 2015 20 11.04 2011 
17130 Annual total precipitation (mm) 12.696 1960 15 13.35 1960 
Annual maximum temperature (°C) 17.596 1997 18 13.35 1997 
Annual mean relative humidity (%) 12.584 – 12 13.35 – 
Annual total evapotranspiration (mm) 39.276 1997 18 13.35 1997 
17664 Annual total precipitation (mm) 4.8957 – 16 10.79 2008 
Annual maximum temperature (°C) 12.701 – 48 10.79 2005 
Annual mean relative humidity (%) 4.1061 – 10.79 – 
Annual total evapotranspiration (mm) 23.283 2005 16 10.79 2005 
Figure 7

Sequential Mann-Kendal test results for annual hydrometeorological variables.

Figure 7

Sequential Mann-Kendal test results for annual hydrometeorological variables.

Close modal

There is one intersection point of and in annual relative humidity series except station 17664. No upward/downward trends are detected as well as no change point by three tests. In general, annual total evaporation series are an upward trend after 2005 by the SQ-MK test. Similarly, an upward trend is observed in the annual maximum temperature that the start of 1997. However, there are no significant trends detected in both annual precipitation (except station 17130) or annual relative humidity series.

Ankara is one of the most critical provinces in Turkey. Therefore, this paper aims to employ new graphical trend analysis as well as the classical method. The three most important meteorological stations evenly distributed in Ankara province are selected. For this purpose, hydrometeorological variables including precipitation, temperature, relative humidity, and evapotranspiration trends are investigated at monthly and annual scales. SNHT, CUSUM, and SQ-MK methods are used to utilize any abrupt change points in the annual series. IPTA (one of the most recent trend analysis approaches reported in the literature), ITA with significance test, and MK (one of the classical approaches of trend analysis) are applied to detect the presence of trends at both monthly and annual scales. The findings of the trend analysis are also compared. While the MK test fails to detect significant trends in the majority of variables, ITA and IPTA capture different trends. This paper results are convenient with previous studies in the literature (Caloiero et al. 2018; Ceribasi & Ceyhunlu 2020; Hırca et al. 2022). In other words, ITA with significance test, IPTA, and MK capture trends in 83, 94, and 0.2% of all months (3 meteorological stations × 12 months) for precipitation, 86, 75, and 22% of all months for temperature, 80, 80 and 30% of all months for relative humidity, and 91, 80 and 47% of all months for evapotranspiration series. These findings indicate that ITA with significance test and IPTA are more sensitive than the MK test with the highest in the precipitation series and lowest in the evapotranspiration series. In addition, three methods are convenient for capturing trends in evapotranspiration series due to higher trends. Thus, the MK test is unable to capture the hidden trend in time series, especially in precipitation series. Therefore, the hidden trend in precipitation is successfully captured by ITA and IPTA graphical methods. Previous studies (Şan et al. 2021; Hırca et al. 2022) also indicate the same result for precipitation data. In addition, although the trend captured by the MK method is observed by the IPTA and ITA methods, the trends that the MK method does not show in some months are shown by the IPTA and ITA methods. The advantage of IPTA graphs shows the trend slope as well as the trend length compared with the MK test. According to IPTA, MK and ITA with significance test indicate a holistic trend. According to the results, for station 17228, it is discovered that there is an increasing tendency in temperature in February, June, July, and September, and a decreasing trend in relative humidity in February and March. Considering all three approaches, evaporation for the same station during the summer (JJA) season shows a significantly increasing trend. Precipitation in March shows an increasing trend, while no trends are detected in May and July for station 17230. The decreasing trend in relative humidity that begins in the winter season continues through the end of March. In terms of MK, ITA, and IPTA procedures, however, the summer season shows an increasing trend that will last through the end of October. In the summer season, there is an increasing trend in evapotranspiration. Furthermore, all three methodologies as well as all stations show an increasing trend in February, July, and September. According to annual time series findings, at stations 17128 and 17664, there is no apparent trend in annual total precipitation, however, at station 17128, there is an increasing trend. At annual mean relative humidity, there is no indication of an increasing or decreasing trend at any of the stations. However, at annual total evapotranspiration, all stations show an increasing trend (10 and 5% confidence levels). According to change point detection, the results reveal that for station 17128 in the annual maximum temperature and annual evapotranspiration series, the SNHT and CUSUM tests fail to detect any abrupt changes. For station 17664, three tests successfully captured a change point in annual evapotranspiration in 2005. According to the SQ-MK test, annual total evaporation series have been increasing since 2005. Similarly, from the beginning of 1997, an increasing trend in annual maximum temperature has been noticed. However, neither the annual precipitation (except station 17130) nor the annual relative humidity series show any significant changes.

No funding was received for conducting this study.

M.E. data gathering, hydrometeorological data trend analysis, interpretation of the findings, manuscript writing, and submission.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.

Abdi
A.
,
Hassanzadeh
Y.
,
Talatahari
S.
,
Fakheri-Fard
A.
&
Mirabbasi
R.
2016
Regional drought frequency analysis using L-moments and adjusted charged system search
.
Journal of Hydroinformatics
19
(
3
),
426
442
.
https://doi.org/10.2166/hydro.2016.228
.
Adelodun
B.
,
Odey
G.
,
Cho
H.
,
Lee
S.
,
Adeyemi
K. A.
&
Choi
K. S.
2022
Spatial-temporal variability of climate indices in Chungcheong provinces of Korea: application of graphical innovative methods for trend analysis
.
Atmospheric Research
280
,
106420
.
https://doi.org/10.1016/j.atmosres.2022.106420
.
Afshar
M. H.
,
Şorman
A. Ü.
,
Tosunoğlu
F.
,
Bulut
B.
,
Yilmaz
M. T.
&
Danandeh Mehr
A.
2020
Climate change impact assessment on mild and extreme drought events using copulas over Ankara, Turkey
.
Theoretical and Applied Climatology
141
(
3
),
1045
1055
.
https://doi.org/10.1007/s00704-020-03257-6
.
Ahmed
N.
,
Wang
G.
,
Booij
M. J.
,
Ceribasi
G.
,
Bhat
M. S.
,
Ceyhunlu
A. I.
&
Ahmed
A.
2022
Changes in monthly streamflow in the Hindukush–Karakoram–Himalaya Region of Pakistan using innovative polygon trend analysis
.
Stochastic Environmental Research and Risk Assessment
36
(
3
),
811
830
.
https://doi.org/10.1007/s00477-021-02067-0
.
Akçay
F.
,
Kankal
M.
&
Şan
M.
2022
Innovative approaches to the trend assessment of streamflows in the Eastern Black Sea basin, Turkey
.
Hydrological Sciences Journal
67
(
2
),
222
247
.
https://doi.org/10.1080/02626667.2021.1998509
.
Alexandersson
H.
1986
A homogeneity test applied to precipitation data
.
Journal of Climatology
6
(
6
),
661
675
.
https://doi.org/10.1002/joc.3370060607
.
Alhathloul
S. H.
,
Khan
A. A.
&
Mishra
A. K.
2021
Trend analysis and change point detection of annual and seasonal horizontal visibility trends in Saudi Arabia
.
Theoretical and Applied Climatology
144
(
1
),
127
146
.
https://doi.org/10.1007/s00704-021-03533-z
.
Allan
R. P.
2021
Climate Change 2021: The Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change
.
WMO, IPCC Secretariat
.
Arab Amiri
M.
&
Gocić
M.
2021
Innovative trend analysis of annual precipitation in Serbia during 1946–2019
.
Environmental Earth Sciences
80
(
23
),
777
.
https://doi.org/10.1007/s12665-021-10095-w
.
Cakmak
B.
&
Şimşek
O.
2010
Drought analysis for 2007–2008 agricultural year of Turkey
.
Tekirdağ Ziraat Fakültesi Dergisi
7
(
3
),
99
109
.
Caloiero
T.
2020
Evaluation of rainfall trends in the South Island of New Zealand through the innovative trend analysis (ITA)
.
Theoretical and Applied Climatology
139
(
1
),
493
504
.
https://doi.org/10.1007/s00704-019-02988-5
.
Caloiero
T.
,
Coscarelli
R.
&
Ferrari
E.
2018
Application of the innovative trend analysis method for the trend analysis of rainfall anomalies in Southern Italy
.
Water Resources Management
32
(
15
),
4971
4983
.
https://doi.org/10.1007/s11269-018-2117-z
.
Cengiz
T. M.
,
Tabari
H.
,
Onyutha
C.
&
Kisi
O.
2020
Combined use of graphical and statistical approaches for analyzing historical precipitation changes in the Black Sea Region of Turkey
.
Water
12
(
3
).
Article 3. https://doi.org/10.3390/W12030705
.
Ceribasi
G.
&
Ceyhunlu
A. I.
2020
Analysis of total monthly precipitation of Susurluk Basin in Turkey using innovative polygon trend analysis method
.
Journal of Water and Climate Change
12
(
5
),
1532
1543
.
https://doi.org/10.2166/wcc.2020.253
.
Cetin
M.
,
Aksoy
H.
,
Onoz
B.
,
Eris
E.
,
Yuce
M.
,
Selek
B.
,
Aksu
H.
,
Hali
,
Eşit
M.
,
Cavus
Y.
&
Orta
S.
2018
Deriving Accumulated Precipitation Deficits from Drought Severity-Duration-Frequency Curves: A Case Study in Adana Province, Turkey
.
Çi̇çek
İ
.
2003
The statistical analysis of precipitation in Ankara, Turkey
.
Fırat Üniversitesi Sosyal Bilimler Dergisi
13
(
1
),
1
20
.
Danandeh Mehr
A.
,
Sorman
A. U.
,
Kahya
E.
&
Hesami Afshar
M.
2020
Climate change impacts on meteorological drought using SPI and SPEI: case study of Ankara, Turkey
.
Hydrological Sciences Journal
65
(
2
),
254
268
.
https://doi.org/10.1080/02626667.2019.1691218
.
Eris
E.
,
Aksoy
H.
,
Onoz
B.
,
Cetin
M.
,
Yuce
M. I.
,
Selek
B.
,
Aksu
H.
,
Burgan
H. I.
,
Esit
M.
,
Yildirim
I.
&
Karakus
E. U.
2019
Frequency analysis of low flows in intermittent and non-intermittent rivers from hydrological basins in Turkey
.
Water Supply
19
(
1
),
30
39
.
https://doi.org/10.2166/ws.2018.051
.
Esit
M.
,
Kumar
S.
,
Pandey
A.
,
Lawrence
D. M.
,
Rangwala
I.
&
Yeager
S.
2021
Seasonal to multi-year soil moisture drought forecasting
.
npj Climate and Atmospheric Science
4
(
1
).
Article 1. https://doi.org/10.1038/s41612-021-00172-z
Gumus
V.
,
Avsaroglu
Y.
&
Simsek
O.
2022
Streamflow trends in the Tigris river basin using Mann–Kendall and innovative trend analysis methods
.
Journal of Earth System Science
131
(
1
),
34
.
https://doi.org/10.1007/s12040-021-01770-4
.
Harka
A. E.
,
Jilo
N. B.
&
Behulu
F.
2021
Spatial-temporal rainfall trend and variability assessment in the Upper Wabe Shebelle River Basin, Ethiopia: application of innovative trend analysis method
.
Journal of Hydrology: Regional Studies
37
,
100915
.
https://doi.org/10.1016/j.ejrh.2021.100915
.
Helsel
D. R.
&
Hirsch
R. M.
2002
Statistical methods in water resources.(USGS Numbered Series No. 04-A3; Techniques of Water-Resources Investigations, Vols. 04-A3). U.S. Geological Survey. https://doi.org/10.3133/twri04A3
Hırca
T.
,
Eryılmaz Türkkan
G.
&
Niazkar
M.
2022
Applications of innovative polygonal trend analyses to precipitation series of Eastern Black Sea Basin, Turkey
.
Theoretical and Applied Climatology
147
(
1
),
651
667
.
https://doi.org/10.1007/s00704-021-03837-0
.
IPCC
2013
AR5 Climate Change 2013: The Physical Science Basis – IPCC
.
Isensee
L. J.
,
Detzel
D. H. M.
,
Pinheiro
A.
&
Piazza
G. A.
2022
Extreme streamflow time series analysis: trends, record length, and persistence
.
Journal of Applied Water Engineering and Research
,
1
14
.
https://doi.org/10.1080/23249676.2022.2030254
.
Ivanovski
K.
,
Hailemariam
A.
&
Smyth
R.
2021
The effect of renewable and non-renewable energy consumption on economic growth: non-parametric evidence
.
Journal of Cleaner Production
286
,
124956
.
https://doi.org/10.1016/j.jclepro.2020.124956
.
Jaiswal
R. K.
,
Lohani
A. K.
&
Tiwari
H. L.
2015
Statistical analysis for change detection and trend assessment in climatological parameters
.
Environmental Processes
2
(
4
),
729
749
.
https://doi.org/10.1007/s40710-015-0105-3
.
Kendall
M. G.
1975
Rank Correlation Methods
, 4th edn.
Charles Griffin
,
London
,
p. 272
.
Kisi
O.
2015
An innovative method for trend analysis of monthly pan evaporations
.
Journal of Hydrology
527
,
1123
1129
.
https://doi.org/10.1016/j.jhydrol.2015.06.009
.
Lehner
F.
,
Deser
C.
&
Sanderson
B. M.
2018
Future risk of record-breaking summer temperatures and its mitigation
.
Climatic Change
146
(
3
),
363
375
.
https://doi.org/10.1007/s10584-016-1616-2
.
Makokha
G. L.
&
Shisanya
C. A.
2010
Trends in mean annual minimum and maximum near surface temperature in Nairobi City, Kenya
.
Advances in Meteorology
2010
,
e676041
.
https://doi.org/10.1155/2010/676041
.
Malik
A.
,
Kumar
A.
,
Najah Ahmed
A.
,
Ming Fai
C.
,
Abdulmohsin Afan
H.
,
Sefelnasr
A.
,
Sherif
M.
&
El-Shafie
A.
2020
Application of non-parametric approaches to identify trend in streamflow during 1976–2007 (Naula watershed)
.
Alexandria Engineering Journal
59
(
3
),
1595
1606
.
https://doi.org/10.1016/j.aej.2020.04.006
.
Mallick
J.
,
Talukdar
S.
,
Alsubih
M.
,
Salam
R.
,
Ahmed
M.
,
Kahla
N. B.
&
Shamimuzzaman
M.
2021
Analysing the trend of rainfall in Asir region of Saudi Arabia using the family of Mann-Kendall tests, innovative trend analysis, and detrended fluctuation analysis
.
Theoretical and Applied Climatology
143
(
1
),
823
841
.
https://doi.org/10.1007/s00704-020-03448-1
.
Mann
H. B.
1945
Nonparametric tests against trend
.
Econometrica
13
(
3
),
245
259
.
JSTOR. https://doi.org/10.2307/1907187
.
Mohorji
A. M.
,
Şen
Z.
&
Almazroui
M.
2017
Trend analyses revision and global monthly temperature innovative multi-duration analysis
.
Earth Systems and Environment
1
(
1
),
9
.
https://doi.org/10.1007/s41748-017-0014-x
.
Onyutha
C.
2016
Statistical analyses of potential evapotranspiration changes over the period 1930–2012 in the Nile River riparian countries
.
Agricultural and Forest Meteorology
226–227
,
80
95
.
https://doi.org/10.1016/j.agrformet.2016.05.015
.
Patakamuri
S. K.
&
Das
B.
2019
trendchange: Innovative Trend Analysis and Time-Series Change Point Analysis
.
Rahman
M. A.
,
Yunsheng
L.
&
Sultana
N.
2017
Analysis and prediction of rainfall trends over Bangladesh using Mann–Kendall, Spearman's rho tests and ARIMA model
.
Meteorology and Atmospheric Physics
129
(
4
),
409
424
.
https://doi.org/10.1007/s00703-016-0479-4
.
Salehi
S.
,
Dehghani
M.
,
Mortazavi
S. M.
&
Singh
V. P.
2020
Trend analysis and change point detection of seasonal and annual precipitation in Iran
.
International Journal of Climatology
40
(
1
),
308
323
.
https://doi.org/10.1002/joc.6211
.
Şan
M.
,
Akçay
F.
,
Linh
N. T. T.
,
Kankal
M.
&
Pham
Q. B.
2021
Innovative and polygonal trend analyses applications for rainfall data in Vietnam
.
Theoretical and Applied Climatology
144
(
3
),
809
822
.
https://doi.org/10.1007/s00704-021-03574-4
.
Sen
P. K.
1968
Estimates of the regression coefficient based on Kendall's tau
.
Journal of the American Statistical Association
63
(
324
),
1379
1389
.
https://doi.org/10.2307/2285891
.
Şen
Z.
2012
Innovative trend analysis methodology
.
Journal of Hydrologic Engineering
17
(
9
),
1042
1046
.
https://doi.org/10.1061/(ASCE)HE.1943-5584.0000556
.
Şen
Z.
2014
Trend identification simulation and application
.
Journal of Hydrologic Engineering
19
(
3
),
635
642
.
https://doi.org/10.1061/(ASCE)HE.1943-5584.0000811
.
Şen
Z.
2017
Innovative trend significance test and applications
.
Theoretical and Applied Climatology
127
(
3
),
939
947
.
https://doi.org/10.1007/s00704-015-1681-x
.
Şen
Z.
2020
Probabilistic innovative trend analysis
.
International Journal of Global Warming
20
(
2
),
93
105
.
https://doi.org/10.1504/IJGW.2020.105387
.
Şen
Z.
,
Şişman
E.
&
Dabanli
I.
2019
Innovative polygon trend analysis (IPTA) and applications
.
Journal of Hydrology
575
,
202
210
.
https://doi.org/10.1016/j.jhydrol.2019.05.028
.
Serinaldi
F.
,
Chebana
F.
&
Kilsby
C. G.
2020
Dissecting innovative trend analysis
.
Stochastic Environmental Research and Risk Assessment
34
(
5
),
733
754
.
https://doi.org/10.1007/s00477-020-01797-x
.
Shadmani
M.
,
Marofi
S.
&
Roknian
M.
2012
Trend analysis in reference evapotranspiration using Mann-Kendall and Spearman's rho tests in arid regions of Iran
.
Water Resources Management
26
(
1
),
211
224
.
https://doi.org/10.1007/s11269-011-9913-z
.
Shokrian
F.
,
Saberi
A.
&
Sabbaq
A.
2022
Investigation of qualitative changes and determination of vulnerability zone of sari plain aquifer using SI model and Man-Kendall non-parametric test
.
Irrigation and Water Engineering
12
(
3
),
419
436
.
https://doi.org/10.22125/iwe.2022.146419
.
Singh
R.
,
Sah
S.
,
Das
B.
,
Potekar
S.
,
Chaudhary
A.
&
Pathak
H.
2021
Innovative trend analysis of spatio-temporal variations of rainfall in India during 1901–2019
.
Theoretical and Applied Climatology
145
(
1
),
821
838
.
https://doi.org/10.1007/s00704-021-03657-2
.
Sneyers
R.
1990
On the statistical analysis of series of observations
.
WMO Publ. 415
,
Technical Note 143, 192
.
Vezzoli
R.
,
Pecora
S.
,
Zenoni
E.
&
Tonelli
F.
2012
Data Analysis to Detect Inhomogeneity, Change Points, Trends in Observations: An Application to Po River Discharge Extremes (SSRN Scholarly Paper ID 2195345). Social Science Research Network. https://doi.org/10.2139/ssrn.2195345
.
Wu
S.
,
Zhao
W.
,
Yao
J.
,
Jin
J.
,
Zhang
M.
&
Jiang
G.
2022
Precipitation variations in the Tai Lake Basin from 1971 to 2018 based on innovative trend analysis
.
Ecological Indicators
139
,
108868
.
https://doi.org/10.1016/j.ecolind.2022.108868
.
Yenice
A. C.
&
Yaqub
M.
2022
Trend analysis of temperature data using innovative polygon trend analysis and modeling by gene expression programming
.
Environmental Monitoring and Assessment
194
(
8
),
543
.
https://doi.org/10.1007/s10661-022-10156-y
.
Yilmaz
B.
2019
Analysis of hydrological drought trends in the gap region (Southeastern Turkey) by Mann-Kendall test and innovative Şen method
.
Applied Ecology and Environmental Research
17
(
2
),
3325
3342
.
https://doi.org/10.15666/aeer/1702_33253342
.
Yuce
M. I.
&
Esit
M.
2021
Drought monitoring in Ceyhan Basin, Turkey
.
Journal of Applied Water Engineering and Research
9
(
4
),
293
314
.
https://doi.org/10.1080/23249676.2021.1932616
.
Yuce
M. I.
,
Esit
M.
&
Karatas
M. C.
2019
Hydraulic geometry analysis of Ceyhan River, Turkey
.
SN Applied Sciences
1
(
7
),
763
.
https://doi.org/10.1007/s42452-019-0800-1
.
Yue
S.
,
Pilon
P.
&
Cavadias
G.
2002
Power of the Mann–Kendall and Spearman's rho tests for detecting monotonic trends in hydrological series
.
Journal of Hydrology
259
(
1
),
254
271
.
https://doi.org/10.1016/S0022-1694(01)00594-7
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).