Evolution Characteristics and Relationship of Meteorological and Hydrological Droughts from 1961 to 2018 in Hanjiang River Basin, China

: In the context of global warming and increasing human activities, the acceleration of the water 3 cycle will increase the risk of basin drought. In this study, to analyze the spatial and temporal evolution 4 characteristics of hydrological and meteorological droughts over the Hanjiang River Basin (HRB); the 5 Standardized Precipitation Index (SPI) and Standardized Runoff Index (SRI) were selected and applied 6 for the period 1961-2018. In addition, the cross-wavelet method was used to discuss the relationship 7 between hydrological drought and meteorological droughts. The results and analysis indicated that: (1) 8 the meteorological drought in the HRB showed a complex cyclical change trend of flood-drought-flood 9 from 1961 to 2018. The basin drought began to intensify from 1990s and eased in 2010s. The 10 characteristics of drought evolution in various regions are different based on scale. (2) During the past 11 58 years, the hydrological drought in the HRB has shown a significant trend of intensification, 12 particularly in autumn season. Also, the hydrological droughts had occurred frequently since the 1990s, 13 and there were also regional differences in the evolution characteristics of drought in various regions. (3) 14 Reservoir operation reduces the frequency of extreme hydrological drought events. The effect of reducing 15 the duration and intensity of hydrological drought events by releasing water from the reservoir is most 16 obvious at Huangjiagang Station, which is the nearest to Danjiangkou Reservoir. (4) The hydrological 17 drought and meteorological drought in the HRB have the strongest correlation on the yearly scale. After 18 1990, severe human activities and climate change are not only reduced the correlation between 19 hydrological drought and meteorological drought in the middle and lower reaches of the basin, but also 20 reduced the lag time between them. Among them, the hydrological drought in the upper reaches of the 21 basin lags behind the meteorological drought by 1 month, and the hydrological drought in the middle and 22 lower reaches of the basin has changed from 2 months before 1990 to 1 month lagging after 1990. 23 results showed that the SPI index has good applicability in humid and sub- humid regions. Gümüşsoy studied the weather of the Beyşehir Lake in Turkey based on the SPI index. Drought conditions, the results show that the SPI index has a better application effect in this semiarid-humid area. The HRB flows through China's Shaanxi Province and Hubei Province.


26
Drought is a globally and common natural disaster, which has a huge impact on human activities and life. 27 In recent years, it has occurred more frequently and causing more serious harm to humans in the context 28 of global climate change and the increasing impact of human activities (Dai et al. 2018; Wyckoff and 29 Bowers 2010). In 2014, the fifth assessment report of the IPCC pointed out that the global impact of 30 drought will be further expanded and extended (Allen et al. 2007). Therefore, drought research has further 31 become a hot issue of concern to scholars worldwide. As the American Meteorological Society has 32 integrated various definitions of drought and divided drought into four types: meteorological drought, 33 hydrological drought, agricultural drought, and socioeconomic drought. (Mishra and Singh 2010)Among 34 them, meteorological drought refers to the phenomenon of water deficit caused by the imbalance of 35 precipitation and evaporation in a certain period of time. It is characterized by insufficient precipitation, 36 and the absolute precipitation of a specific duration is used as a quantitative indicator to determine the 37 degree of drought. Hydrological drought refers to water shortages in which river runoff, reservoirs or 38 groundwater. It affects the three major links of evaporation, infiltration and runoff in the natural water 39 cycle process, involving the surface, soil and groundwater interface, and related to the hydrological cycle 40 and water balance, so it can better reflect the actual drought and the difficulty of drought resistance.

Data source 122
In this study, the meteorological data required are the daily precipitation data of 235 data grids of 123 0.25°×0.25° in the HRB for the period of 1961 to 2018, this data accessed from the CN05.1 dataset 124 provided by China Meteorological Administration(CMA, http://data.cma.cn), that were generated by the 125 interpolation of more than 2400 weather stations in China through anomaly approximation method. The 126 data is of good quality and is widely used in regional climate characteristics and model performance Committee. It has undergone strict data inspection and quality control, which meets the needs of research. 131 The schematic diagram of the HRB and the locations of meteorological grids, hydrological stations are 132 shown in Fig. 1.  133 3 Methodology 134 First, in the studying of temporal characteristics, this paper uses the Mann-Kendall test method to analyze 135 the mutation and trend characteristics of SPI/SRI sequences at different scales. For periodic 136 characteristics, the authors use the Molert wavelet as the mother wavelet and combine the wavelet power 137 spectrum method to analyze the periodicity of SPI/SRI sequences at different time scales. Then, in the 138 research of spatial characteristics, the run theory is used to calculate the drought evaluation index. Finally, 139 the cross-wavelet power spectrum method and the time-lag cross-correlation analysis method are used to 140 analyze the relationship between hydrological drought and meteorological drought. 141

Standardized Precipitation/Runoff Index (SPI/SRI) 142
The accurate description of drought events by the drought index is the basic requirement of drought  It belongs to a subtropical monsoon climate zone. The climate in the basin is relatively mild and belongs 149 to a humid and semi-humid area. Considering the integrity of comprehensive data, this paper finally 150 chose to use the SPI and SRI index to describe the meteorological and hydrological drought in the HRB. 151 The Standardized Precipitation Index (SPI) was used to measure the degree of meteorological 152 drought, which is to calculate the probability of the corresponding distribution function of the 153 precipitation in a certain period, and then carry out the normal standardization process, finally use the 154 standardized precipitation cumulative frequency distribution to divide the drought level . 155 Since the standardization process can eliminate the difference in the temporal and spatial distribution of 156 precipitation, SPI can well reflect the drought situation in the region under different research scales (Zhao 157 et al. 2020). The calculation formula is as follows (Caccamo et al. 2011;McKee T B 1993;Qi et al. 2018): 158 First, suppose that the precipitation(flow) sequence during the study period is x, then the probability 159 function of this sequence is: 160 Where x>0，>0，>0;  and  are the scale parameter and shape parameter of the above probability 162 distribution function, respectively. The parameter estimation method adopts the maximum likelihood 163 method to obtain: 164 Where i x is the period precipitation, x is the average value of the period precipitation. 168 After the above parameters are calculated, combined with formula (1), according to statistical 169 principles, when the precipitation rate in a certain period of time is 0 x , the probability of the time when 170 the rate of the period precipitation x< 0 When the i x =0, there are: 173 , m is the number of periods when i x =0, n is the total length of the sequence. 175 Then, combining equations (5) ~ (6), and substituting the solved probability values into the 176 standardized normal distribution function respectively: 177 At last, the calculation formula of SRI is obtained by the integration of the formula (7): 179 x is the probability of the precipitation distribution associated with 181 the  function. when ( ) 0.5 SRI was used to represent the hydrological drought in this study because its calculation steps and 185 advantages are similar to those of the SPI, the SPI calculation formula rainfall is replaced with flow rate 186 for calculation ). The SPI and SRI classifications of the drought levels are listed in Table  187 1 and correspond to classifications used in previous studies (Hao et al. 2016). 188 respectively. The specific processes of the two calculation methods are as follows (Kendall 1990;Mann 198 Among them, the order column is the sum of the number at the i-th time greater than the number at 220 the j-th time.
According to the above formula, using the SPI sequence and SRI sequence as data input, then the 231 UFk and UBk curves of the studied time series are calculated. If the value of UF is greater than 0, it 232 indicates that the sequence is on an upward trend, which means the drought become serious; and if the 233 value is less than 0, the sequence is on a downward trend, which means the drought has a tendency to

Time-Delay Cross-Correlation Analysis 258
The time-lag cross-correlation analysis is a method of moving a certain time series backward for a period Assuming that the 2 time series x and y have correlation with any lag time k, then the time lag 263 correlation coefficient of the studied series is expressed as: k C x y k  is the covariance of the x series and the time series y+k with increasing lag time 266 k, and x y k    is the mean square error of the two researched series, respectively expressed as:  research results, the HRB transitioned from the rainy period in the 1980s to the drier period in the 1990s, 295 the precipitation has shown an insignificant downward trend in the past 60 years, and the SPI value is 296 directly affected by precipitation. Therefore, the non-significant abrupt change of the scale is also more 297 consistent with the precipitation changes in the HRB. Besides that, due to the calculation difference and 298 time generalizability of different scales, the mutation points of varied scales are different. 299

Results and Discussion
The SPI-1, SPI-3, and SPI -12 changes in the HRB from 1961 to 2018 showed certain periodic 300 characteristics (Fig. 4). It can be seen that the frequency domain scale of the energy center in the wavelet 301 power spectrum does not pass the 0.05 significance level, so the periodicity of SPI-1 is not obvious. On 302 the contrary, the frequency domain scale of the energy center and the extreme value of the wavelet 303 variance of the SPI-3 reaches the extreme value is about 2 years, and it has passed the 0.05 significance 304 level test, so the periodicity of SPI-3 is significant. The periodicity of SPI-12 is more significant, the 305 frequency domain scale of the energy center is mainly concentrated over 4 years, and the SPI-12 value 306 oscillates most strongly at 4 years. Because the time scale of monthly SPI is relatively meticulous, and 307 the sequence fluctuation noise is obvious, which leads to insignificant periodic changes. However, the 308 seasonal and yearly scale SPI eliminates some noises relative to the monthly scale sequence, so 309 significant periodic characteristics are identified (Niu and Chen 2016). 310

Spatial Evolution Characteristics 311
In order to facilitate the subsequent exploration of the relationship between hydrological drought and 312 meteorological drought, the HRB is divided into four spatial areas according to the location of 313 hydrological stations. Below Huangzhuang Station is Area I, above is Area II, and above Huangjiagang 314 Station is Area III, and Baihe Station above is area IV. And Region I correspond to the downstream of 315 the basin, Region II corresponds to the middle reaches of the basin, and Regions III and IV correspond 316 to the upper reaches of the region. 317 The spatial characteristics of the SPI-12 meteorological drought of the HRB are shown in Fig .5. At 318 the yearly scale, the HRB shows a clear trend of aridification in the central region but humidification in the southern and western regions (Fig .5a). The significant aridification areas are mainly distributed in 320 Region II, the proportion of significant aridification grid points in the region is 50.7%, the proportion of 321 aridification grid points is 87.32% of Regions III. And the significant humidification regions are mainly 322 distributed in regions I and IV, where the significant humidification grids in the region is 47.67% and 323 66.7% respectively. From the point of view of the frequency and duration of drought ( Fig .5c & Fig .5d), 324 the frequency of drought events in Region IV is the highest at 38.01%, while the drought duration is the 325 longest at the same time, with an average drought duration of 11.26 months, and it is also with the most 326 occurrence of light drought events. Among them, the frequency of drought in Region I is relatively low, 327 and its duration is the shortest, but the frequency of severe drought and extreme drought events is 328 relatively highest. Overall, although the frequency of drought in Region III is not high, its drought 329 intensity is the highest, and the drought duration is second only to Region IV. It is necessary to prevent

337
From seasonality perspective, the spatial characteristics of meteorological drought in the HRB have 338 obvious differences and variability between the four seasons (Fig .6). In spring, the trend of aridification 339 in areas III and IV in the upper reaches of the HRB is more obvious, but the proportion of significant 340 aridification in the area is relatively low, only 2.87%, where the frequency of drought in area III is higher, 341 light droughts and moderate droughts are occur frequently, and the drought duration is long and drought 342 intensity is high. In summer, the trend of aridification in Region II was more obvious, and the drought 343 frequency was 33.58%. Region I had the highest drought intensity at 1.206, and Region IV had the longest 344 drought duration at 2.046 months. In autumn, the entire basin showed a drought trend, where Region II 345 was the most significant, the drought in Region III was long and strong, the frequency of drought in 346 Region IV was high, and the frequency of moderate drought events was relatively highest. In winter, the 347 aridification trend of area IV is the most significant, and the drought intensity is the highest, area III has 348 the highest drought frequency and the longest drought duration. Taken together, the four seasons of  The M-K test method was used to understand the trend of the hydrological drought index SRI series at 358 seasonal and yearly scales for the 5 hydrological stations within the HRB ( Table 2). The yearly 359 hydrological drought index trend recorded a significant decreasing trend, that is, the aridification trend 360 became more obvious, and the aridification trend of the area near HJG was the most obvious. At a 361 seasonal scale, the spring and autumn droughts aggravated in the entire river basin. The spring droughts 362 at SQ, AH, and BH stations in the upper HRB have a significant trend. The autumn droughts in the control 363 basins of stations other than the SQ station also showed a trend of aridification. And the increasing trend 364 of summer drought in HJG region is more obvious, and it is prone to continuous drought in summer and 365 autumn, while winter in the whole basin shows a trend of insignificant humidification to varying degrees. 366 Wang et al (Wang et al. 2020c) found that the spring drought in Shaanxi Province tended to increase. and 367 different degrees of aridification trends in spring and autumn in recent years, and SQ, AK, and BH all 369 have increased trends in spring and autumn droughts, but the increase in autumn drought at Shiquan 370 Station is not obvious. This is consistent with the conclusion of this paper. 371 Similarly, taking the HJB Station in the middle reaches of the basin as an example, the hydrological 374 drought in the basin also has obvious temporal and periodic characteristics (Fig .7). greater than that on the monthly scale. This is because the difference in data processing during SRI and 381 SPI calculations. The wavelet power spectrum results show that the periodicity of the monthly-scale SRI 382 series in the HRB is not significant, and the yearly and seasonal SRI series have a main period of 8 years 383 (Fig .8).

404
The black solid line on the left picture is the wavelet influence cone, which represents the wavelet spectrum area generated by the

423
From the perspective of drought duration and drought intensity (Fig .10), the Baihe Station has 424 reported the highest drought intensity except during summer; the Shiquan Station has the longest drought 425 duration in spring and summer, especially with the largest drought intensity recorded in summer. The 426 Huangzhuang Station has the longest drought duration in autumn and winter but the drought intensity is 427 not large. It is worth noting that the drought intensity of Huangjiagang Station is the lowest, and also the 428 lowest drought intensity of the light drought in all four seasons. This could be attributed to the release of 429 water during the dry season from the reservoir which has a certain weakening effect on its drought 430 intensity, which lessen the seasonal drought at Huangjiagang Station to a certain extent . 431 432 Fig. 10 The drought duration and drought intensity of five hydrological stations at the seasonal scale in HRB. Table 3 compares the characteristics of hydrological drought events with the corresponding drought 434 characteristics of meteorological drought events at seasonal scale. Except for summer, the frequency of 435 meteorological droughts at other scales is higher than the frequency of hydrological droughts, and the 436 frequency of extreme meteorological droughts events is significantly higher than that of hydrological 437 droughts. In summer, rainfall is abundant, and reservoir storage will increase the occurrence of 438 hydrological droughts. When droughts occur in other seasons, the release of water from the reservoir 439 reduced the occurrence of hydrological drought and also reduced the occurrence of extreme drought 440 events, which further proved that human activities that represented by reservoir water transfer and storage 441 have obvious impacts on hydrological drought in the basin(Xiaoli Yang and Zhongwang Wei 2020). 442

433
The meteorological drought duration and drought intensity of the basin are less than the hydrological 443 drought duration except for Huangjiagang Station, indicating that the release of water from the reservoir 444 in the dry season has a mitigation effect on the later hydrological drought events downstream. In general, 445 the increasing trend of hydrological drought in the basin cannot be ignored, human activities represented 446 by reservoirs have a significant impact on hydrological drought. On the one hand, water storage in the 447 rainy season will increase the frequency of hydrological drought events resulted from increasing water, 448 capturing; while the release of water during the dry season will reduce the occurrence of hydrological 449 drought events. The effect of reducing the duration and intensity of hydrological drought events by 450 releasing water from the reservoir is most obvious at Huangjiagang Station, which is the nearest to 451 Danjiangkou Reservoir (Jiao et al. 2020). 452 Table 3 The comparison of drought characteristic on meteorological drought and hydrological drought events

Correlation between Hydrological Drought and Meteorological Drought 457
Meteorological drought mainly refers to the phenomenon of insufficient regional water caused by the 458 imbalance of atmospheric precipitation and evaporation. It is sudden and can start or end quickly.
Meteorological drought is directly affected by meteorological factors such as precipitation and 460 evaporation, while hydrological drought is different that may be affected by hydrological features. When 461 meteorological drought occurs, human demand for water to production and living causes continuous 462 reduction of surface water and groundwater level, if this situation continues to a certain extent, 463 hydrological drought formed(B et al. 2020). Therefore, the occurrence of hydrological drought is largely 464 affected by meteorological drought, and the two have a certain correlation. 465 In order to further explore the impact of human activities on the drought situation in the HRB, 488 combined with the mutation points of runoff, hydrological drought and related documents, 1990 was 489 selected as the cut-off point, 1961-1990 as the base period, and 1991-2018 as the variation period, then 490 calculate the correlation coefficients of SPI series and SRI series at various scales and different periods 491 in the HRB (Fig .11b). The results show that the correlation between hydrological drought and 492  Therefore, the authors use this method to further analyze the correlation between meteorological drought 507 and hydrological drought in the HRB. Similarly, taking Huangjiagang Station in the middle reaches of 508 the HRB as a representative station, Fig .12 shows the cross-wavelet power spectrum of the SPI sequence 509 and SRI sequence at the monthly, seasonal, and yearly scales at Huangjiagang Station. The Fig .12 shows 510 that both SPI and SRI have a strong correlation at different scales. At the monthly scale, the SPI-1 511 sequence and the SRI-1 sequence have an obvious resonance period in the mid-frequency region, and SRI is significantly improved during the significant resonance period based on the cross wavelet 527 transform, which illustrates the importance of considering different time scales when studying the 528 relationship between hydrological drought and meteorological drought. In general, there is a positive 529 correlation between hydrological drought and meteorological drought at different time scales in the HRB, 530 and hydrological drought is obviously lagging behind meteorological drought, but the synchronization 531 of different scales is not consistent. The specific lag time will be discussed further. 532 Fig. 12 The crossed wavelet power spectrum of monthly, seasonal and yearly SPI and SRI Sequences at Huangjiagang Station.

534
The black solid line represents the wavelet influence cone, the color depth represents the strength of the correlation, the darker 535 the color, the stronger the correlation, the black arrow forward represents advance, and backward represents lag.

536
The occurrence of hydrological drought lags behind meteorological drought, and the correlation 537 between meteorological drought and hydrological drought on the yearly scale is the best. Therefore, the 538 authors taking the yearly scale as an example, where they use the SPI-12 sequence of the same period, 539 lag 1 month, lag 2 months...12 months as the time gradient, calculate the time lag correlation coefficient 540 corresponding to different time gradients, and take the time gradient corresponding to the maximum 541 value of the correlation coefficient as the lag time for SRI to SPI. Fig .13 show that there is a certain lag 542 time between the hydrological drought and meteorological drought in the HRB on a yearly scale. The lag 543 time between the hydrological drought of different hydrological stations and the meteorological drought 544 of the control basin is different. The lag time in the upper HRB hydrological drought lags behind the 545 meteorological drought is about 1 month, while the hydrological drought in the middle and lower reaches 546 lags the meteorological drought by 2 months. 547 548 Fig. 13 The variation of lag time between meteorological and hydrological droughts on yearly scale in HRB. The month with the 549 highest correlation coefficient represents the strongest correlation between hydrological drought and meteorological drought that 550 lags behind this month, that is, the time that hydrological drought lags behind meteorological drought.

551
In order to explore the impact of human activities and climate change on the response of hydrological 552 drought to meteorological drought in the middle and lower reaches of the HRB, similarly, the authors 553 used 1990 as the differentiation point, 1961-1990 as the base period, and 1991-2018 as the variation 554 period. The response changes of the HRB in different periods of hydrological drought and meteorological 555 drought on different scales are showed (Fig .14). It shows that the lagging effects of human activities and 556 climate change on the monthly hydro-meteorological drought are not obvious, and the lag time before and after the variation remains unchanged, while for the seasonal and yearly scales, it significantly 558 reduced in the Huangjiagang station located in the middle and lower reaches of the HRB. The 559 hydrological drought lag time at the Huangjiagang station and Huangzhuang station decreased from 2 560 months lag in the reference period to 1 month lag, which is inseparable from the human activities 561 represented by the Danjiangkou Reservoir located upstream of the two stations. The primary indicator of 562 the occurrence of meteorological drought is insufficient precipitation, which leads to a decrease in 563 watershed runoff. However, the water intake, water use, and water transfer projects surrounding human 564 activities in the basin continue to reduce water volume. The consequence of shortage is that the runoff of 565 the watershed drops rapidly and the hydrological drought advances occurs, that is, the hydrological 566 (2) The hydrological drought in the Hanjiang River Basin showed a significant and serious trend. 586 The autumn drought was the most obvious, mainly in the Huangjiagang area; and followed by the spring 587 drought, which was mainly observed manifested in Shiquan Station and Ankang Station, the summer 588 drought was observed mainly manifested in Huangjiagang Station and Huangzhuang Station. In the area 589 near Huangzhuang Station, the HRB in winter shows a trend of humidification as a whole, and the 590 hydrological drought in the basin has an 8-year cycle. The hydrological drought frequency in the 591 Hanjiang River Basin gradually decreases from upstream to downstream along the main stream. But the 592 abnormal increase of hydrological drought frequency at Huangjiagang Station, 10km downstream of 593 Danjiangkou Reservoir, is closely related to human activities. Drought events in the river basin are mostly 594 light and moderate drought events. The upper and middle reaches of the Hanjiang River Basin have high 595 drought intensity, and the lower reaches have a longer drought duration. The drought at Huangzhuang 596 Station lasted the longest. 597 (3) In the past ten years, the increasing trend of hydrological drought in the river basin cannot be 598 ignored. Human activities represented by reservoirs have a significant impact on hydrological drought. 599 Water storage during the rainy season will increase the frequency of hydrological drought events, while 600 water release during the dry season will reduce hydrological drought events. The effect of reducing the 601 duration and intensity of hydrological drought events by releasing water from the reservoir is most 602 obvious at Huangjiagang Station, which is the nearest to Danjiangkou Reservoir. 603 occurs in arid and semi-arid areas, but also occurs in humid and semi-humid areas, and it often causes 614 more serious damage due to people's neglect (Huang et al. 2013). As the water source of the South-to-615 North Water Diversion, the Hanjiang River is located in a humid-semi-humid area, so the occurrence of 616 drought events has a great impact on the water supply security of the basin. Many scholars have also 617 conducted research on the drought in the Hanjiang River Basin, the results show that the drought situation 618 in the Hanjiang River Basin is present phase changes, and the occurrence of extreme events has increased, 619 which is consistent with the conclusion of this paper (Chen et al. 2013;Xu et al. 2011). In the process of 620 meteorological and hydrological drought in the study area, many studies have shown that there is a certain 621 correlation between meteorological drought and hydrological drought, and climate change and human