RCCC-WBM Model for Calculating the Impact of Abrupt Temperature Change and Warming Hiatus on Surface Runoff in China

Abstract

The global climate shows an obvious warming trend, and the impact on water resources is increasing. Abrupt temperature change and warming hiatus are two important states of temperature change. The quantitative impacts of temperature change and warming hiatus on surface runoff remain unclear. Based on the measured runoff data from 60 representative hydrological stations in China from 1956 to 2016 and the Water Balance Model developed by the Research Center for Climate Change (RCCC-WBC), this paper analyzes the quantitative impacts of abrupt temperature change and warming hiatus on surface runoff. The results showed that the effects of three types of abrupt temperature changes on runoff in different basins in China are significantly different. The effects of abrupt temperature changes and warming stagnation on runoff in northern China are greater than those in southern China, and the effects of abrupt temperature changes and warming stagnation on runoff in the upper, middle, and lower reaches of the same basin are also different. Before the abrupt change in temperature, the influence of temperature on the surface runoff was less than 9%, and the influence of temperature on the runoff in some southern areas was weaker, only affecting less than 3% of the runoff. When the temperature changes abruptly, the influence of air temperature on the surface runoff in a small part of the arid region is up to 30%. The abrupt change in mean maximum temperature has both positive and negative driving effects on runoff in China, and the negative driving effect is concentrated in the areas with abrupt warming, affecting about 8% of the runoff on average. The average influence of abrupt mean temperature change on runoff in China is about 10%, and the area with a large influence on runoff change is concentrated in the area north of 40° N. The abrupt change in temperature in the middle and lower reaches of the Yellow River Basin has a great influence on the runoff change, up to 13%. The maximum impact of abrupt mean minimum temperature on runoff is concentrated in Northeast China, ranging from 9% to 12%. During the period of temperature stagnation, air temperature and runoff showed an obvious reverse trend. During this period, the average negative influence of drastic changes in air temperature on runoff was about 15%, but precipitation and runoff still maintained a good consistency, which may be due to the effect of other influencing factors which offset the negative driving effect of air temperature.

1. Introduction

In recent years, with rapid economic development, the impacts of climate change and human activities on the hydrological cycle have become increasingly severe, which has resulted in a certain change in the spatiotemporal distribution of hydrometeorological elements [1], while the impacts of abrupt temperature change and warming hiatus on surface runoff remain unclear and urgently need to be elucidated.

Changes in river runoff are affected not only by climate change but also by human activities [2]. However, climate change and human activities influence each other and are not independent. On the one hand, climate change has led to the intensification of human activities; on the other hand, human beings have made corresponding changes to adapt to climate change [3]. However, as these effects cannot be strictly divided, the impact of climate change on runoff can only be separated from that of human activities [4]. Under the premise of the independence of both factors, how to quantitatively distinguish the impacts of climate change and human activities on runoff is a hot topic researched by hydrologists. The research methods for quantitatively separating and judging the impacts of climate and human activities on runoff include long-series data comparative analysis [5], comparative test analysis [6], itemized computation combination [7], the elastic coefficient method [8], hydrological simulation, and other methods [9,10].

There have been many quantitative analyses of impacts of climate change and human activities on surface runoff, and the involved study areas include typical watersheds worldwide [11]. The results show that the increase in spring temperature in the western United States is the leading cause of the decrease in spring runoff [12]. The observed runoff trend in the Liard River in northern Canada is related to both meteorological variables and a large-scale oceanic and atmospheric process [13]. There is a certain relationship between runoff changes in California and precipitation and temperature [14,15]. Through analysis of the impact of the Colombian hydrological cycle, it has been pointed out that the annual average measured flow at the Dalles station has decreased by 16.5%, of which 8% to 9% is due to climate change and 7% to 8% is due to irrigation water. The impact of climate on runoff is less than that on erosion, and total sediment transport has decreased by 50%, with only a small component caused by climate change [16]. In China, taking the years from 1963 to 1972 as the base period, the contributions of precipitation to runoff reduction from 1973 to 1999 and 2000 to 2014 were 40.7% and 41.61%, respectively [17]. The variation in runoff into the Nenjiang River from 1956 to 2006 has been characterized by the uneven distribution of annual runoff, and interannual runoff has declined. The main factor affecting runoff is precipitation, followed by evaporation. In addition, the impact of human activities on runoff has also become more obvious [18]. The decrease in summer and autumn runoff in the Yanhe River Basin from 1952 to 2008 was mainly due to a decrease in precipitation in addition to nearby human activities. The increase in winter runoff was mainly due to elevated winter precipitation and increased snowmelt runoff caused by climate warming [19]. Runoff in the Yalong River Basin has shown an increasing trend, and annual average runoff has been projected to increase by 8.9%, 12.5%, and 16.7% under three emission scenarios. Projections of runoff for 2020, 2050, and 2080 varied [20]. If precipitation remains unchanged and the temperature increases by 1 °C, runoff into the Urumqi River and Kaidu River has been projected to increase by 0.24% and 1.90%, respectively [21]. Both climatic factors and human intervention appear to be responsible for the alteration in flow pattern of the Ganga River [22]. Changes in the discharges in the Jhelum River are due to climatic change and anthropogenic activities in the basin [23].

In such a case, it is difficult to perform comparative studies on the extent of impacts on runoff of factors operating over broad spatial scales, since previous studies have employed different methods and periods of comparison. There are currently a few studies on the impacts of climate change and human activities on runoff changes across China that have been performed to evaluate the effects of climatic changes (precipitation and evaporation) and human activities on changes in runoff. It is well known that precipitation is one of the most direct factors affecting runoff [24]. Human activities have also greatly affected runoff in recent years, while temperature has a relatively small impact on runoff. What are the impacts of abrupt temperature change and warming hiatus on runoff? No relevant research has been found. In recent years, in some natural watersheds that are less affected by human activities, runoff trends have not been consistent with precipitation trends. When precipitation has increased or decreased for many years, runoff has shown an opposite trend [25]. What are the effects of abrupt temperature change and warming hiatus on these patterns, and can they provide a basis for revealing this phenomenon? On this basis, the present study performs a quantitative analysis of the effects of abrupt temperature change and warming hiatus on runoff by choosing regions of China as study objects and 60 representative runoff stations and provides a reference for the reasonable development and utilization of water resources, thus being of certain scientific and practical significance.

2. Profiles of the Regions of Interest, Data, and Methods

2.1. Profiles of the Regions of Interest

China is located in the east of Asia, on the west coast of the Pacific Ocean (Figure 1), with a land area of about 9.6 million square kilometers, more than 18,000 km of mainland coastline, more than 14,000 km of island coastline, and more than 4.7 million square kilometers of inland and border waters. The climate is complex and varied, ranging from temperate monsoon, tropical monsoon climate, and tropical rainforest to temperate continental climate and highland Montane, from south to north across the tropical, subtropical, warm temperate, moderate temperate, and cold temperate climate zone.

2.2. Data Sources

The meteorological data used in this study are from the China meteorological data network (http://data.cma.cn/, accessed on 30 August 2022), and finally 622 meteorological stations distributed in China were selected (Figure 1), which can represent the temperature situation in the study area.

Sixty representative hydrologic stations were selected in China. The principle of selection mainly considered that hydrologic stations have relatively complete long series of measured data, and the controlled watershed area of hydrologic stations should cover the whole watershed of China as much as possible and be evenly distributed in the upper, middle, and lower reaches of each river. The annual changes in runoff at these hydrological stations can represent both the regions with large human activities and natural basins with small human activities. Based on the above principles, 60 representative hydrological stations were selected in China, monthly measured runoff data from these stations were collected from their establishment to 2016, and the runoff data were unified from 1956 to 2016. Among them, 5 hydrological stations were established late, and the lack of measured runoff was interpolated. The data came from the China water conservancy system.

2.3. Data Processing and the Method Employed

(1) For the unification of temperature time series, the missing measurement data were subjected to interpolation extension through correlation and regression analysis [26].

(2) For the meteorological data from the runoff stations corresponding to the catchment area, the inverse distance weighting method was used to spatially interpolate the precipitation, air temperature, and evaporation. Each meteorological element was extracted to a grid of 100 m × 100 m, and then the Zonal Histogram tool in ArcGIS was used to extract the average value of the meteorological element corresponding to each watershed.

(3) The Mann–Kendall nonparametric statistical method was used to check abrupt temperature changes and changes in runoff [27].

(4) The years of warming hiatus after abrupt temperature change were identified based on previous studies [28] through an analysis combining a temperature series and its phased trend lines with a 3–5-year sliding value series and its phased trend lines. When the temperature changed abruptly, the rate of temperature change reached its relative maximum; a year was identified as a year of warming hiatus when the temperature change was less than or equal to 0.1 °C/decade from that year to the end of the series (2018) or to a certain year before.

(5) The tendency rate method was used for the time series trend analysis of temperature runoff.

(6) The Penman formula was used to estimate the potential evaporation in China.

(7) For the quantitative identification of runoff change attribution, the quantitative identification method of runoff change attribution based on hydrological simulation was adopted, which includes three parts:

The Mann–Kendall nonparametric statistical method is an effective method for diagnosing sequence variation (mutation) at present and has been widely applied in the analysis of variation (mutation) in hydrological and meteorological sequences [29].

Water volume simulation model (Figure 2). The Water Balance Model developed by the Research Center for Climate Change (RCCC-WBM) of the Ministry of Water Resources of China was selected to simulate the process of reducing natural runoff in the basin. Compared to other models [30,31], this model is a large-scale hydrological model, with the advantages of a simple structure, few parameters, easy comprehension, and low data requirements [32]. The model takes the month as the unit of time, and the inputs include monthly precipitation, surface evaporation, and air temperature. The simulated runoff includes three components, namely, surface runoff, subsurface runoff, and snowmelt runoff [33]. For small-scale basins, the model can simulate the overall hydrological process based on the average values of meteorological elements of the basin; for medium-to-large-scale basins, the model can divide the study basins into grid points and conduct distributed simulations of hydrological processes [34]. The Nash–Sutcliffe model efficiency (NSE) coefficient and the relative error (RE) of the simulated total volume were selected as the objective functions for model parameter optimization [35].

Cause analysis model. The hydrological data before disturbance from human activities (denoted “natural reference period”, assuming no obvious human activity) were used to determine the parameters of the hydrological model, and the meteorological data series after disturbance from human activities (after an abrupt change point, referred to as a “period of significant human activities”) and the model parameters determined by the data from the “natural reference period” were used to simulate and calculate the runoff process during the “period of significant human activities”. The total runoff change before and after an abrupt change point of the hydrological series includes two parts: the influence of changes in climate elements and the influence of human activities. The difference in natural runoff before and after the abrupt change point is caused by climate change, and the difference between the measured runoff process and the simulated natural runoff (calculated based on the parameters determined by the natural reference period) during the period of significant human activities quantifies the influence of human activities. Therefore, the two key links in the quantitative identification method of runoff change attribution are the diagnosis of a point of abrupt change in runoff (i.e., the identification of the natural reference period and period of significant human activities) and the restoration of natural runoff [36]. The quantitative identification model of runoff change attribution is as follows:

$$\Delta {\mathit{W}}_{\mathit{T}}={\mathit{W}}_{\mathit{H}\mathit{R}}-{\mathit{W}}_{\mathit{B}}$$

$$\Delta {\mathit{W}}_{\mathit{H}}={\mathit{W}}_{\mathit{H}\mathit{R}}-{\mathit{W}}_{\mathit{H}\mathit{N}}$$

$$\Delta {\mathit{W}}_{\mathit{C}}={\mathit{W}}_{\mathit{H}\mathit{N}}-{\mathit{W}}_{\mathit{B}}$$

$${\epsilon }_{\mathit{H}}=\frac{\Delta {\mathit{W}}_{\mathit{H}}}{\Delta {\mathit{W}}_{\mathit{T}}}\times 100\%$$

$${\epsilon }_{\mathit{C}}=\frac{\Delta {\mathit{W}}_{\mathit{C}}}{\Delta {\mathit{W}}_{\mathit{T}}}\times 100\%$$

where $\Delta {\mathit{W}}_{\mathit{T}}$ is the total runoff change; $\Delta {\mathit{W}}_{\mathit{H}}$ and $\Delta {\mathit{W}}_{\mathit{C}}$ are the changes in runoff caused by human activities and climate change, respectively; ${\mathit{W}}_{\mathit{B}}$ is the runoff during the reference period; ${\mathit{W}}_{\mathit{H}\mathit{R}}$ and ${\mathit{W}}_{\mathit{H}\mathit{N}}$ are the measured and restored (simulated) runoff, respectively, during the period of influence from human activities; and ${\epsilon }_{\mathit{H}}$ and ${\epsilon }_{\mathit{C}}$ are the contributions of human activities and climate change, respectively, to the change in runoff.

The whole river system of China is composed of the basins of seven rivers, i.e., the Changjiang River, Yellow River, Songhua River, Liaohe River, Haihe River, Huaihe River, and Pearl River. The seven major drainage basins are further divided by their integrity, spatial distribution of water resources, and geographical characteristics into ten water resource regions, i.e., the Songhua River Region, the Liaohe River Region, the Haihe River Region, the Yellow River Region, the Huaihe River Region, the Changjiang River Region, the Region of Southeastern Rivers, the Pearl River Region, the Region of Southwestern Rivers, and the Region of Northwestern Rivers. According to the second water resources assessment for China, the mean volume of surface water resources in China was 2,738,800 million m³ during the period of 1956–2000, of which the surface water resources in the seven river regions (those of the Changjiang River, Yellow River, etc.) accounted for 64.9%. With 60 runoff stations throughout China as the study object, the present study analyzes the characteristics of variation in surface water resources in China over the past 61 years.

3. Analysis and Results

3.1. Multiyear Variation in Runoff

Figure 3 shows the multiyear rate of change in measured runoff at representative hydrological stations; from 1965 to 2016, the runoff measured at most representative hydrological stations in China decreased significantly. Generally, the measured runoff in the northern basins declined significantly faster than that in the southern region. Among these stations, the runoff measured at 21 representative hydrological stations increased for many years. Most of the hydrological stations measuring increased runoff are located in the southern region of the Yangtze River. The fastest rate of increase was 3 mm/10 years, with a nonsignificant increase. The multiyear runoff in the Songhua River Basin decreased the fastest overall, and the fastest rate of decrease was 6 mm/10 years, while the measured runoff in the middle and lower reaches of the Yellow River and the middle reaches of the Changjiang River also decreased rapidly on a multiyear scale; the runoff measurements at some stations in the southeastern coastal water system, the southwestern water system, the Huai River Basin, and the Pearl River Basin increased slightly over the years. On the whole, due to the construction of water conservancy projects and the increase in water consumption for industrial and agricultural development in China, human activities have greatly impacted stream runoff in the 21st century, with such impact principally reflected in runoff reduction. The rivers in the northern arid region are more vulnerable to human activities, so the multiyear reduction in their surface runoff is even more significant.

3.2. Diagnosis of Points of Abrupt Change in Measured Runoff in China

The river hydrological process is the response of a basin under the combined effects of climate change and human social activities. When obvious human activities occur in a basin, the natural process of river hydrology is destroyed, with corresponding gradual or saltatory (abrupt) change. The M-K test was used to check for abrupt changes in annual runoff (Figure 4). The results showed that the runoff measured at some representative hydrological stations in China underwent abrupt change. Abrupt change in surface runoff occurred earliest in the Haihe River Basin between 1962 and 1971, followed by the Songhua River Basin and the Liaohe River Basin, and abrupt changes were concentrated in the period from 1971 to 1984. Generally, most hydrological stations in China began to undergo abrupt changes in the 1980s and 1990s, and the times of abrupt change in runoff were slightly earlier in the northern region than in the southern region. The times of abrupt change were latest in the middle and upper reaches of the Yangtze River Basin and the Pearl River Basin, where they were concentrated after the 1990s. The results of diagnosis of the abrupt change points were basically consistent with the actual human activities in each basin. Under the changing environment, the overall river runoff in China continued to decrease, and the imbalance between the supply and demand of water resources in the northern region was more prominent.

To compare the contributions of climate change and human activities to changes in runoff measured at hydrological stations in different basins in China, this paper divides the reference period with no obvious abrupt changes from the period of influence in a united manner. Generally, the average year of construction of basin reservoirs (roughly equivalent to the average year of abrupt change in the runoff sequences of stations) is treated as the dividing point of the runoff sequence for all hydrological stations in China. Approximately 50–75% of the watershed area reservoirs in China were constructed during or before 1980, and the average construction year of large-scale reservoirs in China was 1972. From the perspective of economic development, China had just started reforming and opening up in 1980, and the construction of water conservancy projects began to develop rapidly to meet the needs of industrial and agricultural production and industrial and agricultural development. Annual runoff in many river basins significantly declined. In summary, this paper uses 1980 as the runoff sequence dividing point for the hydrological station data, which divides the runoff time sequence into period 1 (1956–1979) and period 2 (1980–2016). Period 1 is considered to be the period less affected by human activities, while period 2 is the period greatly affected by human activities.

3.3. RCCC-WBM Model Parameter Calibration and Natural Runoff Process Simulation

The measured runoff sequences at stations before the year of abrupt change in runoff (period 1) were used to calibrate the model parameters and validate the model.

Table 1. Simulation results based on monthly runoff data from representative hydrological stations.

Hydrological Station	Rate on a Regular Basis			Validation Period
	The Data Series	ENS/%	ER/%	The Data Series	ENS/%	ER/%
Ben Zilan	1956–1985	83.12	2.92	1986–1995	83.51	5.24
Cun Tan	1956–1985	68.93	1.15	1986–1996	84.61	5.19
Han Kou	1956–1965	71.48	1.72	1966–1970	62.28	3.24
Du Fengkeng	1956–1970	72.5	2.84	1971–1980	74.37	5.73
Hu Shan	1956–1970	76.26	0.82	1971–1980	77.56	2.17
Cheng Lingji	1956–1965	79.04	2.17	1966–1975	64.31	4.8
Hu Kou	1956–1975	63.33	2.51	1976–1989	73.7	4.78
Gui Gang	1956–1985	63.32	0.93	1986–1996	75.63	2.95
Gui Lin	1956–1990	75.98	3.49	1991–2003	65.25	3.99
He Yuan	1956–1962	73.02	3.64	1963–1968	80.48	5.66
Heng Shan	1956–1985	80.71	1.03	1986–1998	80.76	2.88
Chao An	1956–1965	64.13	3.9	1966–1970	79.47	5.99
Plump reservoir	1956–1970	66.51	0.37	1971–1985	77.55	2.7
Harbin	1956–1974	71.69	0.33	1975–1984	74.45	1.9
Jiamusi	1956–1970	71.41	1.55	1971–1985	83.58	2.65
Chao Yang	1956–1970	80.53	2.06	1971–1985	81.21	2.77
Zhimen Da	1956–1965	75.45	4.46	1966–1970	85.46	5.83
Shi Gu	1956–1970	76.37	2.06	1971–1985	85.69	5.77
Lengshui Jiang	1956–1970	68.24	1.17	1971–1980	60.44	4.22
Xiang Tan	1956–1970	72.88	1.91	1971–1980	61.36	2.73
Tong Guan	1956–1980	71.63	0.83	1981–1990	77.33	1.22
Li Jin	1956–1970	83.98	2.68	1971–1979	61.69	3.41
Wenjia Chuan	1956–1980	83.03	4.3	1981–1991	82.32	5.23
Run Chen	1956–1965	64.2	0.86	1966–1970	67.05	2.41
Mudan Jiang	1956–1970	69.48	0.22	1971–1987	75.97	3.05
Wang Ben	1956–1960	65.2	2.99	1961–1965	66.54	4.28
Tangnai Hai	1956–1980	72.84	3.34	1981–1990	63.19	4.17
Lan Zhou	1956–1970	80.46	0.91	1971–1985	80.06	4.54
Shizui Shan	1956–1970	83.31	1.32	1971–1985	79.88	4.95

shows the simulation results of monthly flow measured at representative hydrological stations in China and indicates that the efficiency coefficients of the Nash–Sutcliffe model of the runoff measured at all stations before abrupt change exceeded 0.5, and the RE values of the simulated monthly measured runoff were also small, within ±5%. Figure 5 shows the measured and simulated monthly runoff processes before the abrupt change in runoff at representative hydrological stations in China and indicates that the measured and simulated monthly runoff processes generally fit the model well. Moreover, in combination with the efficiency coefficients and RE values from the Nash–Sutcliffe model presented in Table 1, these results fully demonstrate that the RCCC-WBM model can better simulate the multiyear hydrological process of representative hydrological stations in China, with high simulation accuracy, and can meet the requirements and be used to simulate and restore natural runoff processes during periods not affected by human activities.

Model parameters calibrated using the measured runoff sequence at each representative hydrological station were used to simulate and analyze the runoff measured during the entire study period from 1956 to 2016. Hydrological stations can be considered to have no “natural runoff” process under the influence of human activities. Figure 6 shows the measured and simulated annual runoff at the representative hydrological stations from 1956 to 2016. The measured and simulated runoff at each representative hydrological station showed accurate simulation overall before the year of abrupt change in runoff, but after the abrupt change year, the simulated runoff was significantly larger or smaller than the measured runoff, indicating that the influence of human activities changed the measured river runoff significantly, but the degree of influence showed clear spatial variation.

3.4. Effects of Abrupt Temperature Change and Warming Hiatus on Surface Runoff

Based on further analysis of the calibration results of the model, although the overall change in temperature had little impact on runoff, the changes in temperature before and after the abrupt temperature change and warming hiatus were relatively large, especially before and after the abrupt temperature change. What was its quantitative effect on runoff change? However, where precipitation changed steadily for years, runoff increased in some natural watersheds that were less affected by human activities; in such a phenomenon, what are the effects of abrupt temperature changes and warming hiatus? To further explore the impacts of three types of abrupt temperature changes and warming stagnation on runoff nationally, the WBM model was used to remove the influence of human activities to simulate the natural runoff from representative stations in China. Taking the abrupt temperature change and warming hiatus as the time nodes, principal component analysis was used to calculate the quantitative impact of temperature on the measured surface runoff under abrupt temperature change, abrupt temperature change/warming hiatus, and after warming hiatus. Combined with the three types of temperature change in the three time periods, the absolute impacts of abrupt temperature change and warming hiatus on surface runoff in China were quantitatively calculated.

Figure 7 shows the percentages of surface runoff change affected by abrupt temperature change and warming hiatus calculated based on the WBM model and indicates that the maximum impact of abrupt temperature change on surface runoff reached 30% of the runoff change. However, only a small number of hydrological stations were affected by abrupt temperature change. Abrupt changes in mean maximum temperatures exerted both positive and negative driving effects on runoff in China. The negative effect was concentrated in the regions where abrupt temperature changes occurred, and the average impact accounted for approximately 10% of the change in runoff. The impact percentages in the northeastern basin and the basin at the foot of the Qilian Mountains were relatively high, which may be due to the large abrupt temperature change in the northeast, with a maximum temperature increase of 5 °C. Coupled with the lack of water resources and less precipitation in the northern area, temperatures have risen sharply, negatively affecting runoff. However, the mean maximum temperatures in some southern areas dropped suddenly, increasing runoff. The average positive driving effect accounted for approximately 8% of the change in runoff and basically showed a spatial distribution pattern with the size of the positive effect increasing from north to south. The impact percentage of the abrupt change in mean temperature on change in runoff was smaller than that of the abrupt change in mean maximum temperature, and the impact was greater in the north than in the south. In terms of the impact of abrupt change in mean temperature on runoff change in China, the average impact accounted for approximately 8% of the change, showing a negative driving effect. The areas largely affected by changes in runoff were concentrated in the area north of 40° N. Sudden changes in mean temperature exhibited greater impacts in arid and semiarid areas. Significant temperature increases lead to increases in evaporation. Since precipitation in arid and semiarid areas is less, the negative driving effect of temperature rise is prominent, which in turn has a greater impact on runoff. In the south, where rainfall is sufficient, the magnitude of the abrupt temperature change was relatively small, so the impact on runoff change was also small. The spatial distribution of the impact of the abrupt change in the mean minimum temperature on the change in measured runoff was similar to that of the abrupt change in mean temperature, but the impact of the abrupt change on the runoff change was smaller than the impact of the abrupt change in mean temperature on runoff. The areas with the greatest impacts on runoff from the abrupt change in mean minimum temperature were concentrated in the northeast region, which accounted for 9–12% of the runoff change, while the southern region was less affected by the abrupt change in the mean minimum temperature. The three measures of abrupt temperature change showed different impacts on the measured runoff changes over the years. Generally, the impact on the measured runoff was greater in the north than in the south, which may be due to greater temperature increase during abrupt temperature change in the northern region compared to that in the southern region. In addition, the northern region is more sensitive to temperature due to the lack of water resources. It is worth noting that the areas where the mean maximum temperature has abruptly decreased have shown a nonnegligible negative driving effect on the measured runoff for many years.

The effect of warming hiatus on runoff change was relatively small. Figure 7 shows that when a warming hiatus occurred, the impact of the average warming hiatus on the multiyear changes in the measured runoff exceeded that of mean maximum temperature, while the impact of the warming hiatus as reflected by mean minimum temperature on the measured runoff was relatively small for many years. Of areas where the mean maximum temperatures have undergone a warming hiatus, the impact on the multiyear runoff change was greatest in the middle reaches of the Yellow River, accounting for approximately 9% of the runoff change. The impact on the runoff change in the northeast was relatively small, accounting for less than 3% of the change in runoff. When the mean temperature underwent a warming hiatus, it exerted a relatively large impact on the multiyear runoff changes in the basins east of 110° E, with an average impact accounting for approximately 9.6% of the runoff change, especially in the lower reaches of the Yellow River and Yangtze River basins. The mean temperature warming hiatus exhibited a larger impact on the runoff change, accounting for 18%, while when the mean minimum temperature underwent a warming hiatus, its impact on the multiyear runoff change was the smallest, accounting for less than 5% of the multiyear runoff change.

4. Results and Discussion

Before the sudden change in temperature in China, the impact on surface runoff was below 9%, while the impact of temperature on runoff in southern regions was even weaker, only below 3%. When there is a sudden change in temperature, the temperature in arid areas has a maximum impact of 30% on surface runoff in a small number of areas. The sudden change in average maximum temperature has both positive and negative driving effects on runoff in China, of which the negative driving effect is concentrated in the areas where temperature change occurs, with an average impact of about 8% of runoff. The rivers in Northeast China and those near the Qilian Mountains have the largest impact on their runoff due to the sudden change in average maximum temperature. The sudden change in average temperature has an average impact on runoff of about 10%, and all show a negative driving effect. The sudden increase in temperature leads to a decrease in runoff. The areas with significant changes in runoff are concentrated north of 40° N. The sudden change in temperature in the middle and lower reaches of the Yellow River Basin has a significant impact on runoff changes, around 13%; the runoff change in the Heilongjiang and Songhua River basins in Northeast China is also greatly affected by the sudden change in average temperature. The region with the greatest impact of sudden changes in average minimum temperature on runoff changes is concentrated in the northeast region, around 9% to 12%, while the southern region is less affected by sudden changes in average minimum temperature. The three types of temperature mutations have had varying degrees of impact on the changes in surface runoff measured over the years in China. Overall, the spatial pattern shows a greater impact on the measured runoff in the north than in the south. It is worth noting that the regions where the average highest temperature undergoes a cooling mutation have a significant negative driving effect on the measured runoff over the years. Compared with the impact of temperature changes on runoff throughout the entire period, when there is a sudden change in temperature, the impact on runoff significantly increases, especially in arid and semiarid regions. Compared with the entire period, the contribution of sudden temperature changes to runoff changes increases by about twice. With the intensification of global climate change and the rapid development of regional social economies, the impact of environmental change on regional water resources is increasingly apparent [37]. Dynamically simulating regional hydrological processes under changing environments and scientifically evaluating the impacts of various driving factors on water resources constitute the basic work for regional water resources evaluation. With the in-depth understanding of regional hydrological cycles and the rapid development of computer technology, basin hydrological simulation technology has become an important and effective tool for evaluating regional water resources [38]. At present, hundreds of basin hydrological models have been proposed worldwide and broadly applied in flood forecasting, water resource simulation, and impact assessment of environmental change [39]. However, due to differences in hydroclimatic characteristics among basins, basin hydrological models based on different runoff mechanisms also have corresponding regional applicability. Therefore, a basin hydrological model can be applied to a specific area only after a sufficient number of representative basins are used to fully test and validate the applicability of the model. The WBM model used in this study has been applied for attribution analysis in China. The simulation results are favorable, with the characteristics of a smaller amount of data to use, relatively easy access, and few parameters. It can be effectively used for attribution analysis in China to further study the impacts of abrupt temperature change and warming hiatus on regional runoff. However, there are some shortcomings, too. For instance, few factors are used. Although the simulation results are accurate, in the later attribution analysis, temperature change will inevitably be attributed to the influence of certain factors, which will lead to improper evaluation of the quantitative impact of temperature. Especially for some stations with stable changes in measured runoff over many years, it fails to accurately distinguish the reference period from the period of impact from human activities so that the timing of the transition can only be speculated by searching the local economic development data for many years. Therefore, the accuracy of the results is affected. The model is more suitable for runoff data with an obvious timepoint of abrupt change. When calculating the impacts of abrupt temperature change and warming hiatus on runoff, there are few influencing factors calculated, which may lead to large calculated values, but their quantitative impacts on runoff cannot be ignored in some climatic types. Multiple methods can be combined to comprehensively quantify the impacts of abrupt temperature change and warming hiatus on surface runoff in subsequent studies.