Risk Assessment of Waterlogging in Major Winter Wheat-Producing Areas in China in the Last 20 Years

Abstract

Against the background of global warming, agricultural meteorological disasters such as waterlogging frequently occur, significantly restricting winter wheat yield and quality formation. Studying the changing trend of meteorological characteristics of waterlogging is beneficial to stabilizing winter wheat yield. We collected meteorological and yield data of China’s main winter wheat production areas in the last 20 years to explore the impact of waterlogging in different growth stages on wheat production. The results showed that waterlogging greatly impacted winter wheat production in the main winter wheat production areas in China, and the degree of influence was larger in the south than in the north. The precipitation in the south was higher, and waterlogging occurred in most growth stages, but waterlogging at the filling stage was more consistent with the yield reduction. On the other hand, the interannual variation in precipitation in the seedling stage in the north varied greatly, which was the critical stage of waterlogging. In conclusion, waterlogging was one of the main factors affecting winter wheat production in China. For southern cities, the filling period was the key period for disaster prevention and mitigation, but it was the seedling stage in the north.

1. Introduction

Wheat supplies one-fifth of dietary calories and protein to the world’s population, with an annual consumption of more than 710 million tons [1]. China is the largest wheat-producing country, with an annual output of 131.4 million tons [2]. Therefore, ensuring wheat-production safety in China is vital for food supply and social stability. Land degradation has been predicted to lead to a decrease of 5% per unit area and 14% in the total wheat yield globally by 2050 [3]. Waterlogging is one of the key soil barrier factors affecting wheat production globally [4,5]. About 5–20% of crops suffer from waterlogging stress, leading to yield loss because of extreme agrometeorological disasters, high groundwater levels, poor drainage facilities, etc. [6]. According to statistics, the global wheat yield loss caused by waterlogging is as high as 25% [7]. On the global scale, the daily extreme precipitation will increase by 5.2% for every 1 °C increase in the average temperature [8]. Hence, with global warming, waterlogging is increasing. Therefore, the meteorological characteristics of waterlogging should be investigated to inform farmers when preventive and remedial measures need to be taken to prevent wheat yield losses.

There are large spatial differences in climate characteristics and agricultural endowments in different wheat-production regions of China [9]. There are five major wheat-producing provinces in China, including Hebei Province, Shandong Province, Henan Province, Jiangsu Province, and Anhui Province, with a wheat-sown area of more than 2000 hm² (http://www.stats.gov.cn/tjsj/ndsj/ accessed on 1 July 2022), located in different regions of North China and Southeast China. Hebei Province (36°03′–42°40′ N, 113°27′–119°5′ E) is located in the North China Plain of China, where heavy rainfall and floods frequently occur in summer, and the frequency of disasters shows an increasing trend [10]. Shandong Province (34°22′–38°24′ N, 114°47′–122°42′ E) is located in the northern coastal area of China, where rainstorms and floods account for 14.12% of meteorological disasters [11]. Henan Province (31°23′–36°22′ N, 110°21′–116°39′ E) is located in the central plains of China, where rainstorms and floods account for 23% of the meteorological disasters [12]. Jiangsu Province (30°46′–34°38′ N, 114°54′–119°37′ E) is located on the southeastern coast of China, where the climate is warm and humid. Affected by the sea–land distribution, atmospheric circulation, and monsoon precipitation, it is a typical drought- and flood-disaster-prone area in China, where rainstorms and floods account for 70% of meteorological disasters [13]. Anhui Province (29°41′–34°38′ N, 114°54′–119°37′ E) is located in the middle latitude zone and belongs to the transition zone of north–south climate in China. The monsoon season is significant, and the weather is changeable. The wheat-growing period is often accompanied by heavy rain and other disasters, adversely affecting wheat production [14]. Frequent rainfall is a typical meteorological feature in South China, while extreme rainfall events have occasionally occurred in northern China in recent years due to global climate change, making waterlogging stress a key limiting factor for wheat production [15]. Therefore, there are great differences in climate among the major wheat-producing provinces in China, and the characteristics of waterlogging and its impact on wheat yield in different wheat-producing areas in China warrant further study.

The degree of the inhibition of waterlogging on wheat yield and quality is different in different growth stages. Waterlogging at the germination stage leads to a reduction in the emergence rate and affects the formation of spike numbers [16]. The temperature in the overwintering stage is low, and wheat grows slowly; thus, waterlogging has a limited impact on the yield [17]. The jointing–booting stage is the key growth stage of wheat, which is closely related to the formation of ear number and grain number per ear. However, the effect of waterlogging at this stage on yield formation is controversial [18,19]. Waterlogging at the booting stage significantly reduces the fertile floret number per panicle, grain number per panicle, and 1000-grain weight [18,20]. The filling period is a period when waterlogging has a great impact on yield [19]. Moreover, the impact of waterlogging on wheat growth is also affected by environmental temperatures, waterlogging duration, soil texture, and other factors [21,22]. Therefore, the decline in winter wheat yield by waterlogging is affected by various meteorological factors at different growth stages. Due to the different climatic conditions in different wheat-production areas in China, the influence of waterlogging at different growth stages should be varied in different wheat-production areas.

Waterlogging is an important factor affecting wheat production, and it is of great significance to study the occurrence of waterlogging. We conducted statistical analysis on the meteorological data and yield data during different wheat growth stages, studied the waterlogging index, the annual occurrence frequency of waterlogging, and the annual average yield-reduction rate of waterlogging to determine (i) the geographical waterlogging risk distribution of the main Chinese winter wheat production areas and (ii) the key growth stage when waterlogging affecting wheat yield in the main wheat-producing areas in China.

2. Materials and Methods

2.1. Data Source

The winter wheat yield, sowing area, ten-day sunshine time, and ten-day precipitation during the wheat-growth period (October to June of the next year) from 2001 to 2020 were collected from the Chinese statistical yearbook (http://www.stats.gov.cn/tjsj/ndsj/, accessed on 1 July 2022). The whole growth period was divided into four stages: seedling stage (October to December), overwintering stage (December to February of the next year), rejuvenation booting stage (February to April), and grain-filling stage (April to June).

2.2. Determination of Study Area

Hebei Province, Shandong Province, Henan Province, Jiangsu Province, and Anhui Province were selected to carry out the study and analysis, where the wheat-sowing area is more than 2000 thousand hectares. The above five provinces are the main wheat-producing areas in China, and the sample representation is significant. The study area includes Shijiazhuang, Jinan, Zhengzhou, Nanjing, and Hefei, which are capitals of Shandong Province, Henan Province, Jiangsu Province, and Anhui Province, respectively (Figure 1).

2.3. Data Processing

2.3.1. Construction of Relative Climatic Yield

To estimate the part of yield reduction in the production fluctuation caused by meteorological conditions, climate yield (y_w) was obtained from the following Equation [23,24,25]:

$$y=y_t+y_w+\epsilon$$

(1)

where y is the actual yield, y_t is trend yield that varies with the social production level, y_w is climate yield that varies with the meteorological conditions over the years, and ε is random error, which is generally ignored. A grey system GM(1,1) model can predict the development and change in a system containing both known and uncertain factors. Therefore, y was supplied to the GM(1,1) model to formula y_t as follows:

$$X^{\left(0\right)}=\left\{x{ }^0\left(i\right),i=1,2,\cdots ,n\right\}=\left\{y\left(i\right),i=1,2,\cdots ,n\right\}$$

(2)

where n represents the total number of samples (n = 20). The sequence X⁽⁰⁾ was accumulated to obtain sequence X⁽¹⁾:

$$x^{\left(1\right)}\left(k\right)={{\displaystyle \sum }}_{i=1}^k{x}_i^{\left(0\right)}, k=1,2,\cdots ,n$$

(3)

$$X^{\left(1\right)}=\{x^{\left(1\right)}\left(k\right),k=1,2,\cdots ,n\}$$

(4)

and then the sequence X⁽¹⁾ was used to set up a differential equation:

$$\frac{d_x^{\left(1\right)}}{d_t}+a{x}^{\left(1\right)}=u$$

(5)

where a and u are the main parameters of the GM(1,1) model. To calculate a and u, the sequence X⁽¹⁾ was sliding averaged as:

$$B=\left[\begin{array}{cc}-\frac{1}{2}\left(x^{\left(1\right)}\left(1\right)+x^{\left(1\right)}\left(2\right)\right)& 1\\ -\frac{1}{2}\left(x^{\left(1\right)}\left(2\right)+x^{\left(1\right)}\left(3\right)\right)& 1\\ ⋮& ⋮\\ -\frac{1}{2}\left(x^{\left(1\right)}\left(n-1\right)+x^{\left(1\right)}\left(n\right)\right)& 1\end{array}\right]$$

(6)

$$Y=\left[\begin{array}{c}{x}^{\left(0\right)}\left(2\right)\\ x^{\left(0\right)}\left(2\right)\\ ⋮\\ x^{\left(0\right)}\left(n\right)\end{array}\right]$$

(7)

and we can set $\widehat{a}$ as an estimating parameter:

$$\widehat{a}={\left(B^{T}B\right)}^{-1}{B}^{T}Y$$

(8)

$$\widehat{a}={\left[a,u\right]}^T$$

(9)

Thus, a and u can be calculated. A trend value (${\widehat{x}}^{\left(1\right)}$) in the prediction model can be obtained as:

$${\widehat{x}}^{\left(1\right)}\left(k\right)=\left(x^{\left(0\right)}\left(1\right)-\frac{u}{a}\right)e^{-a\left(k-1\right)}+\frac{u}{a}, k=1,2,\cdots ,n$$

(10)

Then, restore value ${\widehat{x}}^{\left(0\right)}$ was calculated as:

$${\widehat{x}}^{\left(0\right)}\left(k\right)={\widehat{x}}^{\left(1\right)}\left(k\right)-{\widehat{x}}^{\left(1\right)}\left(k-1\right)$$

(11)

${\widehat{x}}^{\left(0\right)}$ is the y_t in each year. Then, y_w is calculated by Formula (1). To eliminate the “white noise” interference of the production level change on y_w, we adopted relative climatic yield (y_r) for the following analysis, which was calculated from Equation (2):

$$y_r=\frac{y_w}{y_t}\times 100\%$$

(12)

2.3.2. Calculation of Waterlogging Index

Considering the influence of cloudy and rainy weather on crop growth, the crop water requirement, precipitation, and sunshine time in a certain growth period were used to form the waterlogging index in each month (Q_i) [26,27]:

$$Q_i=\frac{P_{re}-E{T}_c}{E{T}_c}-\frac{S-S_0}{S_0}$$

(13)

where P_re is the precipitation in the calculated month, ET_c is the water requirement, S is the sunshine time in the calculated month, and S₀ is the average sunshine time in this month over the year.

The effects of waterlogging in different growth stages on wheat yield are different. The Q_i value of each month is added with different weights as the wet damage index in different growth stages (Q_s):

$$Q_s={\displaystyle \sum }_{i=1}^3{\alpha }_i{Q}_i$$

(14)

where α_i is the weight coefficient of the corresponding month, obtained by the grey correlation degree method [28].

Then, correlation analysis was conducted to compare y_r and Q_s, ten-day precipitation, and ten-day sunshine time at different growth stages.

2.3.3. Calculation of Average Yield Reduction Rate and Frequency of Waterlogging

y_r < 0 was defined as the yield-reduction year caused by adverse climate conditions. The Q_s threshold value was determined when Q_s > 0 and the coincidence rate of y_r < 0 exceeds 60%. Then, the year with Q_s greater than the Q_s threshold value and y_r < 0 is defined as the waterlogging disaster year. Frequency of waterlogging (F_w) was calculated as [13,29]:

$$F_w=\frac{l}{t}\times 100\%$$

(15)

where l is the number of waterlogging disaster years, and t is the number of total samples. The percentage of the waterlogging disaster years to the yield-reduction years (F_wn) was calculated as

$$F_{wn}=\frac{l}{n}\times 100\%$$

(16)

where n is the number of the yield-reduction year. Average yield reduction rate of waterlogging (d_w) was calculated as

$$d_w=\frac{{{\displaystyle \sum }}_{i=1}^l{y}_{ri}}{l},y_{ri}<0$$

(17)

2.3.4. Calculation of Waterlogging Risk Index

The F_w of the whole growing period and d_w were used to construct a waterlogging risk index (K) of different areas [29]:

$$K=d_w\times F_w$$

(18)

2.4. Statistical Analyses

The Pearson correlations conducted to compare y_r and ten-day sunshine time, ten-day precipitation, and Q_s, as shown in Figure 2d,e,f, Figure 3d,e,f, Figure 4d,e,f, Figure 5d,e,f and Figure 6d,e,f, were derived using Origin 2022b software (OriginLab Corporation, Northampton, MA, USA). The extracted event years were determined by the behavior of y_r and ten-day sunshine time, ten-day precipitation, and Q_s excess the mean value ± 1 standard deviation (hereafter SD) [30], and the results are shown in

Table 1. Extracted event years in Shijiazhuang based on the behavior of y_r and ten-day sunshine time, ten-day precipitation, and Q_s excess the mean value ± 1 SD. The red check represents increase, and the black check represents decrease.

Year	Relative Climatic Yield (yr)	Seeding Stage			Wintering Stage			Rejuvenating-Booting Stage			Filling Stage
		Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)
2000–2001		√		√	√	√	√					√
2001–2002						√					√	√	√
2002–2003					√	√	√	√	√		√	√
2003–2004	√		√	√	√		√	√				√
2004–2005						√	√
2005–2006	√		√	√						√
2006–2007									√	√
2007–2008									√			√	√
2008–2009		√									√		√
2009–2010	√		√							√	√	√	√
2010–2011		√		√	√					√
2011–2012						√			√			√	√
2012–2013					√		√	√			√		√
2013–2014	√				√			√		√	√
2014–2015	√				√		√
2015–2016		√
2016–2017		√	√	√
2017–2018	√		√	√	√	√	√
2018–2019									√
2019–2020								√		√		√	√

,

Table 2. Extracted event years in Jinan based on the behavior of y_r and ten-day sunshine time, ten-day precipitation, and Q_s excess the mean value ± 1 SD. The red check represents increase, and the black check represents decrease.

Year	Relative Climatic Yield (yr)	Seeding Stage			Wintering Stage			Rejuvenating-Booting Stage			Filling Stage
		Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)
2000–2001	√	√	√	√	√	√	√
2001–2002	√					√				√	√
2002–2003		√			√		√	√	√	√	√
2003–2004			√	√	√							√	√
2004–2005											√
2005–2006	√	√							√			√	√
2006–2007			√							√
2007–2008	√	√			√		√				√
2008–2009									√	√			√
2009–2010					√			√			√
2010–2011		√	√	√						√		√	√
2011–2012													√
2012–2013						√
2013–2014		√
2014–2015					√				√
2015–2016								√		√
2016–2017		√									√
2017–2018	√				√	√	√		√			√	√
2018–2019												√	√
2019–2020		√			√	√	√	√			√		√

,

Table 3. Extracted event years in Zhengzhou based on the behavior of y_r and ten-day sunshine time, ten-day precipitation, and Q_s excess the mean value ± 1 SD. The red check represents increase, and the black check represents decrease.

Year	Relative Climatic Yield (yr)	Seeding Stage			Wintering Stage			Rejuvenating-Booting Stage			Filling Stage
		Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)
2000–2001						√	√		√	√	√	√	√
2001–2002	√				√	√						√	√
2002–2003	√	√			√	√	√	√		√	√		√
2003–2004		√	√	√					√
2004–2005					√				√
2005–2006	√							√			√
2006–2007	√								√	√		√	√
2007–2008	√							√	√
2008–2009								√	√		√
2009–2010						√	√	√
2010–2011		√	√	√						√		√	√
2011–2012		√	√	√		√						√	√
2012–2013					√
2013–2014		√						√
2014–2015		√			√	√	√		√		√	√	√
2015–2016		√	√	√								√
2016–2017			√	√		√					√
2017–2018	√				√		√	√
2018–2019					√		√
2019–2020						√	√

,

Table 4. Extracted event years in Nanjing based on the behavior of y_r and ten-day sunshine time, ten-day precipitation, and Q_s excess the mean value ± 1 SD. The red check represents increase, and the black check represents decrease.

Year	Relative Climatic Yield (yr)	Seeding Stage			Wintering Stage			Rejuvenating-Booting Stage			Filling Stage
		Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)
2000–2001		√
2001–2002	√										√
2002–2003	√	√			√	√	√	√	√	√	√
2003–2004
2004–2005								√			√
2005–2006	√
2006–2007	√					√
2007–2008	√							√		√	√
2008–2009											√
2009–2010						√	√	√	√	√
2010–2011		√		√	√	√	√	√	√	√
2011–2012												√	√
2012–2013
2013–2014		√	√	√						√
2014–2015		√			√		√					√	√
2015–2016											√
2016–2017		√	√	√	√			√			√
2017–2018								√
2018–2019					√	√	√	√
2019–2020

and

Table 5. Extracted event years in Hefei based on the behavior of y_r and ten-day sunshine time, ten-day precipitation, and Q_s excess the mean value ± 1 SD. The red check represents increase, and the black check represents decrease.

Year	Relative Climatic Yield (yr)	Seeding Stage			Wintering Stage			Rejuvenating-Booting Stage			Filling Stage
		Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)	Ten-Day Sunshine Time	Ten-Day Precipitation	Waterlogging Index (Qs)
2000–2001		√			√						√	√	√
2001–2002	√					√		√			√		√
2002–2003	√				√	√	√	√	√	√	√
2003–2004			√		√			√		√
2004–2005	√										√	√	√
2005–2006
2006–2007												√
2007–2008	√							√	√	√
2008–2009											√
2009–2010									√	√
2010–2011		√	√	√	√	√	√	√	√	√
2011–2012						√	√
2012–2013									√
2013–2014		√	√	√				√	√	√
2014–2015		√			√						√	√	√
2015–2016		√				√	√				√	√	√
2016–2017		√	√	√
2017–2018
2018–2019					√	√	√
2019–2020		√						√

.

3. Results

3.1. Facrors Causing Wataerlogging Disater in Different Growth Stages in Shijiazhuang

As shown in Figure 2, the y_r in Shijiazhuang varied greatly from year to year. y_r increased in 2003–2004, 2013–2015, and decreased in 2005–2006, 2009–2010, and 2017–2018 (Table 1). A negative correlation (not significant) was found between y_r and ten-day sunshine time in each growth stage (Figure 2d).

As shown in Figure 2b, precipitation in the seedling stage was relatively less but raised dramatically in some years (2009–2010 and 2016–2018) (Table 1), and y_r in the corresponding years decreased significantly. Precipitation in the wintering stage and rejuvenating–booting stage was less, and variation over the years was less obvious than other growth stages. Precipitation during the filling stage was greater, and the variation over the years was large. Precipitation during the filling stage increased in 2001–2002, 2003–2004 (y_r increased), 2007–2008, 2011–2012, and 2019–2020, but decreased in 2000–2001, 2002–2003, and 2009–2010 (y_r decreased) (Table 1).

Q_s in the seedling stage and wintering stage was higher than that in other growth stages (Figure 2c). In 2000–2001, 2016–2017, and 2017–2018, Q_s in the seedling stage increased, and y_r in 2017–2018 decreased correspondingly. Q_s in the overwintering stage increased significantly in 2000–2001, 2002–2003, 2004–2005, and 2012–2013 (Table 1), but y_r did not decrease correspondingly.

The above results indicate that the annual difference in y_r in Shijiazhuang is large. The seedling stage had a high incidence of waterlogging in this area, and the increase in Q_s in the seedling stage is consistent with the decrease in y_r.

3.2. Facrors Causing Wataerlogging Disater in Different Growth Stages in Jinan

As shown in Figure 3, the y_r in Jinan was relatively small. y_r increased in 2005–2006 and 2007–2008 and decreased in 2000–2002 and 2017–2018 (Table 2). A significant positive correlation was found between y_r and ten-day sunshine time in the wintering stage (p < 0.05), rejuvenating–booting stage (p < 0.01), and filling stage (p < 0.05) (Figure 3d).

Figure 3b shows that precipitation in the seedling stage was less but increased during 2000–2001 (y_r decrease year) and 2003–2004 (Table 2). There was less precipitation in the wintering stage and little variation among different years. The rainfall during the rejuvenating–booting stage was less, and there was little variation among different years. There was significant precipitation during the filling stage, with a large variation among different years. The precipitation in the filling stage increased sharply during 2003–2004, 2005–20016 (y_r increased), and 2017–2018 (y_r decreased) (Table 2).

As shown in Figure 3c, Q_s at each growth stage had a large annual variation. Only the increase in Q_s of the seedling stage and wintering stage in 2000–2001 and Q_s in the filling in 2017–2018 was synchronous with the reduction in y_r.

In summary, the annual variation of y_r in Jinan was relatively small. The precipitation during the filling period was large, with a large annual variation. The Q_s in different growth periods had a large annual variation, but the increase in Q_s was not synchronized with the decrease in y_r.

3.3. Facrors Causing Wataerlogging Disater in Different Growth Stages in Zhengzhou

The y_r of Zhengzhou was relatively low from 2000 to 2005, and there was little variation in other years. y_r increased in 2005–2008 and decreased in 2001–2003 and 2017–2018 (Table 3). A positive correlation (not significant) was found between y_r and ten-day sunshine time in each growth stage (Figure 4d).

As shown in Figure 4b, precipitation in the seedling stage was less, but the variation among years was larger, which increased in 2003–2004, 2011–2012, and 2015–2017 (Table 3), and y_r in the corresponding years did not decrease. The precipitation during the wintering stage and the rejuvenating–booting stage was less, and there was little variation among different years. There was significant precipitation during the filling period. The precipitation increased sharply during 2001–2002 (y_r decrease year) and 2014–2016.

Q_s in the seedling stage was higher than that in other growth stages and varied greatly over the years (Figure 4c). During 2003–2004, 2011–2012, and 2015–2017, Q_s in the seedling stage increased (Table 3), but y_r did not decrease synchronously. Q_s in the filling stage increased during 2001–2003, 2006–2007, and 2014–2015, while y_r in 2001–2003 decreased correspondingly.

To summarize, the annual variation of y_r in Zhengzhou is small. Precipitation in the seedling and filling stages varied greatly from year to year, and precipitation in the filling stage was significant. Q_s in the seedling stage varied greatly over the years, but the increase in Q_s in the filling stage corresponded with a decrease in y_r.

3.4. Facrors Causing Wataerlogging Disater in Different Growth Stages in Nanjing

The y_r in Nanjing had a large variation (Figure 5a). y_r increased in the years 2005–2008, and decreased in the years 2001–2003 (Table 4). A positive correlation (not significant) was found between y_r and ten- day sunshine time in each growth stage (Figure 5d).

There was little difference in precipitation among different growth stages (Figure 5b). Precipitation in the seedling stage increased sharply in 2016–2017 (Table 4), but the y_r did not decrease. There was less precipitation during the wintering stage, and there was little variation among different years. During 2002–2003 (y_r decrease year) and 2009–2010, the precipitation during the rejuvenating–booting stage increased. During 2014–2015, the rainfall in the filling stage increased sharply and the y_r decreased slightly.

As shown in Figure 5c, there was little difference in Q_s among different growth stages. The Q_s increased in the wintering stage and the rejuvenating–booting stage corresponded with y_r reduction during 2002–2003 (Table 4).

In conclusion, the y_r in Nanjing varied greatly among different years. The precipitation in each growth stage in this region was relatively high, and the increase in precipitation and Q_s occurred in all growth stages, and the occurrence of Q_s increase coincided with the decrease in y_r.

3.5. Facrors Causing Wataerlogging Disater in Different Growth Stages in Hefei

There was little variation in y_r over the years, except for 2001 to 2003, with a significantly lower y_r (Figure 6a), but y_r increased in 2004–2005 and 2007–2008 (Table 5). A significantly positive correlation was found between y_r and ten-day sunshine time in the wintering stage and rejuvenating–booting stage (p < 0.05) (Figure 5d).

As shown in Figure 6b, precipitation in the seedling stage was less, except for the precipitation increasing in 2003–2004 and 2016–2017 (Table 5), but y_r did not decrease correspondingly. Precipitation during the wintering stage was less, and the variation among years was small. The variation in precipitation in the rejuvenating–booting stage was large, which increased in 2002–2003 (y_r decreased), 2009–2010, and 2013–2014. Precipitation in the filling stage was higher than in other growth stages, and the variation in precipitation in the filling stage was large among different years. Precipitation in the filling stage increased during 2014–2016, and y_r did not decrease in the corresponding years.

There was little difference in Q_s at different growth stages (Figure 6c). Q_s increased during the filling stage of 2001–2002, 2002–2003, 2014–2015, 2015–2016, and 2017–2018, and y_r in the corresponding years decreased. Q_s increased in the filling stage in 2001–2002, along with the wintering and rejuvenating-booting stages.

In summary, the variation in y_r in Hefei was relatively small over the years. Precipitation in the filling stage was relatively high, and the variation in different years was large.

3.6. Frequency of Waterlogging and Average Yield-Reduction Rate in Different Growth Stages in Different Cities

F_w in Zhengzhou, Nanjing, and Hefei across all the growth stages was higher: 40%, 50%, and 35%, respectively. Waterlogging occurred in all growth stages in Nanjing and occurred in all growth stages except for the rejuvenating-booting stage in Zhengzhou and Hefei. The d_w in each growth stage in Hefei was higher than that in other cities, while d_w in Zhengzhou was higher during the filling stage. F_w in Shijiazhuang and Jinan was low, where waterlogging mainly occurred in the seedling stage (

Table 6. Frequency of waterlogging (F_w), the percentage of waterlogging disaster years to yield-reduction years (F_wn), and average yield-reduction rate due to waterlogging (d_w) of winter wheat in all the growing periods and different growth stages.

Region	Seeding Stage			Wintering Stage			Rejuvenating-Booting Stage			Filling Stage			The Whole Growing Period
	Fw/%	Fwn/%	dw/%	Fw/%	Fwn/%	dw/%	Fw/%	Fwn/%	dw/%	Fw/%	Fwn/%	dw/%	Fw/%	Fwn/%	dw/%
Shijiazhuang	10.0	16.7	1.9	—	—	—	—	—	—	—	—	—	10.0	16.7	1.9
Jinan	15.0	37.5	3.7	—	—	—	—	—	—	—	—	—	20.0	50.0	3.4
Zhengzhou	35.0	77.8	3.6	20.0	44.4	3.2	—	—	—	10.0	22.2	6.8	40.0	88.9	3.4
Nanjing	20.0	40.0	4.2	10.0	20.0	5.6	40.0	80.0	3.4	30.0	60.0	4.4	50.0	100.0	3.2
Hefei	25.0	62.5	9.0	25.0	62.5	10.3	—	—	—	30.0	75.0	8.8	35.0	87.5	8.3

).

3.7. Geographical Distribution of Winter Wheat Waterlogging

The average yield-reduction rate of the yield-reduction year caused by adverse climate conditions in Shijiazhuang was lower than 3%; in Jinan, Zhengzhou, and Nanjing: it was between 3% and 6%, and in Hefei it was relatively high at more than 8% (Figure 7). Furthermore, the average yield-reduction rate of the waterlogging disaster year (d_w) was slightly higher than that in the yield-reduction year due to adverse climate conditions. F_wn was lower than 30% in Shijiazhuang; in Jinan, it was between 30 and 60%; in Zhengzhou and Hefei, it was between 60 and 90%; and in Nanjing, it was as high as 100% (Figure 8). The results indicated that waterlogging was one of the main reasons for the yield decline in the main winter wheat production areas in China, and the frequency of waterlogging disasters increased gradually from north to south.

Considering the waterlogging risk index (K), the study area was divided into two wet-damage risk areas: a Grade I area is a high-risk area (>1%), including Zhengzhou, Nanjing, and Hefei, and a Grade II area is a low-risk area (<1%), including Shijiazhuang and Jinan (

Table 7. Regional division regarding the waterlogging risk index (K).

Division of Region	Region	Value of Risk Index
Grade I	Hefei, Nanjing, Zhengzhou	>1%
Grade II	Shijiazhuang, Jinan	<1%

).

4. Discussion

4.1. Effect of Waterlogging on Winter Wheat Production in the Main Wheat-Producing Areas in China

The risk assessment of winter wheat waterlogging, based on the yield-reduction rate and the frequency of waterlogging, objectively reflects the magnitude of the disaster of winter wheat waterlogging in the region [27]. In the present study area, the rank of yield reduction caused by adverse climatic factors was Shijiazhuang < Zhengzhou < Nanjing < Jinan < Hefei; the rank of d_w was Shijiazhuang < Nanjing < Jinan < Zhengzhou < Hefei; and the rank of F_wn was Shijiazhuang < Jinan < Hefei < Zhengzhou < Nanjing (Table 2). F_w in Nanjing was the highest, but d_w was lower than that in Hefei and Zhengzhou, which may be due to the high attention paid to waterlogging in Nanjing and the effective prevention and control measures being taken. Then, F_w and the d_w were used to calculate the risk index of waterlogging (K). Based on the K index, Zhengzhou, Nanjing, and Hefei were categorized as high-risk areas of wet waterlogging, and Shijiazhuang and Jinan were categorized as low-risk areas. Therefore, waterlogging is one of the main factors affecting winter wheat production in the main wheat-production areas in China, with the influence being increasingly serious from north to south.

4.2. Precipitation Characteristics at Different Growth Stages in the Main Wheat-Producing Areas of China

The risk of worldwide waterlogging damage in crop production is increasing, as the daily extreme precipitation will increase by 5.2% for every 1 °C increase in the average temperature [8]. The precipitation intensity in China has an increasing trend, the precipitation frequency has a decreasing trend, and the extreme rainfall events have an increasing trend, showing an unequal spatial and temporal distribution of precipitation [31,32] and resulting in frequent drought and waterlogging disasters in different regions of China. In the present study, precipitation and the interannual precipitation variation during the wheat-growth period in different cities differed (

Table 8. Precipitation (mm) and interannual precipitation variation over the whole growing period and different growth stages in different cities.

City	Seeding Stage	Wintering Stage	Rejuvenating-Booting Stage	Filling Stage	The Whole Growing Period
Shijiazhuang	0–17	0.1–4.4	0–8.1	0–21.5	9.6–36
Jinan	0.3–18.2	0.9–9.5	2.2–14.4	8.4–42.3	16.7–69.7
Zhengzhou	0.5–20.3	0.8–10	3.3–10.7	6.1–27.9	14.6–51.0
Nanjing	5.6–52.1	5.8–28.8	8–47.9	15.2–97.9	57.9–154.4
Hefei	3.4–48.6	5.8–26.7	8–40.6	17.6–66.6	54.2–129.7

). The precipitation in each growth stage and the whole growing period of Nanjing and Hefei was higher than that of Shijiazhuang, Jinan, and Zhengzhou. In Nanjing, Hefei, and Jinan, precipitation in the filling period was the highest, followed by the seedling stage and the rejuvenating–booting stage, but precipitation was the lowest in the wintering period. The interannual variation in precipitation in the seedling stage of Nanjing, Hefei, and Jinan was the greatest, which was 9 times, 14 times and 18 times, respectively, while the difference in other growth periods was about 4–9 times. Precipitation in the seedling and filling stages of Shijiazhuang and Zhengzhou was the highest. Moreover, the interannual variation in precipitation in the seedling stage of these two cities was the largest, which was 17 and 20% in Shijiazhuang and Zhengzhou, respectively, and the interannual variation in precipitation in Shijiazhuang during the filling stage was 21%, while that in other growth stages was about 4–10%. Moreover, it has been estimated that precipitation in northern China will increase more than in southern China, the precipitation in autumn and winter will increase more, and extreme precipitation events will increase [33]. Autumn and winter coincide with the wheat seedling stage; in this case, the risk of extreme rainfall during the wheat seedling stage in northern cities will increase. In one word, the maximum precipitation in Nanjing and Hefei was four, two, and three times higher than that in Shijiazhuang, Jinan, and Zhengzhou, respectively, but the interannual variation in precipitation was less, with the interannual variation in precipitation being the largest in the seedling stage of Shijiazhuang, Jinan, and Zhengzhou. Therefore, Nanjing and Hefei were at a high risk of waterlogging, while the risk of waterlogging caused by extreme rainfall in the seedling stage in Shijiazhuang, Jinan, and Zhengzhou was relatively high.

4.3. Occurrence of Waterlogging at Different Growth Stages in the Main Winter-Wheat-Producing Areas of China

The damage due to waterlogging is related to the growth stage when waterlogging occurs. Analyzing the losses caused by waterlogging in different growth stages is conducive to implementing waterlogging prevention measures. After analyzing the frequency and yield-reduction rate of waterlogging in different growth stages, as well as the degree of coincidence between Q_s and y_r, it was found that the occurrence of waterlogging in the main winter wheat production areas in China was significantly different across time and space. F_w in Zhengzhou, Nanjing, and Hefei was higher than in other cities (Table 6), and an increase in Q_s always coincided with a decrease in y_r (Figure 5 and Figure 6). Among them, the precipitation in Nanjing and Hefei was significantly higher than that in other cities, and the waterlogging damage occurred in most of the growth stages, while waterlogging mainly occurred in the seedling stage due to the sharp increase in precipitation. However, d_w of Nanjing was significantly lower than that of the other two cities, which may be related to the application of waterlogging-resistant techniques and varieties. It was concluded that the waterlogging tolerance of wheat varieties in the middle and lower reaches of the Yangtze River decreased with the improvement in varieties from 1967 to 2010 [34], indicating that the current concern about waterlogging is low. Especially in Hefei, the increase in Q_s in the filling stage was consistent with a decrease in y_r (Figure 6). The period before and after flowering is the most sensitive period for wheat to waterlogging [35,36], and waterlogging occurring in the reproductive growth stage of wheat results in a greater decline in yield, greater than that in the vegetative growth stage [7]. Moreover, the precipitation in Nanjing and Hefei during the filling stage was higher than that in other growth stages (Table 8), which may account for the higher yield-reduction rate caused by waterlogging during the filling stage. The occurrence of waterlogging and yield reduction mainly occurred in the seedling stage in Shijiazhuang and (Table 6) Jinan, and the increase in Q_s coincided with a decrease in y_r (Figure 2 and Figure 3).

5. Conclusions

Waterlogging threatens worldwide wheat production. In the present study, waterlogging was one of the important factors affecting the winter wheat yield in the main winter wheat production areas in China, and the impact of waterlogging was more serious in the south than in the north. Precipitation in the southern cities (Nanjing and Hefei) was higher than in the northern cities (Shijiazhuang, Jinan, and Zhengzhou). However, the interannual variation in precipitation in the seedling and filling stages was higher in the northern cities; there was a higher risk of waterlogging caused by the sharp increase in precipitation, especially in Zhengzhou. Waterlogging occurred in most growth periods in Nanjing and Hefei, but the filling stage was more consistent with yield reduction. Waterlogging and yield reduction mainly occurred in the seedling stage in Shijiazhuang, Jinan, and Zhengzhou. Different measures should be taken to relieve waterlogging according to the characteristics of waterlogging in different areas. Prevention measures against waterlogging, such as drainage ditches, should become routine operations in Nanjing and Hefei. Moreover, waterlogging in the seedling stage in Shijiazhuang, Jinan, and Zhengzhou must be taken seriously, and some emergency measures, such as plant growth regulators, must be prepared.