Effects of Yellow River Water Management Policies on Annual Irrigation Water Usage from Canals and Groundwater in Yucheng City, China

Abstract

The Yellow River Water Allocation Management Method was put into place in 1998 to decrease the Yellow River water amount used by upstream areas and provide more water to downstream regions. Rainfall and Yellow River’s infiltration are the main groundwater supply in the downstream area of Yellow River. The groundwater table in the downstream area has continued to decrease since 1979, and the extracted groundwater for irrigation is the main reason for this. Whether the increased river water amount could improve the decreased groundwater level is uncertain. Therefore, we used remote sensing images, groundwater level observations, meteorological data, and unit mean irrigation rate to identify the irrigation events for river water and groundwater, estimate the annual river water irrigation amount and groundwater irrigation amount, and analyze the effects of river water allocation on the groundwater table. Our analysis showed that the area of double-irrigated farmland (farmland that could be irrigated by both groundwater and river water) tended to decrease, while well-irrigated farmland area (farmland that could only be irrigated by groundwater) remained unchanged during the study period. The number of annual irrigation events tended to increase, and the usage of river water remained consistent throughout this period. The increased number of well irrigation events caused annual groundwater usage for irrigation to increase. However, the usage of river water for irrigation remained stable. The increased usage of groundwater for irrigation led the groundwater table to continually decrease from 1998 to 2019. This indicates that there are shortcomings to the current water allocation policy, and that further improvements are needed to prevent continued decrease in groundwater levels.

1. Introduction

Climate change and rapid urbanization have caused the Yellow River runoff to decrease continually since 1979, and a cutoff in the flow of water to downstream regions occurred in 1972 [1]. To increase the runoff and make more water available for downstream areas, the Water Allocation Management Method, Emergency Response Regulations for Yellow River Water Allocation, and Yellow River Water Allocation Policy were issued in 1998, 2003, and 2006, respectively [2]. The Yellow River Water Conservancy Commission of the Ministry of Water Resources was then created with the goal of decreasing the amount of water used in upstream areas of the Yellow River and allocating more water to downstream areas. However, in the downstream area, the volume of allocated river water did not meet the demand such that groundwater had to be extracted.

The downstream area of the Yellow River is part of the North China Plain. Rainfall and Yellow River infiltration are the main sources of groundwater. During recent decades, the groundwater table has decreased in the downstream area of the Yellow River. For example, the groundwater table in the study region decreased 0.46 ± 0.37 m/year for shallow groundwater and 1.14 ± 0.58 m/year for deep groundwater [3,4]. This decline in the groundwater table was caused by groundwater extraction for irrigation, industrial applications, services, and residential use; of these uses, irrigation is thought to be the most important [5]. However, whether water allocation policies would ameliorate the groundwater table decline remains unknown. Thus, estimating the volume of groundwater used for irrigation is critical for analyzing the effects of Yellow River water management policies on the groundwater table decrease.

Most studies have focused only on the estimated total amount of irrigation water without dividing it into irrigation water from canal water versus groundwater. The amount of groundwater used for irrigation is not known, and the effects of the Yellow River water management policies on the groundwater table have also remained unclear. For instance, some studies have estimated potential evapotranspiration with the Penman–Monteith equation based on weather observation data and calculated the amount of irrigation water based on irrigation efficiency [6,7,8]. Some studies have applied machine learning models to estimate irrigation water demand and thereby estimate irrigation water usage. For instance, Mohammad et al. [9] used machine learning models to estimate annual irrigation water demand and irrigation water usage at the county level in California. Wei et al. [10] applied machine learning to estimate the irrigation water consumption on the Kansas High Plain. Parsinejad et al. [11] and Rahimian et al. [12] adopted remote sensing images with the SEBAL model to estimate the amount of irrigation water.

One reason for the inability to estimate the consumption of groundwater versus river water for irrigation is that farmland has not previously been classified into well-irrigated farmland (i.e., farmland that can be only irrigated with groundwater) and double-irrigated farmland. (i.e., farmland that can be irrigated with both canal water and groundwater). In a previous study, we designed a method to classify farmland into various categories, but the numbers of irrigation events that used canal water and groundwater could not be determined [13].

Therefore, in the present study, we designed a method that determines the number of irrigation events using river water and groundwater based on observed groundwater table data and rainfall data and then estimates the annual usage of each water source for irrigation by classifying the farmland into various irrigation types. We used this method to explore the effects of Yellow River water management policies on water use for irrigation and groundwater levels.

2. Study Area

Yucheng City is located in northwestern Shandong Province, north of the downstream plain of the Yellow River. The main sources of irrigation water are the Yellow River and groundwater. The city is composed of 12 townships, with a total area of 990 km² [14]. The altitude in the study area ranges from 19.3 to 27.3 m above mean sea level (Figure 1). Average annual rainfall is 555.5 mm, and annual mean temperature is 13.3 °C. The major crop types in this area are corn and wheat. The soils of the region are dominated by yellow soil and yellow-brown soil [15].

3. Methodology

3.1. Data Collection

Four datasets were used to identify irrigation events and estimate annual irrigation amounts from river water and groundwater during the study period. The first dataset included well-irrigated farmland and double-irrigated farmland. The irrigation categories of farmland were derived from aerial films and high-resolution remote sensing images. The irrigation categories in 1998 were obtained from aerial film before 1998. These aerial films came from Dezhou Bureau of Natural Resource (http://zrzyj.dezhou.gov.cn/, accessed on 1 September 2022). The irrigation categories in 2008 were obtained from high-resolution images taken in 2008, also from the bureau. Irrigation categories in 2017 were derived from Gao Fen Er Hao images at 0.8 m resolution recorded in early May 2018, using the visual interpretation method. The remote sensing images in 2018 were downloaded from the China Centre for Resources Satellite Data and Application (www.cresda.com, accessed on 1 September 2022).

The second dataset included annual irrigation events of winter wheat using river water and groundwater. The dataset was derived from groundwater level data collected every 5 days and meteorological data collected daily. The station used for recording the groundwater levels was located in double-irrigated farmland. The groundwater table data came from the local water resources bureau recorded by the Shandong Water Resources Department (1996–2018). Daily meteorological data for Yucheng were obtained from the weather observation station in the study area, recorded by the State Meteorological Bureau. The third dataset included mean irrigation water usage per hectare per irrigation event. The dataset was obtained from the Academy of Water Resources of Shandong Province [16]. The fourth dataset included annual rainfall data, derived from daily meteorological data obtained from weather observation stations in Yucheng City, recorded by the State Meteorological Bureau.

3.2. Data Analysis

3.2.1. Well-Irrigated Cropland and Double-Irrigated Cropland Extraction

All farmland in the study area can be irrigated by groundwater and could therefore be classified as well-irrigated cropland. However, only some croplands (e.g., lands adjacent to rivers or canals or connected by sublateral canals) can be classified as double-irrigated cropland. Therefore, croplands were classified into two categories: well-irrigated cropland and double-irrigated cropland (i.e., cropland that could be irrigated by both groundwater and canal/river water). The detailed method is described in a previous study [13].

To classify cropland, the croplands were divided into many land parcels by roads, rivers, canals, various landforms, and administrative units. Land parcels were initially extracted from remote sensing images using the visual interpretation method from the high-resolution images. According to our investigation, if a land parcel is within 300 m of the river or canals and the river or canals contain enough water, farmers often connect the land parcel to the river through water belts. Therefore, these land parcels were classified as canal-irrigated croplands using the visual interpretation method. Sublateral canals are a type of connection ditch between rivers/canals and land parcels or between wells and land parcels; these canals are approximately 1.0 m wide. The pan-bands of the Gao Fen Er Hao images and aerial photographs are sensitive to water, and the sublateral canals could be identified from high-resolution images or aerial films. When a land parcel was located far from the river but was connected to it by sublateral canals, it was also classified as canal-irrigated cropland. All other land parcels were classified as well-irrigated cropland. Aerial films taken before 1998 were used to obtain the irrigation categories for 1998. High-resolution images taken in 2008 were used to obtain the irrigation categories for 2008. Gao Fen Er Hao images at 0.8 m resolution recorded in early May 2018 were used to obtain the irrigation categories for 2017.

Our categorizations in 1998, 2008, and 2017 were checked against 30 samples of canal-irrigated cropland and 30 samples of well-irrigated cropland; the irrigation categories were unchanged except in the case of transformed village and urban areas. All samples were randomly selected in the study area, and the percentage of accurately classified irrigated cropland was above 90% (

Table 1. Accuracy of classified cropland.

	Canal-Irrigated Cropland		Well-Irrigated Cropland
	Sample	Accuracy	Sample	Accuracy
1998	30	92%	30	91%
2008	30	94%	30	96%
2017	30	96%	30	95%

), verifying that our classification method was appropriate.

3.2.2. Identification of Irrigation Events and Water Source

The daily rainfall data were combined into the 5-day-interval rainfall data. Next, irrigation events using river water and groundwater were determined based on the relationship between the groundwater table and rainfall data. The main crop system in this area is a rotation of winter wheat and summer corn. Summer corn is planted in late June and harvested in early October and is irrigated immediately after planting. The summer coincides with the rainy season, and irrigation is therefore not needed before the corn is harvested.

For winter wheat, irrigation is typically carried out in mid-October, mid-November, early March of the following year (or, in some years, mid-March or late March), late April, and late June. Each irrigation event lasts 7 days. Winter wheat is planted in mid-October and is irrigated immediately after planting. After winter, winter wheat begins to grow in early March, at which time the soil is typically dry and requires irrigation. In late April, the winter wheat is irrigated prior to the wheat flowering period. Winter wheat is not typically irrigated again before harvest.

During irrigation periods, crops are not irrigated when rainfall exceeds 20 mm before the typical irrigation date. When irrigation is performed using groundwater as a source, the groundwater table decreases. The groundwater table is unaffected or increases if irrigation uses river water as a source. We therefore determined irrigation source based on the relationship between irrigation and groundwater table dynamics (Figure 2). The accuracy of this method was verified in early March 2018 by field survey, and the result proves that this method is valid (Figure 3). Then, the number of annual irrigation events using river water and groundwater was determined for the double-irrigated cropland and well-irrigated cropland, respectively.

The annual amount of irrigation water for double-irrigated farmland was calculated using the following equations:

$${\mathrm{W}}_{\mathrm{irrigation}}={\mathrm{W}}_{\mathrm{river}}+{\mathrm{W}}_{\mathrm{groundwater}}$$

(1)

$${\mathrm{W}}_{\mathrm{river}}={ \mathrm{A}}_{\mathrm{double}}\times { \mathrm{I}}_{\mathrm{river}–\mathrm{unit} }\times { \mathrm{T}}_{\mathrm{river}}$$

(2)

$${\mathrm{W}}_{\mathrm{groundwater} }={ \mathrm{A}}_{\mathrm{double}}\times { \mathrm{I}}_{\mathrm{well}–\mathrm{unit} }\times { \mathrm{T}}_{\mathrm{groundwater}}$$

(3)

where ${\mathrm{W}}_{\mathrm{irrigation}}$ is the annual amount that is used for irrigation (m³), ${\mathrm{W}}_{\mathrm{groundwater}}$ is the annual amount irrigated by groundwater (m³), ${ \mathrm{I}}_{\mathrm{river}–\mathrm{unit}}$ is the mean river water volume per ha used for irrigation except for very dry or very wet years, ${\mathrm{A}}_{\mathrm{double}}$ is the area that is classified as double-irrigated farmland (ha), ${\mathrm{T}}_{\mathrm{river}}$ is the number of annual irrigation events using river water, A_well is the cropland area that is classified as well-irrigated cropland (m²), I_well-unit is the mean groundwater volume per ha used for irrigation, and T_groundwater is the annual irrigation events using groundwater.

Well-irrigated farmland is irrigated only with groundwater, and the number of annual irrigation events is the same as for double-irrigated farmland. The annual groundwater irrigation amount is calculated using Equation (3).

3.2.3. Slope and Intercept of the Linear Trend

The linear trend of irrigation amount is expressed in Equation (4). The slope and intercept of linear trend, groundwater table, and rainfall from 1998 to 2019 were calculated using Equations (4)–(6) [17]:

$$\mathrm{y}=\mathrm{a}\times {\mathrm{x}}_{\mathrm{year}}+\mathrm{b}$$

(4)

$$\mathrm{a}=\frac{\mathrm{n}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{m}}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{i}}-{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{m}}_{\mathrm{i}}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{i}}}{\mathrm{n}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{m}}_{\mathrm{i}}^2-{\left({{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{m}}_{\mathrm{i}}\right)}^2}$$

(5)

$$\mathrm{b}=\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{i}}}{\mathrm{n}} -\mathrm{a} \times \frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{m}}_{\mathrm{i}}}{\mathrm{n}}$$

(6)

where y is the irrigation amount, or irrigation events, or rainfall (m³); x_year is the year; a is the slope of linear trend; X_i is the value of irrigation amount, groundwater table, or rainfall for year i (i = 1, 2, 3…n); and m_i is the sequence number of the year (m₁ = 1, m₂ = 2, m₃ = 3…m_n = n). Positive and negative values indicate increasing and decreasing trends, respectively. b is the intercept. The slope and intercept are all significant at the 5% level.

4. Results

4.1. Areas of Double-Irrigated and Well-Irrigated Cropland

Based on our classification methods, well-irrigated farmland and double-irrigated farmland regions in 1998, 2008, and 2017 were extracted from aerial films and high-resolution images (Figure 4 and

Table 2. Area of cropland irrigation categories.

	Area of Double-Irrigated Cropland (ha)	Area of Well-Irrigated Cropland (ha)	Total Area (ha)
1998	52,955	4724	57,679
2008	52,431	4724	57,155
2017	51,149	4724	55,873

). The total area of farmland in the region followed a decreasing trend from 1997 to 2019. The area of double-irrigated cropland was significantly higher in this region than in other areas, whereas the area of well-irrigated land was minimal. The area of double-irrigated farmland followed a decreasing trend throughout the study period. The area of well-irrigated cropland was 4724 ha and did not change throughout the study period.

4.2. Annual Irrigation Events

Well-irrigated farmland is normally irrigated by groundwater. In double-irrigated farmland, if crops are irrigated by groundwater, the groundwater level decreases throughout the irrigation period. If crops are irrigated by river water, groundwater level remains undisturbed or increases throughout the irrigation period. Annual irrigation events using groundwater and river water were obtained for the period of 1997–2019 (Figure 5, Figure 6 and Figure 7) and were the same for well-irrigated farmland and double-irrigated farmland.

The number of annual irrigation events for crops varied from two to five and increased from 1998 to 2019 with a slope of 0.021. The maximum number of annual irrigation events was four from 1998 to 2009 and increased to five from 2010 to 2019. The trend of annual groundwater irrigation events also increased from 1998 to 2019. The slope of annual groundwater irrigation events was 0.016 from 1998 to 2019. The number of annual river irrigation events varied from zero to four. The trend of annual river water irrigation events remained consistent throughout the study period, varying from zero to three events annually.

4.3. Annual Usage of Irrigation Water

We calculated the river water and groundwater usage per irrigation event for double-irrigated areas and well-irrigated areas (Figure 8). Then, Equations (1)–(3) were used to obtain annual irrigation amounts for groundwater, river water, and total irrigation amounts in the study area. The well-irrigated and double-irrigated farmland areas in 1997 were used to calculate the water usage for 1998–2004, those in 2008 were used for 2005–2012, and those in 2018 were used for 2013–2019. According to Figure 8, total irrigation water in this region varied from 5.4 × 10⁸ m³ to 1.4 × 10⁹ m³. The minimum and maximum amounts of irrigation water occurred in 2007 and 2011, respectively. The trend of total irrigation water volume in this region tended to increase, with a slope of 7 × 10⁶.

The total annual volume of river water used for irrigation varied from 0 to 8.4 × 10⁹ m³ (Figure 9). The minimum amounts of river water used for irrigation occurred in 1999, 2004, 2007, and 2016. The maximum amount was in 2002. There was no clear trend in the annual irrigation usage of river water. In most years, the annual volume of irrigation water was 2.8 × 10⁸ m³ before 2013 and 2.72.8 × 10⁸ m³ after 2013. The decreased volume after 2013 can be attributed to a decline in double-irrigated farmland area due to urban expansion. This shows that this was no improvement in the supply of river water for irrigation after the Yellow River Water Allocation Management Method was implemented in 1998.

The total annual groundwater usage for irrigation for this region varied from 6.7 × 10⁷ to 1.1 × 10⁹ m³ from 1998 to 2019 (Figure 10). The minimum and maximum irrigation amounts from groundwater occurred in 2011 and 2010, respectively. Annual groundwater amount in most years varied between 3.0 × 10⁸ m³ and 1.1 × 10⁸ m³. The trend of total annual irrigation amount tended to increase, with a slope of 6.0 × 10⁶.

4.4. Trending Groundwater Level

Figure 11 shows the trend in annual mean groundwater level from 1998 to 2019. The annual mean groundwater level decreased from 21.8 m in 1998 to 21.3 m in 2002, then increased to 22.1 m in 2004. After 2004, it decreased to 20.8 m in 2019.

5. Discussion

The decreasing trend in annual mean groundwater level was influenced by annual rainfall, river water usage for irrigation, and groundwater usage for irrigation. Annual rainfall in this area tended to increase from 1998 to 2019, with a slope of 2.364 (Figure 12). The minimum annual rainfall was 266.9 mm in 2002, and the maximum annual rainfall was 834 mm in 2018. The increasing trend in annual rainfall indicates that rainfall was not the factor that caused groundwater level to decrease.

The amount of groundwater used for irrigation followed an increasing trend, which may be the causative factor for the decrease in groundwater levels. Although Yellow River water management policies were put into place, and policy violations were punished by state agencies, the amount of river water allocated for irrigation remained consistent from 1998 to 2019. This indicates that there are shortcomings in the management policies, and that further improvements are needed to prevent continued decrease in groundwater levels.

Groundwater is also used for industrial and residential purposes; therefore, it was not possible to determine whether irrigation was the main cause of decreased groundwater levels. Because there is a lack of robust data for groundwater usage for industrial and residential purposes, methods of collecting and monitoring such data should be developed.

6. Conclusions

We collected data on irrigation water source, number of annual irrigation events, annual rainfall, meteorological trends, and river and groundwater levels in the study area from 1998 to 2019. The area of double-irrigated farmland decreased from 52,955 ha in 1997 to 52,431 in 2008 and 51,139 in 2017. The area of well-irrigated farmland remained stable during this period. Total annual irrigation events varied from two to five and followed an increasing trend. The number of irrigation events that used river water remained consistent throughout the study period, while the annual amount of groundwater used for irrigation increased, leading to a continuous decrease in the groundwater table from 1998 to 2019. This trend shows that the Yellow River allocation policy must be improved to maintain groundwater levels.