Grid-quantification study on the effect of rapid urbanization on hydrological processes Open Access

Urbanization is an important indicator to measure the development level of a country or region (Al-Zahrani 2018; Gao et al. 2018).With the rapid development of economic construction and the continuous increase of urbanization level, its influence on the hydrological process may not be ignored (Ren et al. 2006; Wang et al. 2020). On the one hand, in the context of increasing extreme weather, climate change will have a significant impact on hydrological processes (de Moraes et al. 2018). Many studies have shown that climate change altered regional hydrological conditions, and cities amplified the hydrological changes of natural watersheds (Changnon 1979; Jauregui 1995; Wu & Huang 2009; Mei 2019). On the other hand, in the process of urbanization, a large amount of agricultural land or other non-urban land is converted into impervious land, and land-use change completely alters the natural hydrological process (Hu et al. 2020; Kastridis et al. 2020). Changes in urban land cover will lead to changes in the process of runoff generation, and significant changes in runoff depth (net rainfall), evaporation, and infiltration (Liu et al. 2014; Xu et al. 2021; Zhao et al. 2021). Therefore, it is important to study the hydrological effects of cities quantitatively and qualitatively. It plays an important role in flood warning, water resources planning and management, and urban planning and management (Leandro et al. 2016; Martins et al. 2018). But these changes are hard to describe quantitatively.

At present, to quantitatively analyze the impact of urbanization on the process of runoff generation, urbanization is regarded as a process by most studies, and the impervious area is selected as an indicator to measure the impact of urbanization on the process of runoff generation (Shi et al. 2001; Zhang 2013; Brendel et al. 2021). Brun & Band (2000) studied the changes in the runoff process of the Upper Gwens Falls Basin in Baltimore, Massachusetts, USA from 1973 to 1990 based on the HSPF model. The study found that the impact of urbanization on the runoff process increased with the development of urbanization. When the impervious area reached 20% of the entire basin, the runoff process has a more significant change. Wang & Cheng (2002) analyzed the runoff process after urbanization in Hangzhou and also found that under the condition that the impervious area exceeds 20% of the total area, the runoff caused by the design rainfall once in three years has increased by 50%–100%. Bhaduri et al. (2000) developed a long-term hydrological impact assessment (L-THIA) model using the curve number (CN) method to simulate the rainfall-runoff process changes in the Kitty Hawk River Basin from 1973 to 1991, and found that impervious areas And the construction of drainage systems increases the impact of urbanization on runoff processes.

However, in urbanization, how land-use change affects the hydrological process is very complicated, and it will impact all aspects of the runoff process (Al-Zahrani 2018; Du 2020). In addition, urban resources (i.e., population, impervious area, building density, green space, etc.) are unevenly distributed in urban areas. Therefore, the redistribution of rainfall-runoff becomes more complicated and highly dependent on temporal and spatial distribution and land-use changes (Sarauskiene et al. 2020; Wang et al. 2020). Therefore, simply taking the area of impervious area as a research index cannot represent accurate qualitative and quantitative research. There is a need for a systematic approach, combining different land use types and soil types, and based on more accurate hourly rainfall and runoff data, to quantitatively analyze the impact of unbalanced resource distribution within cities on hydrological processes.

Zhengzhou, as one of the central cities in China, has developed rapidly in urbanization in recent years. Different development stages of the city are obvious, and there is a certain regional imbalance phenomenon. Meanwhile, Zhengzhou has a good network of rain stations and hydrologic stations, which can provide complete research data. In this paper, we take the Jialu River Basin as the research area (including the downtown area of Zhengzhou) and divide the research route into 3 steps. (1) We build a Gridded Surface/Subsurface Hydrologic Analysis Model (GSSHA) of the soil-land use index grid. (2) Use the measured precipitation-discharge data to calibrate and verify the parameters of the model. (3) Finally, the impact of land use on key urban hydrological process elements such as evaporation, infiltration and runoff depth is quantitatively described to provide a scientific basis for guiding urban development (Figure 1).

In this study, Zhongmou station control basin was used as the study area, with a total area of 2,276.9 km², including the urban area of Zhengzhou's downtown area of 1,008 km², accounting for 44%. The average annual rainfall in the basin is 642.3 mm. The rainfall from June to September accounts for 60% to 70% of the annual rainfall, mostly short-term heavy rainfall. The annual average temperature is 14.5 °C (Du 2020). There are 31 rainfall stations and 8 weather stations in the basin, which can accurately reflect the spatial distribution and changes of rainfall.

In the basin controlled by the Zhongmou hydrological station, the urban area is mainly concentrated in Zhengzhou, so only the impact of the Zhengzhou incident on the production and confluence process of the basin is considered. In the natural watershed outside the city, the water system is less disturbed by human activities. The water system based on elevation analysis can better reflect the actual situation of the basin, and the water system extracted by the TOPAZ system is used. In urban areas, the water system extracted by the TOPAZ system has been adjusted to adapt to the water system in the urban planning map. ArcGIS to get the elevation of the connection point of the rainwater utilities network and the length between the connection points to calculate the slope. Based on the above method and the obtained data, a land-based confluence model of the basin controlled by the Zhongmu Hydrological Station with a resolution of 1 km × 1 km was established (Figure 2).

To analyze the dynamic changes of land use distribution during urbanization, satellite remote sensing images of 1998, 2003, and 2008 (data from geospatial data cloud, resolution 30 m × 30 m, acquisition months from May to October) were selected. ENVI and ArcGIS were used to process and analyze remote sensing data based on the classification regression tree (CART) algorithm. The CART algorithm can select variables related to classification and improve the accuracy of land use extraction (Chen et al. 2008). Zhengzhou city has a high degree of land use. According to this characteristic, the land use types are divided into construction land, cultivated land, grassland, forest land, and water area.

With the acceleration of Zhengzhou's urbanization process, the urban area sped up its expansion around 1998. Based on this, the urbanization process of Zhengzhou city was divided into a slow urbanization period before 1998 and a sped-up urbanization period after 1998, and the influence of different urbanization development periods on the hydrological process was analyzed (Wang et al. 2020). From 1971 to 1998, urban development in the basin was relatively balanced and land-use change was slow. The land use data at the end of 1998 were used to simulate land use distribution from 1999 to 2012, urban areas expanded rapidly, the land-use conversion rate was fast, and inter-regional development imbalance intensified.

To accurately simulate the land-use situation at this stage, land use data were updated every five years. The watershed grid index map was generated from land use and soil data in 1998, 2003, and 2008, as shown in Figure 3. Among them, the soil data is based on the soil type file in the NRCS SSURGO soil database, and the elevation data of the watershed controlled by the Zhongmu Hydrological Station is processed through ArcGIS (Figure 4).

The rainfall data came from 31 rainfall stations in Zhengzhou, and the average rainfall in the study area was analyzed by the Thiessen polygon method. The rainfall from 1997 to 1998 was selected in the model simulation period, and the rainfall from 1971 to 1972 and 2003 to 2004 was selected in the model verification period (Table 1).

Different rainfall events in the simulation period and the verification period can well reflect the characteristics of the rainfall process in the watershed controlled by the Zhongmu Hydrological Station. The average rainfall is between 18.29 mm and 181.83 mm, the average rainfall intensity is between 0.69 mm/h and 3.55 mm/h, and the maximum rainfall intensity in one hour is between 1.84 mm/h and 21.65 mm/h. The rainfall in 2003062914 and 2004071520 events reached torrential rain level, and the maximum rain intensity in 1 hour reached 15.67 mm/h and 21.65 mm/h respectively.

Under different rainfall distribution characteristics, the impact of land-use change on the hydrological process is different. Therefore, it is necessary to analyze the influence of urbanization on hydrological processes under the condition of controlling soil moisture degree in rainfall basins. The rainfall process of 1978063023, 1998081404, and 2003082520 in the rainfall center of the lower, middle, and upper parts of the basin can reflect the scenarios of most rainfall processes in the basin, so it is selected as a typical rainfall process, and the time distribution characteristics are shown in Table 2. Spatial distribution is shown in Figure 5.

The GSSHA model is a physical-based distributed hydrological model developed and improved by the US Army Engineering Research and Development Center based on the CASC2D model (Li et al. 2017). GSSHA can be coupled with the SUPERLINK module to simulate rainwater pipe drainage flow (Brendel et al. 2021). It is based on 10–1,000 m grid cells, runs on a digital elevation model (DEM), uses finite difference and finite volume methods for each grid cell and channel reach, and simulates surface and groundwater flow, channel flow, channels, rainwater pipe network, evapotranspiration, various hydraulic structures, and nutrient (pollutant) transport scenarios, solve the physics-based partial differential equations describing the runoff process, which can describe the micro-topography, canopy interception, infiltration and Evapotranspiration, two-dimensional overland flow, and one-dimensional channel evolution, mainly including watershed analysis module, precipitation model block, evaporation module, infiltration module, confluence module and water exchange module.

This paper uses DEM data with a resolution of 30 m × 30 m to construct the hydrogeographic base of the basin runoff generation model and extract the flow direction of each channel. Considering the actual situation in the study area, the Deardroff method was selected in the evaporation module. The method parameters are simple, the method has strong applicability, and the required data is easy to obtain; selects the Redistribution Green-Ampt Model, considering the redistribution of soil water to calculate the runoff infiltration process in the infiltration module. The model's built-in multilevel single-linkage (MLSL) method is used to automatically calibrate model parameters using measured rainfall and its corresponding flow data, this method is selected in the optimization module for optimization (Table 3).

The generation of the river basin network index is the most basic link of the GSSHA model. Firstly, a grid of 10–1,000 m is established according to the actual situation of the watershed, natural geographical conditions, and underlying surface conditions. Then, it is classified according to the unique characteristics of the grid's land-use type, soil type, or combination of land use and soil to generate a watershed grid index. The detailed description of the micro-topography, canopy interception, infiltration, and evapotranspiration is completed by setting the grid parameters. The water produced by the grid is discharged out of the basin through the runoff confluence path.

The Nash efficiency coefficient (NSE), relative error of flood peak (REP), and relative error of runoff (RER) for model calibration were selected, and the coefficient of determination (r²) used for model evaluation.

• (1)

  Nash-Sutcliffe efficiency coefficient (NSE)

Based on the 6 flood events at regular intervals, the MLSL method was used to calibrate the model parameters (Yang et al. 2021). The model regularly simulates the hydrological process of the Zhongmou hydrological station control basin with an NSE above 0.80, and the REP is 6.35%-14.76%. The relative error of flood volume is between −2.86% and 10.16%. The calibration results are shown in Figure 6 and Table 4.

The 8 flood events during the verification period were input into the model to get the simulated runoff process and then analyzed with the runoff data measured by the Zhongmou hydrological station. The result proves that the calibrated model can well simulate the hydrological process in the basin controlled by the Zhongmou Hydrological Station. In the verification period, the NSE is above 0.79, the REP is between −12.35% and 12.99%, and the relative error of flood volume is between −10.64% and 10.39%. The verification results are shown in Figure 7 and Table 5.

We scatter the observed and simulated discharge values (Figure 8(a)). The value of r² is 0.87, showing that this model could well reflect the relationship between observed and simulated discharge. Besides, the value is almost near the fit line in the model. However, the model appears some abnormal values. The reason for this phenomenon is that the model has some fluctuations under the sudden changes in rainfall and discharge data. Also, the fitting effect of the cumulative distribution function (Figure 8(b)) is very good, which reflects the good simulation effect of this model.

The changes of a runoff depth corresponding to 1978063023, 1998081404, and 2003082520 rainfall events under the land use distribution in 1993, 1998, 2003, and 2008 were statistically analyzed, and the results were shown in Figure 9. Combined with the change of land use distribution, ArcGIS was used to analyze the simulated runoff depth data, and the results are shown in Table 6.

Comparing the data in the above table, it is found that the change of runoff depth is not only related to the degree of land development and utilization but also affected by the surface roughness. When other land types were converted to building land, the change of runoff depth was greater than grassland, woodland, and cultivated land. The runoff depth increases by 0.94 × 10⁻¹ ∼ 2.42 × 10⁻¹ mm/km² when other land-use types are converted to building land. The runoff depth varies from −0.67 × 10⁻¹ to 1.19 × 10⁻¹ mm/km² when other land uses convert to each other.

The runoff decreased by 0.43 × 10⁻¹ ∼ 0.64 × 10⁻¹ mm/km² when grassland was converted into forest land. In converting cropland to forest and grassland, the runoff depth decreases by 0.49 × 10⁻¹ ∼ 0.73 × 10⁻¹ mm/km² when the cropland is converted to the forest and 0.44 × 10⁻¹ ∼ 0.67 × 10⁻¹ mm when the cropland is converted to grassland.

The corresponding changes of infiltration depth of 1978063023, 1998081404, and 2003082520 rainfall events under land use distribution in 1993, 1998, 2003, and 2008 were statistically analyzed, and the results were shown in Figure 10. Combined with land use distribution, ArcGIS was used to analyze the simulated depth of infiltration, and the results are shown in Table 7.

By comparing the change of infiltration depth caused by the mutual transformation of different land-use types, it can be found that the degree of land development and utilization in the process of urbanization will have a great influence on infiltration, among which the hardening of the underlying surface is the biggest influence. By comparing the changes in the infiltration depth under different land use transformation conditions, it is found that the impact of land-use change on the infiltration process during urbanization increases with the intensification of human exploitation and utilization of the underlying surface. The intensity of development and utilization of other land-use types of building land to building land is greater than that between them, and the change of infiltration process caused by the former is greater than that caused by the latter. When other land-use types are converted to building land, the infiltration depth decreases by 0.80 × 10⁻¹ ∼ 2.18 × 10⁻¹ mm/km². When other land uses are converted to each other, the depth of infiltration varies from −0.95 × 10⁻¹ ∼ 0.58 × 10⁻¹ mm/km².

Through afforestation, returning farmland to forest and grassland, the depth of infiltration can be restored to different degrees. In the process of afforestation, the infiltration depth increased by 0.36 × 10⁻¹ ∼ 0.55 × 10⁻¹ mm/km² when the grassland was converted into forest land. In converting farmland to forest and grassland, the infiltration depth increased by 0.39 × 10⁻¹ ∼ 0.58 × 10⁻¹ mm/km² when the farmland was converted to the forest and 0.35 × 10⁻¹ ∼ 0.53 × 10⁻¹ mm/km² when the farmland was converted to grassland.

Under the land use distribution in 1993, 1998, 2003, and 2008, a statistical analysis was carried out on the evaporation changes of 1978063023, 1998081404, and 2003082520 rainfall events (Figure 11). At the same time, combined with the land use distribution, ArcGIS was used to analyze and simulate the evaporation data (Table 8).

By comparing the data in the table above, it can be found that evaporation is closely related to the area of green plants. The greater the greening degree is, the greater the evaporation will be. On the contrary, when the grassland and woodland farmland is converted to construction land, the evaporation will decrease. When other land-use types convert to construction land, evaporation decreases by 0.14 × 10⁻¹ ∼ 0.28 × 10⁻¹ mm/km². When other land-use types convert to each other, evaporation changes by −0.24 × 10⁻¹ ∼ 0.14 × 10⁻¹ mm/km².

Evaporation area can be increased by increasing greening, such as the conversion of building land or farmland to forest land or conversion of grassland to grassland to forest land. In the process of afforestation, evaporation increases by 0.06 × 10⁻¹ ∼ 0.10 × 10⁻¹ mm/km² when grassland is converted into forestland, and 0.09 × 10⁻¹ ∼ 0.14 × 10⁻¹ mm/km² when farmland is converted into forestland. Evaporation increases by 0.08 × 10⁻¹ ∼ 0.13 × 10⁻¹ mm/km² of cultivated land converted to grassland.

Through the statistical analysis of the contribution rate of different hydrological processes to the total runoff change under different land-use change models, it is found that urbanization mainly affects the hydrological process by influencing the infiltration process. During urban expansion, other land-use types were transformed to building land, and the contribution rate of the decrease of infiltration process to the increase of runoff process was over 80%. In the transformation from forestland to building land, the decrease of infiltration process contributes the most to the increase of runoff process, up to 90%. In the transformation from cultivated land to building land, the contribution rate of the decrease of infiltration process to the increase of runoff process is the highest and the lowest, reaching 82%. Ecological restoration measures, such as afforestation and conversion of farmland to forest, mainly reduce runoff by improving infiltration soil structure. In the transition from forest land to grassland, from cultivated land to forest land, and from cultivated land to grassland, the increase of infiltration contributed over 80% to the decrease of the runoff process.

By longitudinal comparison of data results in Tables 6–8 (as shown in Figure 12), it can be found that for the same type of land use transformation, the change in runoff depth, infiltration depth, and evaporation is consistent. This shows that the three hydrological factors are not only affected by the land-use change but also their correlation with each other cannot be ignored. At the same time, it can be found that the contribution in the infiltration is more than evaporation in the runoff process. When a variety of land-use types are converted to building land, the runoff depth increases, and the runoff velocity speeds up because of the decrease in roughness. The decrease of infiltration is not only because of the hardening of the underlying surface but also has a certain relationship with the change of runoff velocity. The decrease of infiltration depth will also reduce evaporation to a certain extent.

With the acceleration of urbanization, land-use change is mainly manifested as the increase of underlying surface hardening area and the decrease of green area, there is the increase of building land. It is this change that leads to the increase of runoff depth and the decrease of infiltration and evaporation. Through afforestation, the increasing green area can reduce the probability of urban floods to a certain extent. It also provides ideas and reference points for the study of runoff in sponge city construction and future urbanization.

In this study, a soil-land use index GSSHA model was established, and parameter calibration and model validation were carried out by using the measured rainfall-runoff data in Zhengzhou. The innovation of this paper lies in the establishment of GSSHA based on soil land use index grid, which simulates the impact of urbanization on hydrological processes and quantitatively analyzes the impacts and contributions of different land-use transformations on urban hydrological processes. The main findings are as follows:

• (1)

  Land-use change and human use of the urban underlying surface jointly affect hydrological processes. When grassland and forest land are converted into construction land, the changes of runoff infiltration and evaporation are larger than those of grassland, forest and cultivated land.

• (2)

  Afforestation, returning farmland to grassland, and other vegetation restoration measures can reduce the hydrological process changes caused by land-use changes in urbanization to varying degrees.

• (3)

  Land use change affects urban expansion mainly through influencing the infiltration process and hydrological process. Other land-use types change to building land, and the contribution rate of decrease of infiltration process to increase of runoff process is over 80%. In the transition from forest land to grassland, from cultivated land to forest land, and from cultivated land to grassland, the increase of infiltration contributed over 80% to the decrease of the runoff process.

For this research paper with several authors, a brief paragraph specifying their individual contributions was provided. Caihong Hu and Fan Yang developed the original idea and contributed to the research design for the study. Sun Yue handled data collecting. Chenchen Zhao and Jingyi Wang provided guidance and contributed to the research design. Chengshuai Liu and Shan-e-hyder Soomro provided some guidance for the writing of the article. All authors have read and approved the final manuscript.

This work was funded by Key projects of National Natural Science Foundation of China, grant number 51739009, National Natural Science Foundation of China, grant number 51979250.

The authors declare that there is no conflict of interest regarding the publication of this paper.

Data cannot be made publicly available; readers should contact the corresponding author for details.