1. Introduction
Historically, the preferred flood management options are engineered structural solutions, such as dams and embankments. Critics of these traditional approaches argued that although they might meet short-term goals, in the long term they had failed to reduce the economic losses from flooding and when flooding increases it is impractical, expensive and unsustainable to continually heighten floodwalls and strengthen structural defenses [
1]. Also, super abundant hydraulic engineering projects could do more harm for the ecological environment and the local climate [
2]. Some researchers have proposed the concept of sustainable flood management (SFM) which emphasizes a natural and sustainable response to flood risk with the need to link flood risk management, agriculture and land use management [
3]. Other scholars proposed the resilience concept of flood risk management and considered resilience as a promising instrument for preventing and mitigating the impacts of hazards [
4,
5]. Resilience is the ability of a system to return to its equilibrium after a reaction to a disturbance [
6]. Flood risk management implies two types of measures distinguished by their aims: structural and non-structural measures. The aim of structural measures is to modify the flood pattern, while non-structural measures focus on reducing flood impacts [
7]. Even though dams have a big advantage in flood control and play an irreplaceable role in safeguarding cities, they are not ideal for medium or small flood control due to management costs and the waste of water resources.
Non-point flood alleviation of land use management fits this new trend. The concept of non-point flood alleviation is centered on using surface measures to assist point measures (dams and embankments). By changing the land use, flood protection of the basin can be more effective while the focus of non-point flood alleviation is to retain rainfall on the surface and in soil pores.
Land use plays a very important role in the hydrological processes of a certain basin [
8]. Plenty of researches and experiments have proved that different land uses can lead to changes in the runoff process. Beven et al. [
9] compared the land use and the floods happening in the same period with the land use changed in the past to find the historical connection. Woldeamlak [
10] analyzed the pattern of the runoff in northwestern highland Ethiopia with the dynamics method and confirmed that the decrease of forest, the increase of the agriculture land, the grassland degradation and artificial afforestation would all be crucial factors having an influence on the runoff process. McIntyre [
11] testified the connection between floods and land use at the regional scale. Li et al. [
12] quantified the effects of land use change on flood peaks and volumes in Daqinghe River basin with a multiple-linear regression analysis method. Now it has been established that the variable of land use can change the flood pattern, whether it can achieve the aim of flood control by increasing some sort of land use which benefits rainfall retention still remains to be explored. This flood control measure is different from the punctate measure of dams which is called non-point flood alleviation.
The main effect of the land use on runoff/flood is to reduce the rainfall energy by interception, sink filling, transpiration, evapotranspiration and infiltration [
13]. Nie et al. [
14] demonstrated in the upper San Pedro watershed that construction land and the decrease of forests would cause surface runoff to increase. A similar conclusion was made by Lin [
15] in Willow Basin, Canada. The negative changes aforementioned could lead to the degradation of the soil which can lower the infiltration, causing poorer connectivity and impairing the water-holding capacity, thus increasing the flood risk [
16]. According to Feng [
17], grassland can delay the runoff three-fold compared to bare land. The average speed of the runoff on grassland is only 40% of that on bare land. The root system of the grass increases in surface soil voids. The grass covering the surface can also ease the impact of rain drops hitting on soil and compromise the integrity of the soil surface, while also increasing the roughness of the surface soil, so the runoff coefficient becomes significantly smaller [
18]. Fan et al. [
19] compared bare land, grassland and woodland runoff over years with a monthly distribution. Runoff of bare land, grassland and woodland accounted for rainfall over the same period of 18.6%, 5.1%, 3.7%, respectively; and the uniformity coefficients of the bare land, grassland and woodland for the distribution of monthly runoff were 0.99, 0.74, and 0.69 [
19]. Therefore, planting trees and grass reduces runoff and improves the uniformity of the distribution of runoff. As for agricultural land, there has been plenty of studies proving its benefit in preventing floods. Sujono [
20] showed through experiments that some measures of agricultural planting can reduce surface runoff. Evans et al. [
21] showed that certain measures at autumn sowing can lower the frequency of muddy floods. The phenomena can be explained by the fact that with the measures carried out, the soil structure is changed and the land surface is uneven, so the rain either goes to the deep soil layers through the gaps or fills in the depressions [
22]. In addition, the residue covering can protect the land surface from being hit by the rain drops, and slows down the speed of infiltration [
23].
Despite the researches aforementioned, the effects of the land use change on the runoff/flood still remain unclear, especially at basin scale [
24]. For some basins, the flood process is influenced by several factors, such as rainfall, elevation, soil depth, and water steepness [
25], and land use change depends on the former type, the change type, the location, the change time and the like [
26].
Table 1 lists some researches trying to distinguish the effects with various methods.
This paper used the SWAT model to simulate flood processes with different frequency under different land use scenarios in Yongding River Basin, and analyzed the quantitative effect of non-point flood alleviation of each land use scenario. Yongding River basin locates in the upstream region of Beijing, which is an important geographical location with high demand for flood control and is one of China’s four major river flood control areas [
35]. In history, flood was the major disaster in this region. From 1949 to 2001, flood disasters occurred 22 times, once in every 2.4 years [
36]. However, extreme floods rarely occurred as precipitation drastically decreased in recent years while medium and small floods became the main problem. It is reasonable to consider land use management to practice non-point flood alleviation in the Yongding River Basin. The upper reaches of the Yongding River Basin are the Loess Plateau, where soil has a point edge contact support structure, high water permeability and small water storage capacity. Non-point flood alleviation in the Yongding River Basin has the potential to play an important role in flood control.
The objects of this paper are: (1) simulating different floods with different frequencies under the chosen land use scenarios; (2) analyzing the characteristics of flood processes under integrated and single land use changes; (3) determining the effects of non-point flood alleviation and its main influencing factors.
6. Conclusions
To give a general view of the potential of non-point flood alleviation, we averaged the figures separately of the 1-day runoff, 7-day runoff and the runoff depth of all three kinds of floods. The land use of 2005 can on average reduce floods according to the three indicators by 10.7%, 9.7%, 12.3%; the agricultural land was reduced by 20.0%, 20.3%, 14.3%; the grass land was reduced by 3.7%, 10.7%, 20% and the urban/construction land increased by 19.0%, 9.3%, 2.3%.
As a concept, how non-point flood alleviation can be put into use and the practical results remain to be explained. We chose the SWAT model to solve this problem. Through the simulations of land uses of the years 1995 and 2005, we think the results present a relatively clear demonstration of the pattern of non-point flood alleviation.
Non-point flood alleviation is based on the response of runoff/floods to changes in land use. The simulations show that S2 is better in reducing the floods under different rainfall amounts, which indicates that the land use in 2005 functions better in reducing scattered floods than that of 1995. The same goes for the single land use transformation. According to the same three indicators, we can determine that agricultural land and grassland are favorable to reducing scattered floods while urban/construction land is not conducive to mitigating scattered floods in Yongding River basin.
When comparing the return periods of the floods, we discovered that non-point flood alleviation in Yongding River basin has a stronger effect on small floods, and with the return period increasing, the effect is diminished.
Because of land use change, we assume that non-point flood alleviation depends on the change in water-holding capacity. Non-point flood alleviation can fill in the gap in present flood control management as a new measure. One possibility of non-point flood alleviation could consist of a soil water bank on the bottom of the basin, and using various land use measures to prevent floods. For larger floods, the usual engineering projects like dams and reservoirs still are the most effective means. However, for medium and small floods, non-point flood alleviation of land use is attracting more attention. By using land use management to prevent floods, the flood control capacity of Yongding River basin has been improved, the pressure on flood control downstream has been eased and the comprehensive utilization of rainwater has been strengthened at the same time. In addition, non-point flood alleviation is a non-engineering measure, thus, the degree of damage to the ecological environment damage degree has been reduced, which is more conducive to the sustainable development of water resources in the region.