Investigation into impacts of land-use changes on floods in the upper Huaihe River basin , China

To investigate the agricultural land-use change on flood regime, the upper Huaihe River basin above the Dapoling station was selected as the case study site. Based on topography, land-use, hydrological and meteorological data in 1990s and 2010s, the improved distributed Xinanjiang model, with potential evapotranspiration being computed by coupling a dual-source evapotranspiration model with a simplified plant growth model, was adopted to simulate the daily and hourly rainfall-runoff processes over 1990s and 2010s, and then the effects of land-use change on flood volume, flood peak, occurring time of flood peak, the percentage of surface runoff component were investigated respectively. The results was interesting and indicated that impacts of land-use change on flood characteristics varied significantly with land-use types. The outputs could provide valuable references for flood risk management and water resources management in the Huaihe River basin.


Introduction
In the last decades many inundations have occurred in China causing huge loss of human lives and financial damages, which have been aggravated by the intense urbanization and deforestation of flood prone areas.The impact of land use changes on storm runoff generation is currently receiving considerable public attention.It is recognized that there is an impact of land use on hydrology and flood response, however the links between land use change and flooding are still not very clear.The land use and flooding research mainly focused on the effects of urbanization (Suriya and Mudgal, 2012) and deforestation (Green and Alila, 2012), but lacking comprehensive concerns that the effects of agricultural intensification and the impact of land use practices and land management on the risk of flooding.To investigate the agricultural land-use change on flood regime, the upper Huaihe River basin, with a history of flooding over many centuries, above the Dapoling station was selected as the case study site, where the land-use was dominated by the agricultural land-use and it changed significantly in recent decades (Cai et al., 2012).Based on the topography, land-use, hydrologi-cal and meteorological data in 1990s and 2010s, the gridded Xinanjiang model (XAJ) was adopted to simulate the flood processes under different land-use patterns.

Model development
In the paper, an improved gridded XAJ model was selected for the hydrological process simulation.Since the evapotranspiration is a critical variable in determination of local water resources available, reliable estimates of evapotranspiration is essential for the improvement of hydrological simulations.In classic XAJ model (Zhao, 1992), evapotranspiration is always simulated in an empirical or conceptual way or mostly in a lumped way.In the study, the distributed dualsource evapotranspiration model (DESM), considering the soil potential evaporation, vegetation potential evaporation, and evaporation of intercepted rainfall together was adopted to calculate the evapotranspiration as one of the inputs of the gridded XAJ model.However, one of the important input -leaf area index (LAI) of the DSEM, was often estimated by using remotely sensed data on a monthly scale, which could not reflect the actual dynamic evolution of plants.A simplified plant growth model, being capable of calculating the daily LAI, was coupled with the DESM, which not only could capture the actual dynamic evapotranspiration evolution of vegetation, but also reflect the evapotranspiration differences of vegetation types for different land-uses.Thus, a semi-distributed XAJ model coupled with the modified DESM was used to simulate water-balance process and flood process.For detailed description of the XAJ model, it can be referred to Zhao (1992).

Case study site
The Huaihe River Basin is one of seven major river basins in China, and lies in the warm temperature semi-humid monsoon region, which is a transition zone between the climates of North and South China, with a history of flooding over many centuries.With intensified human activities including land-use change in the basin, which has China's highest population density (662 persons per km2), the rainfall-runoff relationship has been altered in terms of flood frequency and magnitude.This paper aims to evaluate impacts of landuse change in the upper Huaihe river basin above Dapoling gauge station on its floods.The study area has a catchment area of 1413 km 2 , and was divided into grids with a size of 0.873 km × 0.873 km to consider the spatial unevenness of precipitation and underlying conditions.

Topography data
The digital elevation model (DEM) data was downloaded from the Global Land One kilometer Base Elevation database with a spatial resolution of 0.873 km × 0.873 km, and the Archydro tool was used to obtain the river net and catchment boundary (Fig. 1).

Land-use data
The land-use maps of 1980s, 1990s and 2000s for Dapoling subbasin were collected from the Chinese Academy of Science (see Fig. 2).It can be seen that the spatial distribution of land-use in 1980s and 2000s was very similar, i.e. being dominated by the dryland in the middle of the basin both in 1980s and 2000s; in contrast, paddy land was the main land-use in 1990s.Since the land-use patterns in the upper Huaihe River basin above the Dapoling station were similar in 1980s and 2000s, the land-use maps in 1990s and 2000s were used to investigate the impact of land-use change on the flood regime in the upper Huaihe River basin above Dapoling station.

Hydrological data
The daily precipitation at 13 stations and daily flow discharge at Dapoling hydrological station during 1990-2010 were collected from the Huaihe River Water Resources Commission for the water-balance simulation of the improved the Xinanjiang model, while hourly precipitation at 13 stations and hourly flow discharge at Dapoling station during 1990-2010 were obtained for the flood process simulation.

Meteorological data
The meteorological data, including daily temperature, solar net radiation, humidity, wind speed, etc. at six weather stations in or neighboring the study area during 1990-2010 were collected from the China Meteorological Administration (CMA).

Water-balance Model
The daily XAJ model was calibrated and validated by the use of observed daily flow, precipitation, and meteorological data in 1990s and 2000s.The calibration periods for two phases (1990s and 2000s) were 1990-1996 and 2000-2006 respectively, and their validation periods were 1997-1999 and 2007-2010 correspondingly (Tables 1-3).The simulation performance disclosed that the daily XAJ model can be simulated with satisfactory accuracy, with the determined co-efficient values (DC) both for calibration and validation periods were more than 0.73 in 1990s and 0.93 in 2000s, and the relative error values (RE) both for calibration and validation periods were generally less than 10 % in 1990s and 2000s.The above suggested that parameters of the daily XAJ model could be used as the initial conditions for the flood simulation model.

Flood simulation model
The hourly XAJ model was calibrated and validated by the use of observed hourly flow, precipitation in 1990s and 2000s respectively (Tables 4-6).In 1990s, the DC value was higher than 0.93, RE less than 10 %, RE peak less than 20 % generally (Table 5) both for calibration and validation periods; while in 2000s, the DC value was higher than 0.73, RE less than 10 %, RE peak less than 20 % generally (  and parameterization schemes of hourly XAJ model could be further used to investigate the impact of the land-use change on the flood characteristics in the Dapoling basin in the upper Huaihe River basin.

Flood events simulation under different land-use patterns
Twelve floods (six in 1990s and six in 2000s) were selected to simulate the effect of land-use change on the flood volume, flood peak, flood peak occurring time, and the ratio of surface runoff to the total runoff by using different parameterization schemes (see Table 4).It is worth noting that same precipitation process was applied during the simulation of the same flood event.To save the space, only two flood process was displayed (Figs.3-4).

Impacts of land-use changes on flood volume
Figure 3 showed the temporal variations of hourly flow under different land-use patterns.From Fig. 3 it can be found that compared with the flood process by using the parameterization scheme of 1990s, the corresponding flood process by using the parameterization scheme of 2000s became more uneven.It disclosed that the temporal distribution of flood was changed apparently when the land-use changed from paddy land to dryland.Flood volume accumulation calculation also revealed that the flood volume increased significantly from 8.143 × 10 7 m 3 to 8.413 × 10 7 m 3 for flood #19960628, and from 1.418 × 10 8 m 3 to 1.452 × 10 8 m 3 for flood #20050828 correspondingly from 1990s to 2000s, with the increased percentage of 3.32 and 2.38 % respectively.Clearly, it can be concluded that the land-use changing from paddy land to dryland will increase flood volume significantly under the same precipitation condition.

Impacts of land-use changes on flood peak
Under the same precipitation condition, the flood process became much thinner and the flood peak became much higher (Fig. 3).From 1990s to 2000s, the flood peak increased significantly from 1582.6 to 1825.8 m 3 s −1 with increased percentage of 15.4 % for flood #19960628, and from 2454.2 to 2750.1 m 3 s −1 with increased percentage of 12.1 % for flood #20050828.Nevertheless, the flood peak occurring time also advanced from 0 to 3 h from 1990s to 2000s.This indicated that land-use changing from paddy land to dryland will result in greater flood peak, and earlier flood peak occurring time.total runoff during earlier time of the flood, especially in period with rapid runoff production.

Conclusions
This paper investigated the impacts of agricultural land-use changes on the flood characteristics by using the gridded XAJ model in the upper Huaihe River basin above the Dapoling station.The results revealed that: (1) the XAJ model performed very well in the water-balance simulation and flood process simulation; (2) land-use changing from paddy land to dryland will increase the flood volume and the flood peak, advance the flood peak occurring time, and lead to higher surface runoff percentage of the total runoff during earlier time of the flood.

Figure 1 .
Figure 1.Sketch map and DEM of the upper Huaihe River above the Dapoling station.

Figure 2 .
Figure 2. Land-uses in three-phrases in the upper Huaihe River above the Dapoling station.

Figure 3 .
Figure 3. Temporal variations of hourly flow for different flood events.

Figure 4
Figure4presented the temporal variations of the percentage of surface runoff component in the upper Huaihe River basin above the Dapoling station.The percentage of surface runoff component in 2000s was generally higher than that in 1990s in earlier time of the flood process, while lower than that in 1990s in later time of the flood event.What's more, the ratio of the surface runoff in 1990s was significantly higher than that in 2000s during the peak stage of the surface runoff.Besides, it can be found that the process of the ratio of surface runoff was similar to that of the precipitation with several hours ahead of the surface runoff.The analysis above demonstrated that the land-use changing from paddy land to dryland will lead to higher surface runoff percentage of the

Figure 4 .
Figure 4. Temporal variations of the percentage of surface runoff component.

Table 1 .
Calibrated parameters for the daily Xinanjiang model in Dapoling sub-basin.

Table 2 .
Statistics of daily calibrated and validated results for Dapoling sub-basin in 1990s.
Notes: I and II for calibration period and validation period respectively.P for precipitation, R o for observed runoff, R c for calculated runoff, AE for abs error, RE for relative error, DC for determined coefficient, AE Peak for absolute error of flood peak, RE Peak for relative error of flood peak, Lag peak for flood peak lag time.

Table 3 .
Statistics of daily calibrated and validated results for Dapoling sub-basin in 2000s.

Table 4 .
Calibrated parameters for the hourly Xinanjiang model in Dapoling sub-basin.

Table 5 .
Statistics of hourly calibrated and validated results for Dapoling sub-basin in 1990s.

Table 6 .
Statistics of hourly calibrated and validated results for Dapoling sub-basin in 2000s.