Flood events play positive roles in river geometry formulation, water resource utilization, pollutant migration and transformation, nutrient exchange between floodplain and river channel, integrity of aquatic ecosystem and so on (
Journal of Geographical Sciences, Volume. 30, Issue 12, 2053(2020)
Investigation on flood event variations at space and time scales in the Huaihe River Basin of China using flood behavior classification
Flood is one of the severest natural disasters in the world and has caused enormous causalities and property losses. Previous studies usually focus on flood magnitude and occurrence time at event scale, which are insufficient to contain entire behavior characteristics of flood events. In our study, nine behavior metrics in five categories (e.g., magnitude, duration, timing, rates of changes and variability) are adopted to fully describe a flood event. Regional and interannual variations of representative flood classes are investigated based on behavior similarity classification of numerous events. Contributions of geography, land use, hydrometeorology and human regulation on these variations are explored by rank analysis method. Results show that: five representative classes are identified, namely, conventional events (Class 1, 61.7% of the total), low discharge events with multiple peaks (Class 2, 5.3%), low discharge events with low rates of changes (Class 3, 18.1%), low discharge events with high rates of changes (Class 4, 10.8%) and high discharge events with long durations (Class 5, 4.1%). Classes 1 and 3 are the major flood events and distributed across the whole region. Class 4 is mainly distributed in river sources, while Classes 2 and 5 are in the middle and down streams. Moreover, the flood class is most diverse in normal precipitation years (2006, 2008-2010 and 2015), followed by wet years (2007, 2013-2014), and dry years (2011 and 2012). All the impact factor categories explain 34.0%-84.1% of individual flood class variations. The hydrometeorological category (7.2%-56.9%) is the most important, followed by geographical (1.0%-6.3%), regulation (1.7%-5.1%) and land use (0.9%-2.2%) categories. This study could provide new insights into flood event variations in a comprehensive manner, and provide decision-making basis for flood control and resource utilization at basin scale.
1 Introduction
Flood events play positive roles in river geometry formulation, water resource utilization, pollutant migration and transformation, nutrient exchange between floodplain and river channel, integrity of aquatic ecosystem and so on (
Flood event variation shows complicated spatio-temporal heterogeneity due to mutual impacts of rainfall patterns, catchment characteristics and anthropogenic activities (
Many methods have been adopted for river or catchment classification (
The Huaihe River Basin is the most densely inhabited basin and the main cropping area of China (
2 Materials and methods
2.1 Study area
As one of the major basins in China, the Huaihe River Basin (111°55°-121°25°E, 30°55°- 36°36°N) is located between the Yangtze River and Yellow River basins. The drainage area is approximately 2.7×105 km2 which is divided into the Huaihe River Catchment (1.8×105 km2) and Yishusi River Catchment (9.0×104 km2) by the paleo-channel of the Yellow River (Old Yellow River) (
Figure 1.
Due to strong precipitation seasonality, rapid discharge from the upper mountainous areas and backwater effect from downstream lakes, the Huaihe River Basin is one of the most frequently and severely threatened basins by flood disasters. Over 350 major floods have happened in the last 2000 years. Particularly, 29 counties were severely stricken in the upper and middle catchment of the Huaihe River during the catastrophic flood of August, 1975 (“75.8” catastrophic flood). The death roll was 26 thousand and the direct economic loss was billions of USD. Since the beginning of the 21st century, major floods happened continuously across the whole basin, e.g., 2001, 2003, 2007, 2009, 2011 and 2020. In order to regulate floods, 9643 reservoirs and 23,767 sluices had been constructed with the storage capacities accounting for over half of the annual runoff magnitude in the Huaihe River Basin by the end of 2018. Therefore, the flow regimes of the Huaihe mainstream and most of the tributaries were regulated remarkably (
2.2 Data sources
The hourly flow observations of 342 flood events at 39 hydrological stations in the three main tributaries (i.e., Shaying River, Hongru River and Southern mountainous rivers) and the Huaihe mainstream are collected from the hydrological yearbooks of Henan and Anhui provinces from 2006 to 2015. There are 89 flood events at 11 stations in the Shaying River, 76 events at 11 stations in the Hongru River, 105 events at nine stations in the Southern mountainous rivers, and 72 events at eight stations in the Huaihe mainstream. The general geographical, major land use and climatic characteristics of controlled catchments are presented in
The general characteristics of controlled catchments, and the selected flood events
The general characteristics of controlled catchments, and the selected flood events
|
2.3 Flood behavior metrics and potential impact factors
Nine flood behavior metrics in five categories of magnitude, duration, timing, rate of changes and variability are adopted to describe the overall characteristics of flood events. The detailed definitions and calculations are given in
Flood behavior metrics used for flood event descriptions
Flood behavior metrics used for flood event descriptions
|
Our study investigates the potential impacts of geographical, land use, hydrometeorological and regulation categories on the flood event classes (Zhang
Potential impact factor categories used to analyze the space and time variations of flood events
Potential impact factor categories used to analyze the space and time variations of flood events
|
The data sources of Geographic Information System (GIS), gauged hydrometeorological observations, and dam regulations are collected for the impact assessments. The GIS data sources include the Digital Elevation Model (DEM) (100 m×100 m resolution), and land use data in 2005, 2010 and 2015 (1000 m×1000 m resolution), all of which are from the Data Center for Resource and Environmental Sciences, Chinese Academy of Sciences (
2.4 Methods
Both the principal component analysis and the cluster analysis are adopted to identify the main flood event classes and their variations. The principal component analysis has advantages of merging multiple correlated metrics into independent composite components without losing the metric information (
2.4.1 Principal component analysis
The nine flood behavior metrics of all the 342 events could be formed as a metric matrix
where
where
Due to the high dimensionality and correlation among different flood behavior metrics, the principal component analysis is adopted to simplify the matrix
where
where
where
If the cumulative variance reaches a certain threshold, the first
2.4.2 Cluster analysis
The hierarchical clustering analysis is adopted to cluster 342 flood events into some representative flood event classes based on the PCA similarity among different flood events. Euclidean distance is used to calculate the PCA similarity and the equation is as follows:
where
2.4.3 Impact factor identification of individual flood event classes
The rank analysis method is adopted to detect the potential relationships between impact factor categories and flood event classes. The typical methods are the Detrended Correspondence Analysis (DCA), Redundancy Analysis (RDA) and Canonical Correlation Analysis (CCA) (
where
The event scores are then used to calculate a new set of impact factor scores following the same procedure. Thus the equation is given as follows:
The impact factor scores are centered and standardized with their mean and variance being zero and one, respectively.
This procedure for calculating impact factor and event scores (i.e., Equations 9-11) is repeated until the scores stabilize. The
3 Results and discussion
3.1 Flood event classification
Five independent PCAs are selected, which explain 83.18% of the entire variance of flood behavior metrics. The first PCA including the characteristic metrics (
The flood events are classified into several clusters (i.e., two to six) based on the similarity of the PCAs (Figures 2 and 3). As the cluster number increases, the
Figure 2.
Figure 3.
3.2 Flood behavior characteristics of individual classes
There are 210 flood events in Class 1, accounting for 61.4% of the total number (
Figure 4.
Figure 5.
There are 19 flood events in Class 2, accounting for 5.6% of the total number. For the flood magnitude,
There are 62 flood events in Class 3, accounting for 18.1% of the total number. For the flood magnitude,
There are 37 flood events in Class 4, accounting for 10.8% of the total number. For the flood magnitude,
There are 14 flood events in Class 5, accounting for 4.1% of the total number. For the flood magnitude,
3.3 Regional variations of individual classes
All the flood event classes exist in the Hongru River, Shaying River and Southern mountainous rivers, except the Huaihe mainstream (
Figure 6.
In the Shaying River, 78.9% of the flood events belongs to Class 1, followed by Class 3 (18.4%), Class 4 (11.8%), Class 2 (6.5%) and Class 5 (1.3%). All the flood event classes (Classes 1-5) exist at XiaGS station, while only one flood event class (Class 1) exists at ZhongM station. In the Hongru River, 60.5% of the flood events also belong to Class 1, followed by Class 4 (21.1%), Class 3 (11.8%), Class 2 (5.3%) and Class 5 (1.3%). The flood event class is most diverse at GuiL station (Classes 1, 3, 5), LiX station (Classes 1, 2, 4), MiaoW and WuGY stations (Classes 1, 3, 4), while only one flood event class is at SuiP station (Class 1) and ZhuMD station (Class 4). In the Southern mountainous rivers, 78.9% of the flood events belong to Class 1, followed by Class 3 (36.8%), Class 4 (15.8%), Class 2 (5.3%) and Class 5 (1.3%). The flood event class is most diverse at JiangJJ station (Classes 1-3, 5) and XinX station (Classes 1-4), while only one flood event class is at HuangNZ station (Class 1). In the Huaihe mainstream, 57.9% of the flood events belong to Class 1, followed by Classes 3 and 5 (14.5%), and Class 2 (7.9%). No flood events exist in Class 4. The flood event class is most diverse at Xix station (four classes, i.e., Classes 1-4), while only one flood event class is at Lutz and Bengb stations (Class 5).
3.4 Interannual variations of individual classes
For all the flood events, the class is most diverse in the normal precipitation years (2006, 2008-2010 and 2015) including all the five classes (
Figure 7.
The flood event classes in the Shaying River shift from Class 1 to Classes 3 and 4 with the decrease of annual precipitation. Class 1 distributes in the whole period, and is dominant particularly in the dry years (2011, 2012 and 2014). There are only five events in Class 2, mainly in the normal precipitation years (2006, 2008 and 2010); 14 events in Class 3, mainly in the normal precipitation and wet years (2007, 2008, 2010, 2013 and 2015); nine events in Class 4, mainly in the wet years (2009, 2013 and 2015).
The flood event classes in the Hongru River and Southern mountainous rivers are relatively diverse, and there are two or three classes in all the years. In the Hongru River, besides the dominant Class 1, Class 4 exists in most years (except 2013) with the ratios from 11.1% to 40.0%. Only four events exist in Class 2, mainly in the normal precipitation years (2006, 2010 and 2015); nine events in Class 3, mainly in the normal precipitation years (2008, 2009, 2012, 2013 and 2015), and only one event in Class 5 (at XiaGS station in 2007). In the South mountainous rivers, Class 3 exists in most years except in 2011 and 2012 with the ratios from 16.7% to 60.0%. Only four events exist in Class 2, in the dry years (2009, 2011 and 2013); 12 events in Class 4, mainly in the normal precipitation and wet years (2008, 2010-2013); only one event in Class 5 (at JiangJJ station in 2015).
In the Huaihe mainstream, the flood event classes vary most obviously in the whole period. There are four classes (except Class 4) in the wet year (2008), but only one class (Class 1) in the normal precipitation and dry years (2012 and 2013). Class 1 is dominant in 2009, 2011 and 2012 with the ratios being over 85%. Class 5 is mainly distributed in 2006-2010 and 2015 with the ratios from 11.1% to 50.0%, and Classes 2 and 3 are mainly distributed in 2013-2015 with the ratios from 6.7% to 50.0% for Class 2, and from 11.1% to 40.0% for Class 3, respectively.
3.5 Potential impacts on the flood event classes
According to the DCA detection, all the lengths of first axises are less than 3.0. Thus, the RDA method is selected. By the RDA analysis, 80% of the impact factors (32/40) are detected to be significantly correlated with the flood event variations (
Figure 8.
Due to the flood event classification, the impact factor categories explain more variations for the individual flood event classes (
Figure 9.
Therefore, the regional and interannual variations of both the individual flood events and classes are mainly impacted by the hydrometeorological category, particularly the total precipitation amount during the events which directly determines all the flood event characteristics (
Furthermore, 27 of all the 39 stations (69.2%) are at the downstreams of reservoirs or sluices (
4 Conclusions
The flood event variations are investigated at both regional and interannual scales based on the flood behavior metrics using the classification approach, and the potential impact categories are further explored to explain the flood event variations using the rank analysis. There are 342 flood events at 39 stations in the upper and middle reaches of the Huaihe River Basin selected for the study. Results show that:
(1) All the flood events are clustered into five flood event classes, i.e., the low discharge, medium variable, single peak flood events with medium rates of changes in the flood season (Class 1), the low discharge, medium variable, multiple peak floods with medium duration in the flood season (Class 2), the low discharge, medium variable, single peak floods with low rates of changes in the pre-flood season (Class 3), the low discharge, extreme variable, single peak floods with high rates of changes in the flood season (Class 4), and the high discharge, stable, single peak floods with long duration in the flood season (Class 5). There are 210, 19, 62, 37 and 14 events in Classes 1-5, respectively, accounting for 61.4%, 5.6%, 18.1%, 10.8% and 4.1% of the total number of events.
(2) The flood event class is most diverse at the source and upstream stations, and becomes single at the downstream stations. Classes 1 and 3 are the major flood events in all the attributes across the basin. Class 2 mainly distributes in the Hongru River and the downstreams of Southern mountainous rivers. Class 4 mainly distributes in the source rivers, and Class 5 mainly distributes in the middle and downstream of Huaihe mainstream. Furthermore, most of the flood event classes exist in the normal precipitation years, followed by the wet years. The flood event class becomes homogeneously distributed in the dry years.
(3) The impacts of geographical, land use, hydrometeorological and regulation categories probably result in the regional and interannual variations for both all the flood events and the individual classes. The contributions of all the impact factor categories range from 34.0% to 84.1%, in which the hydrometeorological category is the most important (7.2%-56.9%). The impacts of geographical, regulation and land use categories should not be ignored, which explain 1.0%-6.3%, 1.7%-5.1% and 0.9%-4.0% of the total variations of flood event classes. Moreover, the combined influences of all the impact factor categories can explain 14.7% of the total flood event variations, 20.4% of the total variations in Class 1, 12.2% in Class 3 and 27.8% in Class 4, respectively.
The results of flood event classes could be beneficial to investigate the flood event variations in a comprehensive manner. For example, the flood event variations in different rivers could be deduced in advance from the identified flood event classes according to the geographical, land use, hydrometeorological and regulation conditions. The variations would be very informative to design plans for flood control and disaster mitigation, water resource utilization at basin scale. However, the potential impacts on the flood event variations can be further explored because 15.9%-92.8% of the total variations in the individual classes are still not explained. Hydrological modelling approach could be adopted to explore the potential impact mechanisms, and quantify their contributions in future studies.
[1] AliG, TetzlaffD, SoulsbyC et al. A comparison of similarity indices for catchment classification using across-regional dataset. Advance in Water Resources, 40, 11-22(2012).
[2] AnctilF. Wavelet analysis of the interannual variability in Southern Quebec Streamflow. Journal of Climate, 17, 163-173(2010).
[3] Boughton WC. A hydrograph-based model for estimating the water yield of ungauged catchments, paper presented at Hydrology and Water Resources Symposium. Inst. of Eng. Aust., Newcastle, N.S.W.(1993).
[4] Brunner MI, ViviroliD, FurrerR, SeibertJ et al. Identification of flood reactivity regions via the functional clustering of hydrographs. Water Resources Research, 54, 1852-1867(2018).
[5] Burn DH, Boorman DB. Catchment classification applied to the estimation of hydrological parameters at ungauged catchments (technical report 118). Wallingford, Oxfordshire, UK: Institute of Hydrology(1992).
[6] Chapman TA. Comparison of algorithms for stream flow recession and baseflow separation. Hydrological Processes, 13, 701-714(1999).
[7] HanD, Cluckie ID, KarbassiounD et al. River flow modelling using fuzzy decision trees. Water Resources Management, 16, 431-445(2002).
[8] Hargreaves GH, Samani ZA. Estimating potential evapotranspiration. Journal of the Irrigation & Drainage Division, 108, 225-230(1982).
[9] HirabayashiY, MahendranR, KoiralaS et al. Global flood risk under climate change. Nature Climate Change, 3, 816-821(2013).
[10] Kennard MJ, Pusey BJ, Olden JD et al. Classification of natural flow regimes in Australia to support environmental flow management. Freshwater Biology, 55, 171-193(2010).
[11] Kumar A RS, Goyal MK, Ojha C SP et al. Application of ANN, fuzzy logic and decision tree algorithms for the development of reservoir operating rules. Water Resources Management, 27, 911-925(2013).
[12] Li ZL, Xu ZX, Li JY et al. Shift trend and step changes for runoff time series in the Shiyang River basin, Northwest China. Hydrological Processes, 22, 4639-4646(2008).
[13] Liao SH, Wen CH. Artificial neural networks classification and clustering of methodologies and applications: Literature analysis from 1995 to 2005. Expert Systems with Application, 32, 1-11(2007).
[14] LyneV, HollickM. Stochastic time-variable rainfall-runoff modelling, I. E. Aust. Natl. Conf. Publ. 79/10, pp. 89-93, Inst. of Eng. Aust., Canberra, ACT(1979).
[15] MaF, YeA, GongW et al. An estimate of human and natural contributions to flood changes of the Huai River. Global & Planetary Change, 119, 39-50(2014).
[16] Ma GB, Jiang WG, LiJ et al. Assessment and validation of short-term flood hazards in China. Geographical Research, 31, 34-44(2012).
[17] McDonnell JJ, WoodsR. On the need for catchment classification. Journal of Hydrology, 299, 2-3(2004).
[18] Mcmanamay RA, Oigbokie CO, Kao SC et al. Classification of US hydropower dams by their modes of operation. River Research and Applications, 32, 1450-1468(2016).
[19] MerzB, Dung NV, ApelH et al. Spatial coherence of flood-rich and flood-poor periods across Germany. Journal of Hydrology, 559, 813-826(2018).
[22] NiedM, Schr?terK, LüdtkeS et al. What are the hydro-meteorological controls on flood characteristics?. Journal of Hydrology, 545, 310-326(2017).
[23] NingY, QianM, WangY. Huaihe River Basin Hydrology Handbook(2003).
[24] Parajka JK, OhnovaS, BalintG et al. Seasonal characteristics of flood regimes across the Alpine-Carpathian range. Journal of Hydrology, 394, 78-89(2010).
[25] Poff NL, Allan JD, Bain MB et al. The natural flow regime: A paradigm for river conservation and restoration. Bioscience, 47, 769-784(1997).
[26] Requena AI, ChebanaF, Ouarda TB. Heterogeneity measures in hydrological frequency analysis: Review and new developments. Hydrology and Earth System Sciences, 21, 1651-1668(2017).
[27] Richter BD, Baumgartner JV, PowellJ et al. A method for assessing hydrologic alteration within ecosystems. Conservation Biology, 10, 1163-1174(1996).
[28] Robinson JS, SivapalanM. Temporal scales and hydrological regimes: Implications for flood frequency scaling. Water Resources Research, 33, 2981-2999(1997).
[29] RoggerM, KohlB, PirklH et al. Runoff models and flood frequency statistics for design flood estimation in Austria - Do they tell a consistent story?. Journal of Hydrology, 456/457, 30-43(2012).
[30] Rosgen DL. A classification of natural rivers. Catena, 22, 169-199(1994).
[31] SauquetE, CatalogneC. Comparison of catchment grouping methods for flow duration curve estimation at ungauged sites in France. Hydrology and Earth System Sciences, 15, 2421-2435(2011).
[32] ShaoQ, CampbellN. Applications: Modelling trends in groundwater levels by segmented regression with constraints. Australian & New Zealand Journal of Statistics, 44, 129-141(2002).
[33] She DX, XiaJ, Shao QX et al. Advanced investigation on the change in the streamflow into the water source of the middle route of China’s water diversion project. Journal of Geophysical Research: Atmospheres, 122, 6950-6951(2017).
[34] ShiP, Ma XX, Hou YB et al. Effects of land-use and climate change on hydrological processes in the upstream of Huai River, China. Water Resources Management, 27, 1263-1278(2013).
[35] Sikorska AE, ViviroliD, SeibertJ. Flood-type classification in mountainous catchments using crisp and fuzzy decision trees. Water Resources Research, 51, 7959-7976(2015).
[36] Singh VP. Entropy, Theory in Hydrologic Science and Engineering. New York, USA: McGraw-Hill Education(2015).
[37] Snelder TH, Biggs BJ, Woods RA. Improved ecohydrological classification of rivers. River Research and Applications, 21, 609-628(2005).
[38] SunP, Wen QZ, ZhangQ et al. Nonstationarity evaluation of flood frequency and flood risk in the Huai River basin, China. Journal of Hydrology, 567, 393-404(2018).
[39] WangW, WeiJ, ShaoQ et al. Spatial and temporal variations in hydro-climatic variables and runoff in response to climate change in the Luanhe River basin, China. Stochastic Environmental Research and Risk Assessment, 29, 1117-1133(2015).
[40] XiaJ, DuH, ZengS et al. Temporal and spatial variations and statistical models of extreme runoff in Huaihe River Basin during 1956-2010. Journal of Geographical Sciences, 22, 1045-1060(2012).
[41] XiaJ, She DX, Zhang YY et al. Spatial-temporal trend and statistical distribution of extreme precipitation events in Huaihe River Basin during 1960-2009. Journal of Geographical Sciences, 22, 195-208(2012).
[42] YangP, XiaJ, Zhan CS et al. Discrete wavelet transform-based investigation into the variability of standardized precipitation index in Northwest China during 1960-2014. Theoretical and Applied Climatology, 132, 167-180(2018).
[43] Yang WZ, Zhang BW, Fu RM. A summary of flood disaster in 2016. China Flood & Drought Management, 27, 26-29(2017).
[44] Zbigniew WK, ShinjiroK, Sonia IS et al. Flood risk and climate change: Global and regional perspectives. Hydrological Sciences Journal, 59, 1-28(2013).
[45] ZhangQ, Chen YQ, Chen XH et al. Copula-based analysis of hydrological extremes and implications of hydrological behaviors in the Pearl River basin, China. Journal of Hydrologic Engineering, 16, 598-607(2011).
[46] Zhang SJ, Jiang YH, Yang HJ et al. An hydrology-process based method for antecedent effect rainfall determination in debris flow forecasting. Advances in Water Science, 26, 34-43(2015).
[47] Zhang YY, Chen QT. Characteristics of main flood event types and their temporal-spatial variations in the upper and middle reaches of the Huai River Basin. Progress in Geography, 39, 627-635(2020).
[48] Zhang YY, Shao QX, XiaJ. Changes of flow regimes and precipitation in Huai River Basin in the last half century. Hydrological Processes, 25, 246-257(2011).
[49] Zhang YY, Shao QX, Zhao T TG. Comprehensive assessment of dam impacts on flow regimes with consideration of interannual variations. Journal of Hydrology, 552, 447-459(2017).
[50] Zhang YY, XiaJ, Bunn SE et al. Classification of flow regimes for environmental flow assessment in regulated rivers: The Huai River Basin, China. River Research and Applications, 28, 989-1005(2012).
[53] Zheng HX, ZhangL, Liu CM et al. Changes in stream flow regime in headwater catchments of the Yellow River basin since the 1950s. Hydrological Processes, 21, 886-893(2007).
Get Citation
Copy Citation Text
Yongyong ZHANG, Qiutan CHEN, Jun XIA. Investigation on flood event variations at space and time scales in the Huaihe River Basin of China using flood behavior classification[J]. Journal of Geographical Sciences, 2020, 30(12): 2053
Category: Research Articles
Received: Apr. 29, 2020
Accepted: Aug. 2, 2020
Published Online: May. 7, 2021
The Author Email: ZHANG Yongyong (zhangyy003@igsnrr.ac.cn)