1.3.3.1. Scaling issues
The scaling issues have been previously summarized by Zhang et al. (2013). Briefly, the scaling issues arise from the variation due to space, time or other dimensions in the real world. Following this concept, finite and discrete measurements are used to understand the infinite variables and the continuous system. Thereby, information about environmental features can be stored, recalled and analyzed. There are three types of scale: operation scale, measurement scale, and modeling scale. An operation scale is a scale which concerns the operation of a physical process in a natural environment. A measurement scale concerns the spatial resolution used to determine an object. There are two extents of a measurement scale for a data set: spatial (the space between samples) and temporal (the integration time).
Besides, a modeling scale relates to both the natural process and the applied models. A modeling scale can involve four different extents, including geographic scale (the research area), temporal scale (the time period of research), measurement scale of parameters (the resolution of input data) and the model scale (the temporal and spatial scale when a model
36
was established). The spatial scales can be local or plot scale (~1m), hillslope or research scale (~100m), catchment scale (~10km) and regional scale (~1000km), while the temporal scales can be event scale (~1 day), seasonal scale (~1 year) or long-term scale (~100 years).
There are many causes leading to scaling problem (Harvey, 2000; Heuvelink, 1998). The most common and fundamental cause is the existence of spatial heterogeneity and relevant process nonlinearities. Indeed, the spatial scale depends on both climatic input data and land-surface parameters, which are in turn derived from various measurements including temperature, precipitation, topography, land uses, soil physical and chemical properties and other hydrological properties. Thus, it is difficult to aggregate large-scale behavior from local processes. The second reason is that the predominant processes can differ in different scales.
The correlations derived at one scale might not be applicable to another scale. Therefore, more processes need to be considered by the scaling method. The third cause is the cross-scale interactions between small-cross-scale and large-cross-scale parameters, suggesting that scaling should be only applied over a limited range of scales and specific context. Another cause for scaling problems is the lack of information on process linkages in a dynamic environment.
The alterations of the scale measurement can significantly influence the variability of model parameters. The physical meaning of model parameters is associated with its corresponding processed. If a parameter is estimated at its process scale, its value will be able to approximate the reality. Otherwise, its value will be less realistic if the measurement scale is larger or smaller than the process scale.
1.3.3.2. The temporal variability
The temporal variability of hydrological phenomena can be caused by climatic and human factors. There are different types of temporal variability, including the diurnal, annual, inter-annual or irregular temporal variations. Among those, diurnal and seasonal characteristics are the important source of variations that affect directly the flood process.
The diurnal variability of hydrological phenomena can result in changes in river pattern, for example, due to snowmelt, evaporation or water management operations. The diurnal variability can influence the behavior of discharges in terms of timing, relative magnitude, and shape (Lundquist and Cayan, 2002). For example, in rivers where water is added diurnally, the discharges will be characterized by daily sharp runoff rise and gradual decline.
37
Meanwhile, in rivers where water is removed diurnally, the discharges will be characterized by daily gradual runoff rise and sharp decline.
The seasonal variability of hydrological phenomena can differ across different area. In particular, this type of variability also depends on climatic factors, for example, on seasonal snow accumulation and release, or seasonal rainfall changes. The inter-annual variability is also caused by climatic phenomena, such as the atmospheric-ocean phenomena like El Nino or Pacific Decadal Oscillation (Woods, 2006).
Another important type of spatial variability is caused by irregular phenomena, including storms and floods (Woods, 2006). These extreme weather phenomena can affect the streamflow directly. Especially, this type of temporal variability is now becoming more important nowadays, due to the more frequent extremes foreseen under the climate change conditions.
1.3.3.3. The spatial variability
The spatial variability of hydrological phenomena are the results of spatial patterns in climate, land use, topography, geology, and soil characteristics.
Climate factors are indeed an important cause of spatial variability. Climate conditions vary across different regions, which strongly affect hydrological processes. For example, rainfall accumulation tends to be higher in the high mountainous area than in the downhill area. Climate conditions can also vary within a specific region. For instance, rainfall accumulation can be greatly different between the upwind and downwind side of the mountains, due to differences in air moisture (Woods, 2006).
Geological and soil characteristics are characterized by complex spatial variability. This type of variability can affect hydrogeological characteristics such as porosity and permeability, which in turn affect the rate of infiltration and water movement throughout soil layers. The variation of topography within and across the region can also affect hydrological processes, as it can influence the streamflow direction and velocity. Last but not least, vegetation is also an important source of spatial variability, as it can affect water transport, evapotranspiration, soil water content. The presence of plants in different regions also reflect the climatic conditions.
38