Spatial variation and scale variation in soil and water loss in heterogeneous landscape: A review
QIU Yang~1, FU Bo-Jie~2(1. Department of Resources and Environmental Science, Beijing Normal University, Beijing 100875, China; 2. Department of Systems Ecology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China)
This paper reviewed the studies on the theory and methods for the spatial variation and scale variation in landscape pattern and soil and water loss, and discussed the advance in scale and scaling of soil and water loss. Actually, the continuum between simple scaling and multiscaling is more common in nature. Whether both scaling-up and scaling-down are possible or not is related to the scale variation property. In general, it is probable for scaling up or scaling down with simple scaling, while it is not with multiscaling.For the soil and water loss, a complicated process with spatiotemporal variation, it shows a significant multiscaling characteristic. However, traditional studies on soil and water loss were conducted at plot scale or catchment scale solely,so these studies could not explore the scaling up or scaling down and the processes analysis on soil and water loss. In recent decades, only a little attention was paid on the comparing analysis of soil and water loss between plots with different size and the first-order catchment. However, it is scare though necessary to study the continuous scaling characteristics of soil and water loss and their relation to the environmental factors. In China, much attention was given to the study on soil erosion at plot scale and the simulation of soil and water loss at catchment scale and their relation to the environmental factors, while little was done with scaling variation in landscape pattern or soil and water loss.There are three types of methods for study on spatial variation and scaling variation: statistical model and empirical modeling, physically based model and processes modeling, and integrative model and synthetically modeling. No model can ever be expected to out-perform all other models on all reasonable data sets examined. Thus we insist that the best combination of several methods is different for the different purpose, objects, space and time studied. For the soil erosion models, the physically based model with spatial distribution can give more satisfactory results for the simulation of cross-scaling relationships in soil erosion than the empirical model can do. It is urgent to study the spatial variation of key parameters of model (i.e. rainfall, topography, soil, plant and land use) and to explore the scaling property of soil and water loss. In addition, it is also necessary to study the influencing of resolution and extent of model upon the output.Obviously, recent studies provide much information in theory and methods to study the scaling in landscape pattern and loss of soil and water and their relation to the influencing factors. In the future, we should pay much attention to explore both the spatial variation and the scaling variation in the loss of soil and water and related environmental factors based on "scale-pattern-processes", and to improve the soil erosion models based on these studies on the spatial and scale variation. All these studies will let us to know much about both the spatial variation of soil and water loss and its influencing factors and the scaling variation of soil and water loss and its mechanisms, and therefore will be helpful to determine the appropriate scale for study and management on soil and water loss.