Author(s): |
Jon Olav Skøien, European Commission, Joint Research Centre (IT)
G. Blöschl, Vienna University of Technology, Institute for Hydraulic and Water Resources Engineering (AT)
G. Laaha, BOKU Vienna, Institute of Applied Statistics and Computing, Vienna, Austria (AT)
E. J. Pebesma, Institute for Geoinformatics, University of Muenster (DE)
J. Parajka, Vienna University of Technology, Institute for Hydraulic and Water Resources Engineering (AT)
A. Viglione, Vienna University of Technology, Institute for Hydraulic and Water Resources Engineering (AT)
|
Abstract: |
Runoff related variables (runoff, runoff statistics, temperature, concentrations) are often modelled through conceptually or physically based models. The methods are usually data intensive, and require calibration. Geostatistical methods can be regarded as more data driven methods, with a more limited need for data. However, such methods have rarely been applied for hydrological modelling, as the methods were traditionally based on point observations or observations with a regular support.
There has recently been a development of geostatistical methods for non-point support, particularly within health modelling (Goovaerts, 2006) and hydrology (Gottschalk, et al., 2006; Skøien, et al., 2006). In this paper we will demonstrate an R package (rtop) that implements the methods presented by Skøien et al. [2006], and extended with suggestions from Gottschalk et al. [2006].
Taking advantage of the existing methods in R for manipulating spatially referenced objects (points, lines, polygons, grids), and the extensive possibilities for visualizing the results, rtop makes it considerably easier to apply geostatistical interpolation methods to observations with a non-point support, in comparison to former implementations of the method. Variogram fitting differs strongly from that in traditional geostatistics, but we will present methods for automatic and manual variogram fitting, and for visualizing the model fits. Another feature of the kriging method is that it can handle observation uncertainty, either as a result of measurement uncertainty, or when interpolating statistics, such as annual mean or floods with a certain return period.
Goovaerts, P. 2006. Geostatistical analysis of disease data: accounting for spatial support and population density in the isopleth mapping of cancer mortality risk using area-to-point Poisson kriging. International Journal of Health Geographics, 5, art. no. 52.
Gottschalk, L., I. Krasovskaia, E. Leblois, and E. Sauquet (2006), Mapping mean and variance of runoff in a river basin, Hydrology and Earth System Sciences, 10, 469-484.
Skøien, J. O., R. Merz, and G. Blöschl (2006), Top-kriging - geostatistics on stream networks, Hydrology and Earth System Sciences, 10, 277-287.
|
