Advanced search
1 file | 7.18 MB Add to list

Spatial prediction of water table dynamics in Flanders

(2012)
Author
Promoter
(UGent)
Organization
Abstract
Shallow water tables are one of the most important land characteristics. They determine the potential land use: the depth is essential in the evaluation of suitability for agriculture and in nature development projects, it is also crucial to identify possible problems with other land uses. Although it is easy to derive the water table on a specific location at one point in time, the depth is highly variable in time (due to its dependence on precipitation surplus) and in space (as is easily observed when strolling through the countryside). The common method for estimating the water table depth in Flanders was by using estimates based on the existing natural drainage class map of Flanders. This natural drainage class map is based on data collected during the national soil survey. It was derived from the depth of gley mottles and the reduction horizon and the position in the landscape and is often used to predict the depth and variability of the water table. Rather than duplicating this methodology, a different method was applied based on actual measurements of the water table. The evaluation was done in three case studies in different part of Flanders: The Dijle Valley, Kluizen and Damme. These encompass some of the major regions where the water table is of great importance: sandy Flanders, river valleys in central Flanders, the coastal region and areas with temporary water tables. The method consisted of three steps: (1) Locations with sufficient water table measurements were used as reference time series. Their data were combined with daily precipitation surplus data using time series analysis to derive climate representative water table statistics. The most important statistics derived were the mean highest water table (MHW) defined as the mean value of the three highest water table levels measured biweekly (or semi-monthly) and the mean lowest water table (MLW) by using the three lowest water table levels. A continuous time series model (PIRFICT) was chosen and implemented, using the precipitation and potential evapotranspiration as driver variables. (2) The data derived from these reference time series were combined with data from shorter time series. In the case of Dijle, additional series were already existing. In contrast for Kluizen and Damme, 500 locations were visited twice, once during winter and once during summer, when the water table is supposed to be on its highest/lowest respectively. This was used to estimate the MHW and MLW in these locations. (3) Finally these locations were used to create areawide predictions. Two major methods were used: remapping and relabeling methods. Relabeling methods use the old soil drainage class map and apply new attributes to the existing polygons without changing their borders. Remapping uses ancillary maps as a basis: from the digital elevation model relevant properties like slope and wetness index were derived and combined with other properties such as the horizontal and vertical distance to channels. To map MHW, MLW and drainage class, point measurements and these ancillary maps were used in a regression kriging approach. All methods proved to be superior to estimates using only the old soil drainage class maps. Nevertheless, the timing of measurements was important, and a large number of summer measurements taken in Damme during an unusually dry summmer could not be used in the further analysis, leading to much higher uncertainties associated with these maps. Apart from testing the updating methodologies in different study areas, a second subject was an evaluation of the predictive quality of the current drainage class map in different parts of Flanders. This predictive quality was evaluated by combining data from networks with good spatial but poor temporal coverage with data from networks with better temporal but poor spatial coverage. Point predictions for MHW and MLW were derived by applying time series modelling (in networks with a good temporal coverage) and by using total least squares regression (to expand to the other sites). The resulting MHW and MLW point data set was used to evaluate the currency of the existing map and to identify regional differences. The quality of the current map is moderate, and large differences occur between regions. Especially the Campine region shows large and systematic differences, whereas the southeastern hills and chalk-loam region are relatively accurate. If more weight is given to errors in the wetter drainage classes, about 50% of the area of Flanders would benefit from remapping.
Keywords
Flanders, hydropedology, Geostatistics, Water Table, Phreatic groundwater, drainage class, Digital Soil Mapping, Belgium

Downloads

  • (...).pdf
    • full text
    • |
    • UGent only
    • |
    • PDF
    • |
    • 7.18 MB

Citation

Please use this url to cite or link to this publication:

MLA
Van de Wauw, Johan. Spatial Prediction of Water Table Dynamics in Flanders. Ghent University. Faculty of Sciences, 2012.
APA
Van de Wauw, J. (2012). Spatial prediction of water table dynamics in Flanders. Ghent University. Faculty of Sciences, Ghent, Belgium.
Chicago author-date
Van de Wauw, Johan. 2012. “Spatial Prediction of Water Table Dynamics in Flanders.” Ghent, Belgium: Ghent University. Faculty of Sciences.
Chicago author-date (all authors)
Van de Wauw, Johan. 2012. “Spatial Prediction of Water Table Dynamics in Flanders.” Ghent, Belgium: Ghent University. Faculty of Sciences.
Vancouver
1.
Van de Wauw J. Spatial prediction of water table dynamics in Flanders. [Ghent, Belgium]: Ghent University. Faculty of Sciences; 2012.
IEEE
[1]
J. Van de Wauw, “Spatial prediction of water table dynamics in Flanders,” Ghent University. Faculty of Sciences, Ghent, Belgium, 2012.
@phdthesis{3040427,
  abstract     = {{Shallow water tables are one of the most important land characteristics. They determine the potential land use: the depth is essential in the evaluation of suitability for agriculture and in nature development projects, it is also crucial to identify possible problems with other land uses. Although it is easy to derive the water table on a specific location at one point in time, the depth is highly variable in time (due to its dependence on precipitation surplus) and in space (as is easily observed when strolling through the countryside). The common method for estimating the water table depth in Flanders was by using estimates based on the existing natural drainage class map of Flanders. This natural drainage class map is based on data collected during the national soil survey. It was derived from the depth of gley mottles and the reduction horizon and the position in the landscape and is often used to predict the depth and variability of the water table.
Rather than duplicating this methodology, a different method was applied based on actual measurements of the water table.
The evaluation was done in three case studies in different part of Flanders: The Dijle Valley, Kluizen and Damme. These encompass some of the major regions where the water table is of great importance: sandy Flanders, river valleys in central Flanders, the coastal region and areas with temporary water tables.
The method consisted of three steps:
(1) Locations with sufficient water table measurements were used as reference time series. Their data were combined with daily precipitation surplus data  using time series analysis to derive climate representative water table statistics. The most important statistics derived were the mean highest water table (MHW)  defined as the mean value of the three highest water table levels measured biweekly (or semi-monthly) and the mean lowest water table (MLW) by using the three lowest water table levels. A continuous time series model (PIRFICT) was chosen and implemented, using the precipitation and potential evapotranspiration as driver variables.
(2) The data derived from these reference time series were combined with data from shorter time series. In the case of Dijle,  additional series were already existing. In contrast for Kluizen and Damme, 500 locations were visited twice, once during winter and once during summer, when the water table is supposed to be on its highest/lowest respectively. This was used to estimate the MHW and MLW in these locations.
(3) Finally these locations were used to create areawide predictions. Two major methods were used: remapping and relabeling methods. Relabeling methods use the old soil drainage class map and apply new attributes to the existing polygons without changing their borders. Remapping uses ancillary maps as a basis: from the digital elevation model relevant properties like slope and wetness index were derived and combined with other properties such as the horizontal and vertical distance to channels. To map MHW, MLW and drainage class, point measurements and these ancillary maps were used in a regression kriging approach.
All methods proved to be superior to estimates using only the old soil drainage class maps. Nevertheless, the timing of measurements was important, and a large number of summer measurements taken in Damme during an unusually dry summmer could not be used in the further analysis, leading to much higher uncertainties associated with these maps.
Apart from testing the updating methodologies in different study areas, a second subject was an evaluation of the predictive quality of the current drainage class map in different parts of Flanders. This predictive quality was evaluated by combining data from networks with good spatial but poor temporal coverage with data from networks with  better temporal but poor spatial coverage. Point predictions for MHW and MLW were derived by applying time series modelling (in networks with a good temporal coverage) and by using total least squares regression (to expand to the other sites). The resulting MHW and MLW point data set was used to evaluate the currency of the existing map and to identify regional differences. The quality of the current map is moderate, and large differences occur between regions. Especially the Campine region shows large and systematic differences, whereas the southeastern hills and chalk-loam region are relatively accurate. If more weight is given to errors in the wetter drainage classes, about 50% of the area of Flanders would benefit from remapping.}},
  author       = {{Van de Wauw, Johan}},
  isbn         = {{9789461970770}},
  keywords     = {{Flanders,hydropedology,Geostatistics,Water Table,Phreatic groundwater,drainage class,Digital Soil Mapping,Belgium}},
  language     = {{eng}},
  pages        = {{170}},
  publisher    = {{Ghent University. Faculty of Sciences}},
  school       = {{Ghent University}},
  title        = {{Spatial prediction of water table dynamics in Flanders}},
  year         = {{2012}},
}