Print Email Facebook Twitter Witnessing wetness Title Witnessing wetness: Spatial and temporal variation of soil water content in the Savutalele Catchment, Fiji Author Eeman, Sara (Delft University of Technology, Faculty of Civil Engineering and Geosciences) Project Molengraaff Fonds Date 2004-04 Abstract Soil conservation is an important issue in large parts of the world. Controlling or preventing erosion is one of the most important ways to achieve this. In designing erosion prevention measures for a specific area, modelling can be very useful. One such model is LISEM, which has been used in a number of projects worldwide. An area in which extreme erosion occurs is the Pacific, combining steep relief, shallow soils and heavy rainfall. The CROPPRO project is set up in the Pacific region, aiming to hand calibrate and apply the LISEM model. This model can then be used to design measurements which can decrease the stress that increasing population and economic growth are putting on this constrained island environment. In modelling erosion, a few parameters are of governing importance. Apart from infiltration and hydraulic conductivity, initial soil moisture content is such a parameter. This is defined as the soil moisture content just before the start of a rainfall event. This research aims to improve the possibility to predict this initial soil moisture content, which is currently done by monitoring the soil moisture content every few weeks and ‘educated guessing’ based thereon. Better prediction can be reached by relating the soil moisture content to other parameters such as precipitation, soil properties, geography and vegetation. Depending on the information and time available, there are two options to obtain such a relation; namely statistics or physical modelling. Statistical modelling is less elaborate but is only possible when soil moisture is measured continuously over at least one season: an ARX model can than relate other known parameters to the soil moisture content. It was found that the predictive value of such a model is reasonable (40% explained variance), especially for situations which are not extremely dry or wet. Physical modelling (in this research done with HYDRUS 1-D and 2-D versions) can also predict the soil moisture content quite well. Of course physical information, especially soil properties, are necessary for using this method. Boundary conditions are essential to make the model run properly. Initial conditions will improve the accuracy of the absolute values, but not so much the trend caused by the different parameters (weather, soil properties and geometry), given that sufficient initialisation time is provided. A time-series of soil moisture content for a whole growing season was available. This series was used in this research, to proof the use of the numerical model. Availability of such a series of measurements is useful, but not in general necessary. A few control points (in time and space) should be sufficient, given that physical information is sufficiently accurate. To reference this document use: http://resolver.tudelft.nl/uuid:969d6ea6-e751-43f3-bd65-e7cb500e97b9 Publisher Delft University of Technology Part of collection Geoscience Reports Document type report Rights (c) Sara Eeman Files PDF Eeman (2003).pdf 4.67 MB Close viewer /islandora/object/uuid:969d6ea6-e751-43f3-bd65-e7cb500e97b9/datastream/OBJ/view