next up previous
Next: 1.3 Dataset investigations Up: 1 Introduction Previous: 1.1 Dataset background


1.2 Dataset particulars

The dataset contains data from around Australia, excluding the westernmost edge of Western Australia, the northernmost tips of the Northern Territory and Queensland, and the southern half of Tasmania. The dimensions are 82 in the \( x \) direction (West-East), 60 in the \( y \) direction (South-North) and 16 in the \( z \) direction (vertical slice planes). The dataset is spatially irregular, that is, the data points are at arbitrary positions in 3D space1, rather than at regular intervals as is the case with regular and rectilinear data2. In addition, a separate file gives topographic information, that is, the surface's elevation above sea-level at each \( (x,y) \) point in the dataset.

The dataset covers the 47 hour period between 00:00 UTC3 29 August, 1996 (10:00 AEST4 29 August, 1996) and 23:00 UTC 30 August, 1996 (09:00 AEST 31 August, 1996), at 1 hour intervals. Thus the dataset can be thought of as being 4 dimensional, 3 in space and 1 in time.

At each point in the 4D dataset, three variables are available, being the wind velocity (a 3-vector in \( x \), \( y \) and \( z \) in km/hr), the temperature (in degrees Kelvin) and the relative humidity (that is, the humidity relative to the dew-point absolute humidity).

The dataset suffers from a peculiarity, however, and that is a distinct lack of data for the \( z \)-component of the wind velocity. Whether this is due to the original wind velocity measurements lacking \( z \)-components or the model's possible inability to deal with \( z \)-components of wind velocity is unknown. For the purposes of the visualisations in this project, this meant that the winds appear to be ``trapped'' in horizontal planes. This is unfortunate, but doesn't detract a great deal from the visualisations themselves.


next up previous
Next: 1.3 Dataset investigations Up: 1 Introduction Previous: 1.1 Dataset background
Kevin Pulo
2000-08-22