• No results found

The GIS interface. Shown are different management units and the

APPLICATIONS IN THE AGRICULTURAL LANDSCAPE

7.6 The GIS interface. Shown are different management units and the

land cover selection box

To visualize different types of land cover, we use three-dimensional models that represent features typical for that type of land cover. If we want a management unit to represent forest, we would fill that unit with trees. To render the landscape in PA we use a base model (e.g. terrain with existing buildings), and sets of three-dimensional models for each land cover option and management unit, which are used to represent the predefined land cover options. Whenever a management unit is changed in the GIS, the appropriate predefined three-dimensional model is loaded into the scene graph.

In a workshop we will typically have one user controlling the GIS interface. The two-dimensional view will include the management units and other features like roads and buildings for orientation. During the workshop the participants will suggest land use change and the GIS user will change the management units accordingly. Having just one person changing the landscape avoids simultaneous changes that may confuse the meeting.

Navigation

The GIS interface can also be used for navigation in the VR view. The GIS user can click onto any point on the two-dimensional map and the viewpoint in the three-dimensional view will be updated. Another option is to select a location from a list of predefined viewpoints. All workshop participants are also able to select a viewpoint from the predefined list using the PDA devices. This allows a greater number of community members to navigate around the landscape—but some control is necessary.

The PDA interface also allows the workshop participants to ‘walk’ around the virtual landscape. The screen of the PDA is divided into quarters, i.e. top, bottom, left and right.

Holding the stylus on the screen will result in continuous movement in the three-dimensional view, e.g. holding the stylus onto the top part of the screen will result in continuous forward movement. The other quarters will result in backward movement and left and right turns. We have elected to use only this simple navigation model at this stage because in order to achieve smooth, continuous movement in the three-dimensional view a lot of information has to be sent over the network, i.e. the current viewpoints have to be constantly updated. To limit network traffic the preferred option is predefined viewpoint selection or viewpoint setting on the GIS interface. The continuous movement functionality on the PDA devices should be used as a supplementary possibility to vary the view a little at a given location, but not to move over large distances.

Voting

Feedback from the workshop participants is an important feature of our approach. Those present can, of course, express opinions verbally at any time. However, there are also occasions when a more formal expression of preference is appropriate. This can be done using the PDA devices. At a given stage a question can be initiated via the GIS interface.

The question is typed in and sent to the PA-server and distributed to PDA device. A vote dialogue will appear showing the question and giving possible answers. Each member of the workshop will then be able to enter his or her preference. Once the voting is finished the results are sent back to the PA-server and then displayed on the VR view. The results will also be written into a log file for later evaluation.

Our system currently supports two types of votes. The first is a simple ‘yes/no’

question. For example, is the currently shown land cover scenario a viable option for the study area? The choice is ‘yes’, ‘no’,? and ‘undecided’. The second type of vote seeks a rating from one to five. For example, how sustainable is the current scenario?

The number of PDA devices is likely to be smaller than the number of workshop participants: people will have to share the PDAs. Each participant will vote and then will pass the PDA on to another person. Once everybody has voted, a ‘done’ button will be

Applications in the agricultural landscape 133

selected on each PDA device. The voting procedure is based on an honour system, i.e. it relies on workshop participants only voting once. Once every PDA has recorded that the voting is finished, the results are sent to the server. The server will evaluate the results and send the result to the main VR view. This will either be the proportion of yes/no votes or the average rating value. In this way the workshop participants will receive instant feedback and can discuss the results further if they so desire.

Impact models

The result of any land use change is not only visual. The quality of life of the local people may be changed in other ways through effects on their environment. Their concerns may include water quality and quantity, salinity, weeds, soil acidity, soil erosion and population numbers. Our system provides an immediate estimation of the non-visual impacts of landscape changes. For each predetermined indicator we can run a model on the GIS that estimates the comparative effect of any changes.

Since our system runs in real time we cannot use very complex models. A computing time of ten seconds may be acceptable but anything above that will stretch the patience of the audience. Thus, we cannot produce very accurate models, but we can give a correct general indication. The outputs are not absolute. Our models will indicate if a scenario increases or decreases each indicator relative to present conditions, or to some other scenario. For example, a scenario using extensive grazing may produce better water quality than a scenario using intensive irrigated horticulture. While for most impact models we will have slightly different ways of computing the values, the common approach will to associate a certain value with each land use type for each impact of interest. Those values will be determined from talking to experts in the respective fields.

Finally, those values will be integrated in one way or the other to give a global estimate for the whole study area.

At any time the GIS user can select a tool on the GIS interface to compare two management unit layers (scenarios). The user can then select the impact models to be compared. The GIS will compute the relevant values and send the results to PA-server for display on the VR view (Figure 7.7). For each impact model we have a small two-dimensional icon that is displayed on the VR screen. The icon will be coloured according to the difference between the two compared scenarios. If there is a change to a larger value the icon will be bright green or dark green (bright green being the stronger change).

If there is a change to a lower value the icon will be red or orange (with red being the stronger change).

The impact modelling combined with the visual assessment, through the VR view, should allow users to quickly assess any scenario. Simplified impact models could occasionally give a false indication. We accept this as a trade-off for real-time performance. More complex process models may produce greater accuracy but these are never accessible in a workshop environment. In seeking to provide people with an interactive tool for exploring and envisioning the future some analysis is better than none.

A more thorough analysis should be undertaken on preferred scenarios.

7.7 Screenshot of a virtual landscape