How do you evaluate GIS data when it's uncertain?

One thing I've found helpful in research when GIS is uncertain is, that I like to assess and test other parts of the environment to help make GIS assessments make sense. For what is considered problem areas, I tend to monitor it through other seasons. I do this to see if it is an isolated problem or if it is a year-round issue to approach. If it is isolated I can begin a solution and if not can begin trial and error.

Look at the source and the metadata, who made it and the date, how it was made, plot the data and analyze it visually. See if it makes sense based on known features in orthophotos or in vector datasets. Ground truth it

You can visualize data uncertainty in many ways. Think of a scenario where you know your information will be +/- 20% (plus or minus 20 percent) different than a certain number. You can display a 'high', 'median', and 'low' in colors like green, yellow, and red in your data to show the possible spread of uncertainty in your information. Displaying uncertainties in a different way can show the possible outcomes better in some situations and can allow you to prepare for different conditions.

You can utilize algorithms, simulations, scripts, or other methods to test for uncertainty. For example, imagine you must make a dataset to predict the flow of a waterway - a future path. You can utilize simulations like the Monte Carlo Simulation to predict multiple possible outcomes. These outcomes are your uncertainties, which can be predicted to some degree. Being able to show these uncertainties in your data is a projection and could have percentage error that needs to be accounted for.

You do research on the data source. Make sure the source is reliable and know your data. How it was acquired, the scale, by who it was acquired. Look at comments fields and Metadata. You can ground truth data or compare it to known locations. For example, I found parcel boundary discrepancies using GIS analysis between what the Land Titles Office has recorded, and what surveyors had recorded. Because of my work, LTSA corrected the boundaries.

If you are in a position of control over the data acquisition team its wise to build in some data validation activity as a project deliverable. For instance defining the position of a nearby permanent object (eg trig point) provides data that can be validated at a later date to check equipment error. Acquisition teams should also provide full specification for applicable co-ordinate reference systems and document transform tests to lat/long, particularly important for teams working UTM in offshore environments where end users may need to transform to local co-ordinate systems.