Existing methods for land surface form characterisation often focus on relatively simple landform element classifications and do not evaluate results with large scale participant experiments. This piece of research takes a semantics-grounded approach to characterising the landform valley. Based on definitions three algorithms to characterise valleys in a fuzzy way are introduced. Comparison of the results to assessments regarding the degree of being in a valley gained from over 800 participants in a questionnaire survey yields significant amounts of explained variance (R2 D 0.35–0.37). Furthermore, accounting for very ambiguously perceived stimuli showing vast low places leads to markedly improved regressions (R2 D 0.45–0.49), weighting of the data with a measure of uncertainty in judgment even more so (R2 D 0.50–0.55).