A comparison of soil texture-by-feel estimates: Implications for the citizen soil scientist

TitleA comparison of soil texture-by-feel estimates: Implications for the citizen soil scientist
Publication TypeJournal Article
Year of Publication2018
AuthorsSalley S.W., Herrick JE, Holmes C, Karl JW, Levi MR, McCord S, VAN DE WAAL CORNELIS, Van Zee J
JournalSoil Science Society of America
Start Page1526
Date Published08/2018
ARIS Log Number345276

Estimating soil texture is a fundamental practice universally applied by soil scientists to classify and understand the behavior, health, and management of soil systems. While the accuracy of both the soil texture class and the estimates of percent sand and clay is generally accepted when completed by trained soil scientists, similar estimates by “citizen scientists” or inexperienced seasonal resource scientists are often questioned. We compared soil texture classes determined by texture-by-feel and laboratory analyses for 2 groups: professional soil scientists who contributed to the USDA-NRCS National Soil Characterization Database and seasonal field technicians working on rangeland inventory and assessment programs in the Western United States and Namibia. Texture accuracy was compared using a confusion matrix to evaluate classification accuracy based on the assumption that laboratory measurements were correct. Our results show that the professional soil scientists predicted the laboratory-determined texture class for 66% of the samples. Accuracy for seasonal field technicians was between 27% and 41%. When a “correct” prediction was defined to include texture classes adjacent to the laboratory-determined texture based on a standard USDA texture triangle, accuracy increased to 91% for professionals and 71% to 78% for seasonal field technicians. These findings highlight the need to improve options for increasing the accuracy of field-textured estimates for all soil texture observers, with relevance to career soil scientists, seasonal technicians, and citizen scientists. Opportunities for improving soil texture accuracy include training, calibration, and decision support tools that go beyond simple dichotomous keys.