Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Sep 24;9(1):13763.
doi: 10.1038/s41598-019-50376-w.

Digital soil mapping including additional point sampling in Posses ecosystem services pilot watershed, southeastern Brazil

Bárbara Pereira Christofaro Silva  1 , Marx Leandro Naves Silva  2 , Fabio Arnaldo Pomar Avalos  2 , Michele Duarte de Menezes  2 , Nilton Curi  2
Affiliations

Digital soil mapping including additional point sampling in Posses ecosystem services pilot watershed, southeastern Brazil

Bárbara Pereira Christofaro Silva et al. Sci Rep. .

Erratum in

Abstract

This study aimed to evaluate the performance of three spatial association models used in digital soil mapping and the effects of additional point sampling in a steep-slope watershed (1,200 ha). A soil survey was carried out and 74 soil profiles were analyzed. The tested models were: Multinomial logistic regression (MLR), C5 decision tree (C5-DT) and Random forest (RF). In order to reduce the effects of an imbalanced dataset on the accuracy of the tested models, additional sampling retrieved by photointerpretation was necessary. Accuracy assessment was based on aggregated data from a proportional 5-fold cross-validation procedure. Extrapolation assessment was based on the multivariate environmental similarity surface (MESS). The RF model including additional sampling (RF*) showed the best performance among the tested models (overall accuracy = 49%, kappa index = 0.33). The RF* allowed to link soil mapping units (SMU) and, in the case of less-common soil classes in the watershed, to set specific conditions of occurrence on the space of terrain-attributes. MESS analysis showed reliable outputs for 82.5% of the watershed. SMU distribution across the watershed was: Typic Rhodudult (56%), Typic Hapludult* (13%), Typic Dystrudept (10%), Typic Endoaquent + Fluventic Dystrudept (10%), Typic Hapludult (9.5%) and Rhodic Hapludox + Typic Hapludox (2%).

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
(a) Location of the study area and the soil survey sampling sites and their associated additional observations. (b,c) Detail of the field and additional observations clusters.
Figure 2
Figure 2
(a) Allocation of additional observations (AO) to field observations (FO) in Google Earth for the soil mapping unit (SMU) Typic Endoaquent + Fluventic Dystrudept. Source: Google Earth Pro 7.3, https://www.google.com.br/earth/download/gep/agree.html. (b) Original and final proportions of the training datasets by SMU.
Figure 3
Figure 3
Soil map production flow chart. MLR: multinomial logistic regression, C5-DT: C5 decision tree, RF: random forest, DSM: digital soil map, MESS: multivariate environmental similarity surface.
Figure 4
Figure 4
Mean overall accuracy of the tested models and the effects of additional observations (AO accuracy) in relation to field observations (FO accuracy). Bars are standard errors from 5-fold cross-validation after 100 repetitions. C5-DT: C5 decision-tree, RF: random forest, MLR: multinomial logistic regression.
Figure 5
Figure 5
Importance of selected covariates by soil mapping unit. TWI: topographic wetness index; DI: direct insolation; NO: negative openness; MRVBF: multi-resolution index of valley bottom flatness.
Figure 6
Figure 6
Occurrence signatures of soil mapping units projected on the feature-space of selected covariates. X-axis: covariate value, Y-axis: change of predicted class probability. R-squared values indicate goodness-of-fit between feature and prediction space projections (solid curves). TWI: topographic wetness index; DI: direct insolation; NO: negative openness; MRVBF: multiresolution index of valley bottom flatness.
Figure 7
Figure 7
Multivariate environmental similarity surface (MESS) of the chosen model (random forest with additional observations), extrapolation level increase along with negative values.
Figure 8
Figure 8
Soil mapping units distribution in the watershed by the RF* model.
Figure 9
Figure 9
Extrapolation level of the digital soil map (lighter colors) by soil mapping unit.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

  • Disentangling Jenny's equation by machine learning.
    Prieto-Castrillo F, Rodríguez-Rastrero M, Yunta F, Borondo F, Borondo J. Prieto-Castrillo F, et al. Sci Rep. 2023 Nov 27;13(1):20916. doi: 10.1038/s41598-023-44171-x. Sci Rep. 2023. PMID: 38017030 Free PMC article.

References

    1. Brevik EC, et al. Soil mapping, classification, and pedologic modeling: History and future directions. Geoderma. 2016;264:256–274. doi: 10.1016/j.geoderma.2015.05.017. - DOI
    1. Brevik EC, Hartemink AE. Early soil knowledge and the birth and development of soil science. CATENA. 2010;83:23–33. doi: 10.1016/j.catena.2010.06.011. - DOI
    1. Keys K, Sterling SM, Guan Y. Using historic soil survey data to map water erosion hazard for land-use planning in Nova Scotia. Can. J. Soil Sci. 2015;95:299–304. doi: 10.4141/cjss-2014-113. - DOI
    1. Jenny, H. Factors of Soil Formation: A System of Quantitative Pedology. (McGraw-Hill, 1941).
    1. McBratney A, Mendonça Santos M, Minasny B. On digital soil mapping. Geoderma. 2003;117:3–52. doi: 10.1016/S0016-7061(03)00223-4. - DOI

Publication types