Figures & data
Figure 2. Flood predictors: (a) slope angle; (b) topographic position index (TPI); (c) land use; (d) plan curvature; (e) topographic wetness index (TWI); (f) aspect; (g) convergence index (CI); (h) hydrological soil group; (i) distance from rivers; (j) rainfall; (k) elevation; and (l) lithological category
![Figure 2. Flood predictors: (a) slope angle; (b) topographic position index (TPI); (c) land use; (d) plan curvature; (e) topographic wetness index (TWI); (f) aspect; (g) convergence index (CI); (h) hydrological soil group; (i) distance from rivers; (j) rainfall; (k) elevation; and (l) lithological category](/cms/asset/ae833ee1-857e-4ff6-bbb0-fcc588065710/thsj_a_1842412_f0002a_oc.jpg)
Figure 3. Frequency distribution of flood pixels within the flood predictor classes
![Figure 3. Frequency distribution of flood pixels within the flood predictor classes](/cms/asset/56b55efd-a06d-495b-b59a-4dc44e16dbf8/thsj_a_1842412_f0003_oc.jpg)
Figure 4. General ANFIS architecture of first-order Takagi-Sugeno fuzzy model
![Figure 4. General ANFIS architecture of first-order Takagi-Sugeno fuzzy model](/cms/asset/c34a7707-188d-4689-8d01-bcfc9c029e72/thsj_a_1842412_f0004_oc.jpg)
Figure 5. Flowchart of the developed methodology. ANFIS - Adaptive Neuro-Fuzzy Inference System; AHP - Analytical Hierarchy Process; CF - Certainty Factor; WOE - Weights of Evidence; RMSE - Root Mean Sqaured Error; K-index- Kappa Index; FPI - Flood Potential Index
![Figure 5. Flowchart of the developed methodology. ANFIS - Adaptive Neuro-Fuzzy Inference System; AHP - Analytical Hierarchy Process; CF - Certainty Factor; WOE - Weights of Evidence; RMSE - Root Mean Sqaured Error; K-index- Kappa Index; FPI - Flood Potential Index](/cms/asset/fb7166bf-4798-415e-ab24-32f6d7110357/thsj_a_1842412_f0005_oc.jpg)
Figure 6. Performance in terms of RMSE calculated for the training dataset: (a) ANFIS-AHP, (c) ANFIS-CF and (e) ANFIS-WOE, and the validation dataset: (b) ANFIS-AHP, (d) ANFIS-CF and (f) ANFIS-WOE
![Figure 6. Performance in terms of RMSE calculated for the training dataset: (a) ANFIS-AHP, (c) ANFIS-CF and (e) ANFIS-WOE, and the validation dataset: (b) ANFIS-AHP, (d) ANFIS-CF and (f) ANFIS-WOE](/cms/asset/9717d225-f727-4933-b2f0-d8dfed8708b8/thsj_a_1842412_f0006_oc.jpg)
Table 1. Random consistency index (RI; after Saaty Citation1980)
Table 2. Multicollinearity analysis and predictive ability of independent variables. TPI: topographic position index; TWI: topographic wetness index; TOL: tolerance; VIF: variance inflation; IG: information gain
Table 3. Pairwise comparison matrices
Table 4. Pairwise comparison matrix properties. CI: consistency index; CR: consistency ratio; RI: random consistency index; TPI: topographic position index; TWI: topographic wetness index
Table 5. Certainty factor (CF) and weight-of-evidence (WOE) coefficients. TPI: topographic position index; TWI: topographic wetness index
Table 6. Statistical metrics used to evaluate model performance. TP: true positive; TN: true negative; FP: false positive; FN: false negative
Figure 7. Relative importance of flood conditioning factors
![Figure 7. Relative importance of flood conditioning factors](/cms/asset/dee97779-3fc8-4b2a-ad6b-085f745f38ca/thsj_a_1842412_f0007_oc.jpg)
Figure 8. Map of FPIANFIS-AHP values across the Trotuș River basin
![Figure 8. Map of FPIANFIS-AHP values across the Trotuș River basin](/cms/asset/cc62fbd8-d72c-4425-b2c4-45afd19f9915/thsj_a_1842412_f0008_oc.jpg)
Figure 9. Map of FPIANFIS-CF values across the Trotuș River basin
![Figure 9. Map of FPIANFIS-CF values across the Trotuș River basin](/cms/asset/2f54cd33-0dea-4be1-b0bd-9c754c3b13aa/thsj_a_1842412_f0009_oc.jpg)
Figure 10. Weights of the FPI classes
![Figure 10. Weights of the FPI classes](/cms/asset/48fe7f99-d0ee-4941-aea8-fe6d552e4460/thsj_a_1842412_f0010_oc.jpg)
Figure 11. Map of FPIANFIS-WOE values across the Trotuș River basin
![Figure 11. Map of FPIANFIS-WOE values across the Trotuș River basin](/cms/asset/f63c9df4-b73c-4c3b-8f5b-db2f4b9fe6e9/thsj_a_1842412_f0011_oc.jpg)
Figure 12. ROC curves: (a) success rate and (b) prediction. AUC: area under curve
![Figure 12. ROC curves: (a) success rate and (b) prediction. AUC: area under curve](/cms/asset/e1f11004-e833-4cbd-a936-1ea292f9ddd7/thsj_a_1842412_f0012_oc.jpg)
Table 7. Flood point distribution (%) within FPI classes