![]() ![]() NS prob improves in 8 out of the 14Ĭatchments. (R) might either improve or worsen model performance. However, in the remaining 14 catchments, improving only the regionalization procedure X1353) and 95% CI (from 70% to 74% in catchment C8Z1, and from 45% to 52% in catchment Performance of NS prob (from 0.68 to 0.71 in catchment C8Z1, and from 0.65 to 0.72 in catchment ![]() Improving regionalization (case R in figure 6a and 6c) leads to an improvement in the Suitable and adequate representation of the PCs based on observations (see comment above), X1353, where the model is a suitable representation of the regionalized information and a The figures show that for basins C8Z1 and Quantified by 95% CI for the cases BC, R, M, and MR. The parcels a)Īnd b) in this figure summarize the change in accuracy and precision, which is given by NS prob,įor the cases BC, R, M, and MR and the parcels c) and d) do the same for reliability, which is To further investigate the impact of regionalization and model suitability and adequacy in theįlow predictions, posterior values of NS prob and 95% CI are shown in the Figure 6. Very high BFs equal to 53,051 and p-values equal to 1 for all catchments. ![]() Additionally, in the case MR reg, improving the regionalization quality leads to Further, Figure 5a andĥb show that the model is suitable (p-values over 0.05) and adequate (BF over 1) in all the basinsįor the case M M. Regionalization is also suitable for all the catchments, as the p-value is 1. (the value is the same for each basin, as it is based on a fixed error distribution). Regionalization procedure adds 53,051 times more information compared to prior knowledge (3) the model is adequate (Figure 5a and 5b case BC M ) simultaneously only in catchment C8Z1.įor case R reg, BFs in Figure 5b reveal that the strength of evidence is decisive (Table 2), as the Intersection of results for cases BC M and BC reg).įurthermore, (1) the model reproduces the regionalized PCs with high probability (Figure 2.5aĬase Reg&Mod), (2) the regionalization procedure is adequate (Figure 5a and 5b case BC reg ), and The regionalization procedure and model are a suitable and adequate representation of the PCsīased on observations for only two basins (out of 16), basins C8Z1 and X9257 (Figure 5a and 5b, Over 1, Figure 5b case BC M) in catchments C8Z1 and X1353. In the catchment – i.e., the model adds enough information about the PCs based on data (BFs (p-value over 0.05 Figure 2.5a case BC M), and (3) an adequate representation of the PCs observed Then, the model is simultaneously (1) a suitable representation of the regionalized PCs (FigureĢ.5a item Reg&Mod), (2) a suitable representation of the PCs observed in the catchment The p-value for the model suitability test (BC M) is above 0.05 for basin AN520, but the Bayes’įactor is 1 therefore, the model cannot be considered either informative or disinformative. Representation of PCs in three catchments (X1353, X9257, and C8Z1). Therefore, the model is at the same time a suitable and adequate Theįigure 4: Regionalized residuals for PC2 and PC3 for a selected catchment: (a) a fitted joint Gaussian distribution, and (b) a scatterplot. Of information observed in the catchment (i.e., based on observations of flows) in six catchments However, the BC M case shows that model structure is a suitable representation of the four pieces These 12 catchments, the regionalization passes the suitability test and is adequate at the same That the regionalization procedure is adequate in 12 basins (Bayes’ Factor is over 1 – Figure 5b,Ĭase BC reg) and is inadequate for four basins (X1404, AN433, X1353, and X9040). While the adequacy test for the case BC reg shows Representation of the PCs based on observations of flows is accepted for all basins, as p-valuesĪre greater than 0.05 (Figure 5a, item BC reg). In the case of BC reg, the hypothesis that the derived regionalization procedure is a suitable Selected model is not likely to reproduce the four pieces of regionalized information (PCs) at the PCs is rejected for the other 13 basins (Figure 5a, item Reg&Mod). Hypothesis that the model is able to represent the regionalized PCs for 3 catchments,Ĭatchments X1353, AN433, and C8Z1 the model suitability hypothesis based on the regionalized The hydrological model is found suitable (p-value is greater than 0.05) for the null Hydrological model is a suitable representation of the regionalized information in the ungaugedĬatchment. When flow observations are not available, the suitability test for the model shows whether the ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |