Flash-flood forecasting has emerged worldwide due to the catastrophic socio-economic impacts this hazard might cause and the expected increase of its frequency in the future. In mountain catchments, precipitation-runoff forecasts are limited by the intrinsic complexity of the processes involved, particularly its high rainfall variability. While process-based models are hard to implement, there is a potential to use the random forest algorithm due to its simplicity, robustness and capacity to deal with complex data structures. Here a step-wise methodology is proposed to derive parsimonious models accounting for both hydrological functioning of the catchment (e.g., input data, representation of antecedent moisture conditions) and random forest procedures (e.g., sensitivity analyses, dimension reduction, optimal input composition). The methodology was applied to develop short-term prediction models of varying time duration (4, 8, 12, 18 and 24 h) for a catchment representative of the Ecuadorian Andes. Results show that the derived parsimonious models can reach validation efficiencies (Nash-Sutcliffe coefficient) from 0.761 (4-h) to 0.384 (24-h) for optimal inputs composed only by features accounting for 80% of the model’s outcome variance. Improvement in the prediction of extreme peak flows was demonstrated (extreme value analysis) by including precipitation information in contrast to the use of pure autoregressive models.

Information on the spatiotemporal rainfall occurrence, its microphysical characteristics, and its reflectivity–rainfall (Z–R) relations required to provide rainfall mapping based on rain radar data is limited for tropical high mountains. Therefore, this study aims to analyze rainfall types in the Andes cordillera to derive different rain-type Z–R relations using disdrometer observations at three study sites representative for different geographic positions and elevations (2610, 3626, and 3773 m MSL). Rain categorization based on mean drop volume diameter (Dm) thresholds [0.1 < Dm (mm) ≤ 0.5; 0.5 < Dm (mm) ≤ 1.0; 1.0 < Dm (mm) ≤ 2.0] was performed using drop size distribution data at a 5-min time step over an approximate 2-yr period at each location. The findings are as follows: (i) Rain observations characterized by higher (lower) Dm and rain rates are more frequent at the lower (higher) site. (ii) Because of its geographic position, very light rain (drizzle) is more common at higher altitudes with longer-duration events, whereas rainfall is more convective at the lower range. (iii) The specific spatial exposition regarding cloud and rain formation seems to play an important role for derivation of the local Z–R relationship. (iv) Low A coefficients (≤60) for the first rain type resemble typical characteristics of orographic precipitation. (v) Greater values of A (lowest and highest stations for Dm > 1.0 mm) are attributed to transitional rainfall as found in other studies. (vi) Rain-type Z–R relations show a better adjustment in comparison with site-specific Z–R relationships. This study is the first contribution of Z–R relations for tropical rainfall in the high Andes.

No abstract