The forecast of river stream flows is of significant importance for the development of early warning systems. Artificial intelligence algorithms have proven to be an effective tool in hydrological modeling data-driven, since they allow establishing relationships between input and output data of a watershed and thus make decisions data-driven. This article investigates the applicability of the k-nearest neighbor (KNN) algorithm for forecasting the mean daily flows of the Ramis river, at the Ramis hydrometric station. As input to the KNN machine learning algorithm, we used a data set of mean basin precipitation and mean daily flow from hydrometeorological stations with various lags. The performance of the KNN algorithm was quantitatively evaluated with hydrological ability metrics such as mean absolute percentage error (MAPE), anomaly correlation coefficient (ACC), Nash-Sutcliffe efficiency (NSE), Kling-Gupta efficiency (KGE') and the spectral angle (SA). The results for forecasting the flows of the Ramis river with the k-nearest neighbor machine learning algorithm reached high levels of reliability with flow lags of one and two days and precipitation with three days. The algorithm used is simple but robust to make short-term flow forecasts and can be integrated as an alternative to strengthen the daily hydrological forecast of the Ramis river.

Gridded meteorological datasets represent a valuable alternative for hydrometeorological applications, especially in areas with limited ground observations. However, it is important to evaluate these datasets to quantify their accuracy, error, and bias in estimates. The objective of this research was to assess the hydrological performance of gridded meteorological products in Peruvian Altiplano basins. Three evaluation approaches were employed: 1) pixel-to-point, 2) basin-averaged values, and 3) as forcings in hydrological modeling. Using precipitation (P) and potential evapotranspiration (PET) data from AgERA5, ERA5-Land, MERRA-2, and PERSIANN-CDR, comparisons were made with 33 local meteorological stations using statistical metrics such as correlation coefficient (CC), root mean square error (RMSE), and percentage bias (PBIAS). Additionally, the GR4J hydrological model was employed in four watersheds to assess model performance with different data combinations. Results showed that AgERA5 and ERA5-Land excelled in daily precipitation estimation, while MERRA-2 had the lowest PET bias. Using reference data for P and PET as model forcings yielded consistent results across basins. However, the performance declined when using gridded products for both P and PET, revealing significant limitations in replicating hydrological processes and emphasizing the need to enhance these products for areas with limited observational data.

La presente investigación se realizó en la cuenca del río Ramis, ubicado en la región hidrográfica del Titicaca, Perú, con el objetivo de calibrar y validar el modelo hidrológico Sacramento (SAC-SMA) desde un enfoque semidistribuido. La información hidrometeorológica utilizada de precipitación pluvial, temperatura y caudal, corresponden a una serie de registro 2005 – 2016. La metodología de interpolación espacial de datos meteorológicos en la estación virtual, fue estimada mediante el procedimiento de Shepard y la evapotranspiración potencial por el modelo de Turc, estas metodologías vienen incorporados en la plataforma RS-MINERVE y son estimaciones automatizadas. La fase de calibración y validación del modelo, se realizó de forma aleatoria con el 70% y 30% del total de datos respectivamente. La evaluación estadística de eficiencia y error fueron medidos a través del coeficiente de Nash, coeficiente de Nash para valores del logaritmo y raíz del error medio cuadrático relativo. Los resultados son satisfactorios y se afirma que las salidas del modelo hidrológico representan adecuadamente los caudales de avenida y estiaje, constituyéndose como una alternativa para el fortalecimiento del pronóstico hidrológico a paso de tiempo diario del rio Ramis.

Reference evapotranspiration (ETo) is crucial for agriculture and is traditionally estimated using the Penman–Monteith (PM) method, which relies on multiple climatic variables. This study assessed machine learning (ML) techniques to estimate ETo at the Yauri meteorological station in Peru. Two ML models—K-nearest neighbors (KNN) and artificial neural networks (ANN)—were tested and compared against both the PM and the Hargreaves–Samani (HS) methods. Their accuracy was measured using metrics such as mean absolute error (MAE), anomaly correlation coefficient (ACC), Nash–Sutcliffe efficiency (NSE), Kling–Gupta efficiency (KGE), and spectral angle (SA). The results indicate that ML techniques can effectively estimate ETo, providing robust alternatives in areas with limited meteorological data, thus enhancing water resource management.