Several methods have been tried to estimate air temperature using satellite imagery. In this paper, the results of two machine learning algorithms, Support Vector Machines and Random Forest, are compared with Multiple Linear Regression and Ordinary kriging. Several geographic, remote sensing and time variables are used as predictors. The validation is carried out using two different approaches, a leave-one-out cross validation in the spatial domain and a spatio-temporal k-block cross-validation, and four different statistics on a daily basis, allowing the use of ANOVA to compare the results. The main conclusion is that Random Forest produces the best results (R 2 = 0.888 ± 0.026, Root mean square error = 3.01 ± 0.325 using k-block cross-validation). Regression methods (Support Vector Machine, Random Forest and Multiple Linear Regression) are calibrated with MODIS data and several predictors easily calculated from a Digital Elevation Model. The most important variables in the Random Forest model were satellite temperature, potential irradiation and cdayt, a cosine transformation of the julian day.

Se propone una metodología para realizar proyecciones espacio-temporales de la evapotranspiración de referencia (ET 0) y el posible impacto de cambio climático en cuencas semiáridas. El estudio se desarrolla en la Demarcación Hidrográfica del Segura, utilizando para estimar la ET 0 el modelo de Hargreaves calibrado mensualmente mediante el método de Allen. Se parte de una regionalización de temperaturas máximas y mínimas a partir de los modelos globales MPEH5 y MPEH5C y los escenarios de emisión SRESA2, SRESA1B, SRESB1 y E1. Se ha evaluado la tendencia temporal y la distribución espacial de ET 0 mediante Theil-Sen, Mann-Kendall y 3 métodos de interpolación (regresión lineal múltiple, krigreado ordinario y regresión-krigeado). Se observa un incremento significativo en los 4 escenarios entre los 2,73 mm para el SRESA2 y los 0,63 mm para E1. Se observa además un patrón espacial caracterizado por un aumento mayor de la ET 0 en zonas de cabecera y menor en la zona litoral.

Large ensembles of climate models are increasingly available either as ensembles of opportunity or perturbed physics ensembles, providing a wealth of additional data that is potentially useful for improving adaptation strategies to climate change. In this work, we propose a framework to evaluate the predictive capacity of 11 multi-model ensemble methods (MMEs), including random forest (RF), to estimate reference evapotranspiration (ET0) using 10 AR5 models for the scenarios RCP4.5 and RCP8.5. The study was carried out in the Segura Hydrographic Demarcation (SE of Spain), a typical Mediterranean semiarid area. ET0 was estimated in the historical scenario (1970–2000) using a spatially calibrated Hargreaves model. MMEs obtained better results than any individual model for reproducing daily ET0. In validation, RF resulted more accurate than other MMEs (Kling–Gupta efficiency (KGE) M=0.903, SD=0.034 for KGE and M=3.17, SD=2.97 for absolute percent bias). A statistically significant positive trend was observed along the 21st century for RCP8.5, but this trend stabilizes in the middle of the century for RCP4.5. The observed spatial pattern shows a larger ET0 increase in headwaters and a smaller increase in the coast.

Currently, water demands from urban and agricultural use are increasing, especially in arid and semiarid regions, such as the Mediterranean. This situation is expected to become worse with the climate change projections for the region, increasing the pressure, in both quantity and quality, on fresh water resources. Evapotranspiration (ET0) is a hydrologic variable with high uncertainty and considered incorrect in water balance estimations. However, its accurate assessment is essential to obtain the real value of available water to satisfy water demands, especially in extended agricultural areas such as the southeast of Spain. ET0 can be obtained using different equations with different levels of input data requirements, among them the Penman-Monteith option is the one recommended by the FAO (PMFAO), but its input data requirements are high. On the other hand, there are simpler options, such as the Hargreaves equation (ET0,HG), but there is not such a big agreement about its accuracy in the scientific literature. The main objection to the use of PMFAO is the lack of some of the required meteorological variables in most climate stations, forcing the use of simpler alternatives. This paper presents an R-CRAN code where the ET0,HG, parameterized by Samani, is calibrated and validated with the Allen model considering 18 statistical contrasts. Both ET0,HG results (pre-and post-calibrated) are compared with daily, monthly and annual results of the PMFAO. All meteorological data was provided by the CA52 Cartagena La Aljorra weather station, managed by the Agricultural Information System of the Murcia region (SE Spain). The main results show that daily, monthly and annual ET0,HG results after the Allen calibration and validation are similar to the PMFAO. However, a moderate underestimation of ET0,HG compared to PMFAO was identified. To sum up, the presented R-CRAN code provides an alternative to apply the ET0,HG method with few meteorological input requirements and, once calibrated, can be applied to extended data networks in other regions.