02057nam a2200253 a 450000100080000000500110000800800410001902400410006010000190010124500910012026001270021152012320033865000200157065000230159065000180161365000220163165000240165370000200167770000280169770000200172570000220174570000180176770000180178521483592023-01-17       2022    bl uuuu     u00u1 u    #d7 a10.1109/IJCNN55064.2022.98924682DOI1 aRIBEIRO, V. P.  aBayesian network for hydrological modelban inference approach.h[electronic resource]  aIn: INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2022, Padua, Italy, Proceedings...[S.l.: s.n.], 2022.c2022  aAbstract: According to the Food and Agriculture Organisation, there are growing concerns about the availability and use of water in agriculture. The hydrological model generates a water balance and the resulting value indicates the amount of available water in a given area. The calculation of the water balance is fundamental for the development of new strategies for the management of water resources. One of its main adversities is the estimation of evapotranspiration, which may be considered a fundamental component. This factor considers climatological variables collected from weather stations that are spread over large areas. However, there are frequent cases of long periods of missing data. We evaluated the performance of a Bayesian Network inference model for estimating evapotranspiration in a large agricultural region in Brazil. To this end, the method considered factors such as accuracy, missing data, and model portability. The results indicate that the model achieves up to 86% accuracy when comparing estimated values to expected values derived from the Penman-Monteith equation. The results show that wind speed and relative humidity are the most critical climatological variables for accurate estimation.  aBayesian theory  aEvapotranspiration  aWater balance  aBalanço Hídrico  aEvapotranspiração1 aPADOVANI, C. R.1 aBALESTIERI, J. A. P. B.1 aCUNHA, A. S. M.1 aMARQUES, P. A. A.1 aDUARTE, S. N.1 aMACIEL, C. D.