Open AccessMulti-step-ahead forecasting of hourly potential evapotranspiration for irrigation triggering in horticultural nurseries under oceanic climate
Author(s) -
Rousseau Tawegoum
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/958/1/012025
Subject(s) - evapotranspiration , irrigation scheduling , environmental science , kalman filter , context (archaeology) , data assimilation , autoregressive model , meteorology , climatology , climate change , econometrics , statistics , geography , mathematics , ecology , archaeology , biology , geology , soil science , soil water
Predicting hourly potential evapotranspiration is particularly important in constrained horticultural nurseries. This paper presents a three-step-ahead predictor of potential evapotranspiration for horticultural nurseries under unsettled weather conditions or climate sensor failure. The Seasonal AutoRegressive Integrated Moving Average model based on climate data was used to derive a predictor using data generated according to prior knowledge of the system behavior; the aim of the predictor was to compensate for missing data that are usually not considered in standard forecasting approaches. The generated data also offer the opportunity to capture variations of the model parameters due to abrupt changes in local climate conditions. A recursive algorithm was used to estimate parameter variation, and the Kalman filter to model the state of the system. The simulations for steady-state weather and unsettled weather conditions showed that the predictor could forecast potential evapotranspiration more accurately than the standard approach did. These results are encouraging within the context of predictive irrigation scheduling in nurseries.