Development of a Sap Flow Prediction Model for 'Shingo' Pear Trees Using Growth and Meteorological Information

• DOI: 10.37727/jkdas.2025.27.2.385
• Publication Date: 2025-04-28T01:47:40Z
• Authors: Won Gi Choi; Dong-Eon Kim; Sang Hyun Lee; Min Soo Kim

Understanding sap flow is crucial for comprehending tree growth. However, it impractical all farms to directly measure flow. Although many studies have attempted predict flow, most of them relied on hard-to-obtain data in addition weather information. This study aims propose a model predicting ‘Shingo’ pear trees using only easily accessible variables, civil twilight time, and DAFB (Days After Full Bloom). Additionally, account the time-lag effect nonlinear relationship between we performed lagged Spearman correlation analysis. For solar radiation, coefficient was found peak at 4-hour lag. prediction model, used machine learning models such as Random Forest XGBoost, well deep including LSTM, GRU, BiLSTM, hybrid CNN-GRU-BiLSTM model. All achieved R² values above 0.94, demonstrating that feasible proposed variables. In particular, best with an MAE 225.4, RMSE 427.6, 0.9550. We concluded well-suited environments seasonal localized climate phenomena, South Korea. Through this, expect provide data-driven irrigation standard across country.