Scholarlink

Abstract <p>Effective rainfall (<i>P</i><sub>eff</sub>) plays an important role in irrigation management, but measured <i>P</i><sub>eff</sub> data are often scarce. This study introduces a method using linear regression to determine the coefficients of the empirical formula for computing <i>P</i><sub>eff</sub> in the CROPWAT model based on the rainfall data (1981–2020) at three stations Hung Yen, Bai Thuong, and Nha Trang in Vietnam. Rainfall data were clustered using the 80th percentile threshold, two regression approaches were applied: clustered regression (<i>P</i><sub>eff</sub> <i>= aP</i> when <i>P</i> ≤ <i>z</i>; <i>P</i><sub>eff</sub> <i>= cP + d</i> when <i>P &gt; z</i>) and non-clustered regression (<i>P</i><sub>eff</sub> <i>= aP</i>). The model’s performance was evaluated using statistical indices, including <i>R</i><sup>2</sup>, <i>RMSE</i>, and <i>MAE</i>. The results showed that the clustered regression model achieved superior performance (<i>R</i><sup>2</sup>: 0.976–1.0, <i>RMSE</i>: 0–12.65 mm, and <i>MAE</i>: 0–1.97 mm), particularly under heavy rainfall conditions (error: 0–3.61% compared to the USDA SCS formula). In contrast, the non-clustered regression model was more suitable for areas with a distinct rainy season, such as Nha Trang (<i>R</i><sup>2</sup>:&#xa0;0.892–0.924, <i>RMSE</i>: 26.58–36.77 mm, and <i>MAE</i>: 20.89–30.79 mm). Sensitivity analysis showed that the most sensitive coefficient a (±10% variation in <i>P</i><sub>eff</sub>), provided a basis for model calibration. The application of irrigation water requirement calculation for rice in 2019 showed errors ranging from 2.37 to 25.54%, thereby demonstrating its practical value for optimizing irrigation water use. Overall, the research provides an effective solution to support irrigation planning in areas lacking <i>P</i><sub>eff</sub> data, contributing to improved agricultural production efficiency in Vietnam.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Determining Empirical Coefficients for Estimating Effective Rainfall in the CROPWAT Model Using Linear Regression

  • 期刊论文

  • 发表日期：

    2026年5月4日

查看全文
  • Nguyen Quang Phi,
  • Tran Quoc Lap

摘要

Abstract

Effective rainfall (Peff) plays an important role in irrigation management, but measured Peff data are often scarce. This study introduces a method using linear regression to determine the coefficients of the empirical formula for computing Peff in the CROPWAT model based on the rainfall data (1981–2020) at three stations Hung Yen, Bai Thuong, and Nha Trang in Vietnam. Rainfall data were clustered using the 80th percentile threshold, two regression approaches were applied: clustered regression (Peff = aP when P ≤ z; Peff = cP + d when P > z) and non-clustered regression (Peff = aP). The model’s performance was evaluated using statistical indices, including R2, RMSE, and MAE. The results showed that the clustered regression model achieved superior performance (R2: 0.976–1.0, RMSE: 0–12.65 mm, and MAE: 0–1.97 mm), particularly under heavy rainfall conditions (error: 0–3.61% compared to the USDA SCS formula). In contrast, the non-clustered regression model was more suitable for areas with a distinct rainy season, such as Nha Trang (R2: 0.892–0.924, RMSE: 26.58–36.77 mm, and MAE: 20.89–30.79 mm). Sensitivity analysis showed that the most sensitive coefficient a (±10% variation in Peff), provided a basis for model calibration. The application of irrigation water requirement calculation for rice in 2019 showed errors ranging from 2.37 to 25.54%, thereby demonstrating its practical value for optimizing irrigation water use. Overall, the research provides an effective solution to support irrigation planning in areas lacking Peff data, contributing to improved agricultural production efficiency in Vietnam.