Scholarlink

<p>A quasi-periodic model was developed to analyze and forecast monthly precipitation in the Alto Atoyac Basin, México. The model used 40 years (1982–2022) of CHIRPS v2.0 data. This model uses linear regression and a six-harmonic Fourier series in a novel way to jointly capture seasonal variability and long-term trends. The Augmented Dickey-Fuller (ADF) test revealed that the monthly series is stationary (<i>p</i>-value = 0.01) with quasi-periodic fluctuations. In contrast, the annual series is non-stationary (<i>p</i>-value = 0.51) and exhibits a significant upward trend in accumulated precipitation. To determine model complexity, harmonic configurations from three to six components were tested. The six-harmonic model performed best, achieving a Pearson r of 0.90 and a Mean Percentage Error (MPE) of -0.30% during the training period. During the test period (2023–2024), the model showed consistent accuracy (Pearson r = 0.84 and MPE = 1.68%). Residual diagnostics confirmed statistical assumptions of homoscedasticity, normality, and independence, e.g., t-test (<i>p</i>-value = 0.99) and Ljung-Box (<i>p</i>-value = 0.39). Significance tests on the linear trend parameters confirmed the intercept and slope as highly significant (<i>p</i>-value &lt; 0.001), which reinforces the evidence of a persistent positive trend. The model also provides intervals ±1 and ±2 standard deviations, which support uncertainty analysis. This adaptable, computationally efficient model integrates seamlessly into hydrological simulation tools and decision-support systems. Its accurate representation of climate variability makes it valuable for water resource planning, flood control, drought mitigation, and agricultural management, especially under changing climate conditions and in other comparable basins. </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

Quasi-periodic model: an alternative approach for forecasting monthly precipitation in a high-mountain basin to optimize water management

  • 期刊论文

  • 发表日期：

    2025年7月8日

查看全文
  • Adriana Serrano-Santiago,
  • René Ledesma-Alonso,
  • Daniela Cortes-Toto,
  • Maria Elena Cerecedo-Arroyo,
  • Adriana Palacios

摘要

A quasi-periodic model was developed to analyze and forecast monthly precipitation in the Alto Atoyac Basin, México. The model used 40 years (1982–2022) of CHIRPS v2.0 data. This model uses linear regression and a six-harmonic Fourier series in a novel way to jointly capture seasonal variability and long-term trends. The Augmented Dickey-Fuller (ADF) test revealed that the monthly series is stationary (p-value = 0.01) with quasi-periodic fluctuations. In contrast, the annual series is non-stationary (p-value = 0.51) and exhibits a significant upward trend in accumulated precipitation. To determine model complexity, harmonic configurations from three to six components were tested. The six-harmonic model performed best, achieving a Pearson r of 0.90 and a Mean Percentage Error (MPE) of -0.30% during the training period. During the test period (2023–2024), the model showed consistent accuracy (Pearson r = 0.84 and MPE = 1.68%). Residual diagnostics confirmed statistical assumptions of homoscedasticity, normality, and independence, e.g., t-test (p-value = 0.99) and Ljung-Box (p-value = 0.39). Significance tests on the linear trend parameters confirmed the intercept and slope as highly significant (p-value < 0.001), which reinforces the evidence of a persistent positive trend. The model also provides intervals ±1 and ±2 standard deviations, which support uncertainty analysis. This adaptable, computationally efficient model integrates seamlessly into hydrological simulation tools and decision-support systems. Its accurate representation of climate variability makes it valuable for water resource planning, flood control, drought mitigation, and agricultural management, especially under changing climate conditions and in other comparable basins.