Scholarlink

<p>Predicting crop yield is complex due to its dependence on multiple meteorological variables. Remote sensing (RS) has become the prime source of climate parameters due to detailed spatial and temporal resolutions; however, its product needs further quality enhancement due to associated errors. The primary aim of this study is to incorporate the EEMD, as a signal modification technique, with the LSTM model, aiming to anticipate crop yield of four strategic crops, namely barley, lentils, pea, and wheat in all provinces of Iran. The annual crop yield of these crops was extracted from Iran’s Ministry of Agriculture data over 15 years, spanning from 2005 to 2020. In the context of EEMD-LSTM prediction, four influential climate parameters, including minimum and maximum temperature, precipitation, and the SPEI, were considered as the input variables. The analysis shows that applying EEMD generally enhances the predictive performance of the LSTM model for most agricultural products. For barley, lentils, and peas, EEMD substantially improves accuracy compared with the baseline model. Although its impact on wheat is not statistically significant, EEMD still reduces RMSE by approximately 42%. For lentils, the method yields notable improvements, with reductions of 15.7, 13.8, and 8% in MAE, RMSE, and PCC, respectively. Additionally, when noisy wheat data are removed, the error distribution slightly increases, indicating that wheat exhibits the least improvement among all evaluated crops. The spatial assessment reveals clear geographic differences in the effectiveness of EEMD. The method substantially reduces prediction errors for barley in the northwestern region. A similar, though smaller, improvement is observed for lentils in the same area. Conversely, EEMD increases prediction errors for chickpea production in both the northwest and eastern regions, highlighting strong spatial variability in the model’s performance across the study area. The utilization of EEMD, as a noise removal tool, evidently leads to a reduction in model errors and a notable enhancement in the performance of the estimation model. The findings of this study can be utilized in the development of food security policies and enhancement of agricultural product performance across various locations, ultimately increasing productivity.</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

Climate-responsive crop forecasting: an EEMD-LSTM fusion approach for improved strategic crop yield simulation

  • 期刊论文

  • 发表日期：

    2025年12月26日

查看全文
  • Seyed Babak Haji Seyed Asadollah,
  • Yusef Kheyruri,
  • Ahmad Sharafati,
  • Asaad Shakir Hameed

摘要

Predicting crop yield is complex due to its dependence on multiple meteorological variables. Remote sensing (RS) has become the prime source of climate parameters due to detailed spatial and temporal resolutions; however, its product needs further quality enhancement due to associated errors. The primary aim of this study is to incorporate the EEMD, as a signal modification technique, with the LSTM model, aiming to anticipate crop yield of four strategic crops, namely barley, lentils, pea, and wheat in all provinces of Iran. The annual crop yield of these crops was extracted from Iran’s Ministry of Agriculture data over 15 years, spanning from 2005 to 2020. In the context of EEMD-LSTM prediction, four influential climate parameters, including minimum and maximum temperature, precipitation, and the SPEI, were considered as the input variables. The analysis shows that applying EEMD generally enhances the predictive performance of the LSTM model for most agricultural products. For barley, lentils, and peas, EEMD substantially improves accuracy compared with the baseline model. Although its impact on wheat is not statistically significant, EEMD still reduces RMSE by approximately 42%. For lentils, the method yields notable improvements, with reductions of 15.7, 13.8, and 8% in MAE, RMSE, and PCC, respectively. Additionally, when noisy wheat data are removed, the error distribution slightly increases, indicating that wheat exhibits the least improvement among all evaluated crops. The spatial assessment reveals clear geographic differences in the effectiveness of EEMD. The method substantially reduces prediction errors for barley in the northwestern region. A similar, though smaller, improvement is observed for lentils in the same area. Conversely, EEMD increases prediction errors for chickpea production in both the northwest and eastern regions, highlighting strong spatial variability in the model’s performance across the study area. The utilization of EEMD, as a noise removal tool, evidently leads to a reduction in model errors and a notable enhancement in the performance of the estimation model. The findings of this study can be utilized in the development of food security policies and enhancement of agricultural product performance across various locations, ultimately increasing productivity.