Scholarlink

<p>Accurate crop yield estimation is crucial for assessing crop insurance claims, financial planning, storage and transportation logistics, and supporting policy decisions related to food trade and minimum support price (MSP) fixation for Rabi pulses. This study focuses on predicting the yields of key Rabi pulse crops viz. gram and field pea in the drought-prone Bundelkhand region of Uttar Pradesh, India, using historical remote sensing-based multispectral and meteorological data from 2001 to 2017. The results reveal that crop classification using the Random Forest classifier achieved an overall accuracy of 82.83% and a kappa coefficient of 0.76. Four crop yield prediction models were developed and validated for both Rabi pulses crops: (i) Weighted Statistical Yield Model, (ii) Random Forest Regression, (iii) Stepwise Multiple Linear Regression, and (iv) Phenological Metrics-based Yield Model. Among these, the Weighted Statistical Yield Model, relying solely on weather variables and their derived indices, outperformed others with the lowest RMSE of 0.040 t/ha for gram and 0.037 t/ha for field pea. The findings demonstrate that remote sensing data, coupled with limited ground-truth and ancillary datasets (crop yield data from MoA&amp;FW DAC yield records), can effectively substitute field-based observations, enabling reliable regional-scale crop yield estimation.</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

Development of Yield Modelling Techniques for Rabi Pulses using Geospatial Techniques

  • 期刊论文

  • 发表日期：

    2025年6月17日

查看全文
  • Abhishek Danodia,
  • T. Udaya Bhaskar Jayanth,
  • N. R. Patel,
  • Dipanwita Haldar

摘要

Accurate crop yield estimation is crucial for assessing crop insurance claims, financial planning, storage and transportation logistics, and supporting policy decisions related to food trade and minimum support price (MSP) fixation for Rabi pulses. This study focuses on predicting the yields of key Rabi pulse crops viz. gram and field pea in the drought-prone Bundelkhand region of Uttar Pradesh, India, using historical remote sensing-based multispectral and meteorological data from 2001 to 2017. The results reveal that crop classification using the Random Forest classifier achieved an overall accuracy of 82.83% and a kappa coefficient of 0.76. Four crop yield prediction models were developed and validated for both Rabi pulses crops: (i) Weighted Statistical Yield Model, (ii) Random Forest Regression, (iii) Stepwise Multiple Linear Regression, and (iv) Phenological Metrics-based Yield Model. Among these, the Weighted Statistical Yield Model, relying solely on weather variables and their derived indices, outperformed others with the lowest RMSE of 0.040 t/ha for gram and 0.037 t/ha for field pea. The findings demonstrate that remote sensing data, coupled with limited ground-truth and ancillary datasets (crop yield data from MoA&FW DAC yield records), can effectively substitute field-based observations, enabling reliable regional-scale crop yield estimation.