Modeling and monitoring cotton production using remote sensing techniques and machine learning: a case study of Punjab, Pakistan.
Saved in:
| Title: | Modeling and monitoring cotton production using remote sensing techniques and machine learning: a case study of Punjab, Pakistan. |
|---|---|
| Authors: | Hasan, Sher Shah1 (AUTHOR) shershah538@gmail.com, Goheer, Muhammad Arif1 (AUTHOR) arifgoheer@gmail.com, Uzair, Muhammad2 (AUTHOR) uzairkhokhar005@gmail.com, Fatima, Saba2 (AUTHOR) sabafatima1101@gmail.com |
| Source: | Environment, Development & Sustainability. Apr2026, Vol. 28 Issue 4, p8659-8680. 22p. |
| Subject Terms: | *Remote sensing, *Machine learning, *Agricultural forecasts, *Linear statistical models, *Cotton trade, *Vegetation monitoring, *Agricultural climatology |
| Geographic Terms: | Pakistan, Punjab (Pakistan) |
| Abstract: | Agriculture is the backbone of Pakistan's economy and makes up 24 percent of the national GDP and half the labor force. This makes crop estimation studies extremely vital for a country's economic growth and food security. Cotton is one of the most important cash crops in Pakistan contributing 2.4% to the total value addition in agriculture. Remote Sensing (RS) and Geographic Information Systems (GIS) techniques can be used to effectively estimate crop yields even before harvesting. The objective of this study was to utilize RS/GIS, and machine learning to create a model for predicting cotton production; as well as identifying the impacts of climate-related factors on the growth and yield of cotton. Data from MODIS product MOD13A1, with a 16-day temporal resolution from 2011 to 2021 was used to calculate eleven vegetation indices in cotton-dominated districts of Punjab. These indices, along with rainfall data, temperature data, and historical yield data served as input to the machine learning models. Automatic Linear Modeling (ALM) and Artificial Neural Networks (ANN) were used to forecast the yields. The study also created a correlation between climate factors (rainfall and temperature) and cotton seasonal production. Pearson correlation coefficient of − 0.319 indicated a significant influence of maximum temperature on observed yields, while the Automatic Linear Modeling showed both maximum temperature and participation as a predictor for yield. These results underscore the vulnerability of cotton to climate change, proving cotton's sensitivity to temperature and rainfall. Comparison of both models based on their predictive yield placed ALM model's accuracy above ANN's at 44.1%, providing insights into the effectiveness of traditional linear modeling versus neural network approaches in predicting cotton yields. [ABSTRACT FROM AUTHOR] |
| Database: | Energy & Power Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Agriculture is the backbone of Pakistan's economy and makes up 24 percent of the national GDP and half the labor force. This makes crop estimation studies extremely vital for a country's economic growth and food security. Cotton is one of the most important cash crops in Pakistan contributing 2.4% to the total value addition in agriculture. Remote Sensing (RS) and Geographic Information Systems (GIS) techniques can be used to effectively estimate crop yields even before harvesting. The objective of this study was to utilize RS/GIS, and machine learning to create a model for predicting cotton production; as well as identifying the impacts of climate-related factors on the growth and yield of cotton. Data from MODIS product MOD13A1, with a 16-day temporal resolution from 2011 to 2021 was used to calculate eleven vegetation indices in cotton-dominated districts of Punjab. These indices, along with rainfall data, temperature data, and historical yield data served as input to the machine learning models. Automatic Linear Modeling (ALM) and Artificial Neural Networks (ANN) were used to forecast the yields. The study also created a correlation between climate factors (rainfall and temperature) and cotton seasonal production. Pearson correlation coefficient of − 0.319 indicated a significant influence of maximum temperature on observed yields, while the Automatic Linear Modeling showed both maximum temperature and participation as a predictor for yield. These results underscore the vulnerability of cotton to climate change, proving cotton's sensitivity to temperature and rainfall. Comparison of both models based on their predictive yield placed ALM model's accuracy above ANN's at 44.1%, providing insights into the effectiveness of traditional linear modeling versus neural network approaches in predicting cotton yields. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 1387585X |
| DOI: | 10.1007/s10668-024-05331-9 |
