January 1, 2025International Journal of Multidisciplinary Research and Growth Evaluation0 citations

Reconstruction of MODIS EVI Time Series to Map Maize Cropping Patterns in Sloping Areas: A Case Study in Son La of Vietnam

PHPham Viet HoangViet Nam University Of Traditional Medicine DHDinh Thai HoangUniversity of Technology Sydney NLNguyen Van LocVietnam National University of Agriculture

Key Points

The model mapped maize cropping patterns with an overall accuracy of 81.6%, demonstrating the effectiveness of remote sensing methods.
Using Savitzky-Golay and Support Vector Machine algorithms, the study detected fluctuations in maize acreages over 15 years (2003-2018).
Estimated maize acreages showed a strong correlation with official data, with a reliability metric of r = 0.75.
Further improvements in mapping accuracy could be achieved with high-resolution remote sensing data and better training samples.

Abstract

Mapping maize cropping patterns and acreage is important to many stakeholders in Vietnam where official statistical data is not up-to-date. This study demonstrates the use of Savitzky - Golay and Support Vector Machine algorithms on a time series of Moderate Resolution Imaging Spectroradiometer Enhanced Vegetation Index data (2003 - 2018) to detect and map the maize cropping patterns on the sloping area of Son La province of Vietnam. The method was able to map the spatial distribution of maize cropping patterns in areas where cultivated acreages were highly fluctuated recently, with an overall accuracy of 81.6%. Districts' estimated maize acreages also show a good agreement with the official data at r = 0.75. Our findings suggest that a further improvement of the maize cropping pattern map can be made when higher spatial resolution remote sensing data are available and better presentative training samples for the whole region could be considered.

KI fragen

Bookmark

View Full Paper