Land-use land-cover dynamics of Hari-Ke-Pattan wetland using machine learning algorithm, remote sensing and GIS techniques in google earth engine
Wetlands are a dynamic ecosystem hosting diverse species of plants and animals. Wetlands are being altered because of environmental degradation and climatic variables. However, anthropogenic activities are the reason for disturbing the ecological harmony of wetlands and altering them for their own...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Action for Sustainable Efficacious Development and Awareness
2025-05-01
|
| Series: | Environment Conservation Journal |
| Subjects: | |
| Online Access: | https://journal.environcj.in/index.php/ecj/article/view/3072 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Wetlands are a dynamic ecosystem hosting diverse species of plants and animals. Wetlands are being altered because of environmental degradation and climatic variables. However, anthropogenic activities are the reason for disturbing the ecological harmony of wetlands and altering them for their own use. Hari-ke-Pattan wetland is one such example, situated at the border of Tarn Taran and Ferozepur districts of Punjab. It is assigned as a Ramsar site of the state. This study aims to understand the Land Use Land Cover (LULC) dynamics of Harike wetland from 2001 to 2021. It provides an insight into LULC changes that have happened within 5 kilometers of the study area, applying techniques of remote sensing, GIS, and machine learning. Landsat 5 TM for 2001 and Landsat 8 OLI/TIRS for 2021 were utilized within the data catalog of Google Earth Engine for analyzing Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), and LULC changes. To calculate NDVI, near-infrared and red bands were used; for NDWI, Gao’s formula of near-infrared (NIR) and short-wave infrared bands (SWIR) was used. JavaScript was used in the Earth Engine code editor to calculate these indices. The random forest classifier was utilized for the LULC classification. Statistics of LULC calculations were also done in the code editor of Earth Engine. QGIS 3.34.6 helped in mapmaking. Accuracy assessment of the classification showed an accuracy of 91% and 89% for the 2001 and 2021 LULC classifications, respectively. Kappa coefficient values were 0.89 and 0.86 for the 2001 and 2021 classifications, respectively. Analysis highlights that agricultural areas, built-up areas, and barren land registered an increase in their extent, whereas wetlands, vegetation, and aquatic vegetation registered a significant decrease in their areal extent. NDVI and NDWI values significantly decreased, indicating that vegetation and water content in the study have undergone negative changes.
|
|---|---|
| ISSN: | 0972-3099 2278-5124 |