Mapping cashew monocultures in the sub-districts of Sawantwadi and Dodamarg on Google Earth Engine
Due to the high potential of cashew as a cash group, global agricultural land devoted to cashew has increased rapidly over the past decade. To make space for these plantations, natural landscapes such as forests are cleared. One ecologically sensitive area in which cashew plantations may be con...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/156733 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Due to the high potential of cashew as a cash group, global agricultural land devoted to cashew
has increased rapidly over the past decade. To make space for these plantations, natural
landscapes such as forests are cleared. One ecologically sensitive area in which cashew
plantations may be contributing substantially to deforestation trends is the sub-districts of
Sawantwadi and Dodamarg, located within the Western Ghats belt in India. To monitor long term changes in the landscape attributed to cashew plantations, we planned to use remote
sensing data to create a map which can accurately map cashew monocultures in our study site,
for the year 2010. The map can then help researchers and policymakers understand the long term trends in the area devoted to cashew cultivation and determine the impact cashew
plantations have on forest cover in the region. We created 9 different maps, with 3 classification
algorithms (RF, CART, and SVM) and 3 different datasets (optical data, radar data, and a
combination of both), to determine which parameters can create the map that can most
accurately identify cashew monocultures. We determined that a model with RF as the
classification algorithm and the combined optical and radar dataset as the input data created
the map with the greatest overall accuracy of 97%. Therefore, we established that cashew
plantations spanned an area of 40 664 hectares in the region constituting 29% of the total
region. The methodology in this study can be replicated in future remote sensing studies which
may create similar maps for different time periods, thus making it possible for scientists and
policy-makers to accurately monitor land cover changes in the region. |
|---|