Mapping of oil palm land cover using integration of cloud computing, machine learning and big data

Oil palm is one of the agricultural crops that produces high amount of biomass, in which contributes to support the Sustainable Development Goals (SDGs). Furthermore, oil palm is a climate-friendly product that can generate energy in a more efficient way than using harmful eleme...

Full description

Saved in:
Bibliographic Details
Main Author: Shaharum, Nur Shafira Nisa
Format: Thesis
Language:English
Published: 2019
Subjects:
Online Access:http://psasir.upm.edu.my/id/eprint/84394/1/FK%202019%20145%20-%20ir.pdf
http://psasir.upm.edu.my/id/eprint/84394/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Oil palm is one of the agricultural crops that produces high amount of biomass, in which contributes to support the Sustainable Development Goals (SDGs). Furthermore, oil palm is a climate-friendly product that can generate energy in a more efficient way than using harmful element such as fossil fuel. However, the expansion of oil palm plantation has been recognised as a threat to the wildlife species and had caused massive amount of deforestations. A tropical country with humid weather, Malaysia has been listed as one of the top three countries with the highest rate of deforestation in the world. Obtaining information of oil palm plantation over a large area will be very intensive, costly and time consuming. Thus, this study had utilised cloud-based platforms, namely Google Earth Engine (GEE) and Remote Ecosystem Monitoring Assessment Pipeline (REMAP) to map the oil palm areas. Random Forest (RF) machine learning algorithm was utilised to produce and classify the land cover maps covering the Peninsular Malaysia. By using Landsat data obtained in Period 1 (1999 – 2003) and Period 2 (2014 – 2017), both cloud-based platforms were able to produce the land cover maps of Peninsular Malaysia. The overall accuracies produced by the GEE and REMAP for Period 1 data were 78.60% and 79.52% respectively. Meanwhile, the overall accuracies produced for Period 2 data were 79.77% and 80.00% for GEE and REMAP respectively. The changes of oil palm plantations noted from Period 1 to Period 2 using both cloud-based platforms were assessed, and the results showed oil palm plantation in Peninsular Malaysia is at sustainable state. For the first time, cloud-based platforms such as REMAP and GEE were being assessed for monitoring the changes to oil palm land cover in Peninsular Malaysia. Furthermore, the utilisation of REMAP and GEE were implemented to validate each other and to see the consistency of the results produced. This is a new paradigm shift from the normal approach utilising desktop-based big archives data analysis over huge areas that consumes tremendous amount of time and computing resources. In conclusion, both GEE and REMAP were able to produce the oil palm land cover maps and the changes of the oil palm can be analysed. In the future, the produced oil palm land cover maps of Peninsular Malaysia can be integrated with other ancillary data in a Geographic Information System (GIS) which later can assist the authorities in producing better decision-making and action plans.