Mapping of rice farms using sentinel data and machine learning in Andhra Pradesh, India | |
| Author | Chiluvoori, Shivaji Raju |
| Call Number | AIT RSPR no.RS-22-04 |
| Subject(s) | Crops--India--Remote sensing Artificial satellites in agriculture--India |
| Note | A research submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Remote Sensing and Geographic Information Systems |
| Publisher | Asian Institute of Technology |
| Abstract | Mapping the Paddy rice is a major staple meal for billions of people worldwide. Mapping the regional distribution of rice and estimating yields are critical for global food policies. Remote sensing and GIS technology have been widely developed for agricultural cultivation monitoring and management throughout the last three decades. This research explores the use of multi-temporal C-band SAR (Sentinel-1) and optical MSI (Sentinel-2) satellite imagery datasets to map paddy rice area in the West Godavari district of Andhra Pradesh, India during the cultivation season of Rabi from 2017 till 2021. Along with the optical imagery and pixel-based technique, we also developed SAR dataset. Maps of paddy rice acreage were produced using the Random Forest classifier model. The SAR imagery classified map has performed to be reliable to the optical MSI dataset by the geographic distribution and accuracy assessment of paddy rice classification maps. In principle, the produced paddy acreage maps were quite reliable and thus, the accuracy of predicted paddy yields. The findings demonstrate that the suggested technique has a high overall accuracy assessment of more than 90% and a kappa coefficient statistic of more than 0.90. The method evaluated paddy rice growth phases and proved that the majority of paddy rice cultivation areas in Andhra Pradesh region of India has multiple seasonal cropping patterns, establishing paddy rice as a principal crop in agriculture. Notably, the anticipated area was quite similar to the provided data at the district level. It may be concluded that remote sensing and GIS satellite imagery datasets performed significantly and can effectively utilize the traditional field survey data collection approach for paddy rice area mapping. The random forest classification developed model has the capability of extracting paddy rice signatures from SAR satellite datasets and provide a credible estimate of paddy area. The selected technique, which employs basic statistical regression yield models, might be useful to policymakers in delivering timely yield information. |
| Year | 2022 |
| Type | Research Study Project Report (RSPR) |
| School | School of Engineering and Technology |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Remote Sensing and Geographic Information Systems (RS) |
| Chairperson(s) | Sarawut Ninsawat |
| Examination Committee(s) | Tripathi, Nitin K.;Mozumder, Chitrini |
| Scholarship Donor(s) | AIT Fellowship |
| Degree | Research Studies Project Report (M.Eng.) - Asian Institute of Technology, 2022 |