Multi-Harvesting Rice Monitoring Using Optical and Synthetic Aperture Radar Satellite Data

0
142

Multi-Harvesting Rice Monitoring Using Optical and Synthetic Aperture Radar Satellite Data

Authors: Tahmid Anam Chowdhury, Nusrat Jahan Ety, Warefta E Murshed, Abed Chaudhury

Author for correspondence: Abed CHAUDHURY
“kanihati@gmail.com”

Project: Krishan Foundation, Kanihati, Hafiza Khatun Memorial Trust.
Village: Hazipur, Upazila: Kulaura, District: Moulvibazar, Bangladesh

 

Abstract

Staple to about half of the world’s population, rice is the most vital and nutrient primary food crop in the world. Multi-harvesting rice variety is one such way of cultivation that can open a new dimension to food security and economic benefit to farmers. In this method, the rice plants are not uprooted from the field so that they can grow multiple times. Geographic Information Systems (GIS) and Remote Sensing have proven their efficiency in monitoring rice using optical and Synthetic Aperture Radar (SAR) satellite datasets. Time series of Normalized Difference Vegetation Index (NDVI) from optical remote sensing and VV polarisation from SAR remote sensing has been applied to monitor the variety of multi-harvesting Panchavrihi (PV). Optical satellite data includes Sentinel 2 and Landsat 9 satellites, whereas SAR data includes Sentinel 1 satellite. Time series curve changes with PV rice growth stages. The results exhibit during the leaf growing stage NDVI and VV polarisation backscatter show low values (0.3 to 0.4 for NDVI, -12 to -13 dB for SAR). There is an increase in both parameters towards the milk stage. NDVI and VV backscatter peak around (0.8 to 0.9) and (-6 to -7 dB), respectively, until the milk stage. Both parameters begin to fall towards the ripening stage. This scenario was repeated until four multi-time harvests. Meanwhile, single-growing varieties from farmer's plot were also studied by NDVI and VV backscatter as control variety. Thus, in this study, NDVI and VV polarisation give efficient interpretation to monitor the multi-harvest PV rice variety. The obtained results were verified by fieldwork data during the satellite data acquisition time throughout the experiment timeline from the selected rice fields.

Introduction

Agriculture is the most essential and vital sector to ensure food security. Rice (Oryza sativa L.) is one of the leading food crops crucial for the world's food supply (Kennedy, 2002). Rice production has increased significantly in the lastseveral decades with limited cultivable land (FAO, 2009). 21% of global human per capita energy is provided by rice (Kennedy, 2002). In 2019, global rice production was close to 755 million tons, covering an area of 162 million hectares (Tayyib, 2021). By the end of 2050, the world population will be almost 10 billion (Nations, 2019), which indicates that rice production must thrive to ensure food security and to achieve Sustainable Development Goal (SDG) 2. The development of different satellites has opened a new dimension of Spatiotemporal study, including land cover, natural disasters, meteorology, hydrology, ecology, crops etc. Satellite images are being efficiently used for crop mapping, monitoring and yield prediction. Satellite-based rice study includes MODIS (Moderate Resolution Imaging Spectroradiometer), Landsat and Sentinel images (Zhao et al., 2021). Normalized Difference Vegetation Index (NDVI) is a popular index derived from the optical satellite images. It is widely used to map, monitor, and predict rice (Franch et al., 2021; Misra et al., 2020; Soh et al., 2022; Zhao et al., 2021). However, the use of optical satellite images still has some challenges in terms of time series development, including atmospheric correction errors, decreased reflectance by shadows and cloud presence (Soriano-González et al., 2022). Due to the limitations of optical remote sensing in cloudy weather, Synthetic
Aperture Radar (SAR) remote sensing in the microwave region with sensitivity to various geometrical and biophysical parameters in all weather conditions has gained popularity for rice monitoring (Dave et al., 2017; Dineshkumar et al., 2019; Haldar et al., 2018; Harfenmeister et al., 2021; Ramathilagam & Haldar, 2022; Sharifi & Hosseingholizadeh, 2020). Sentinel 1 derived different polarisations (e.g. VV, VH) have been proven effective to map and monitor crop growth stages in different places of the world (Clauss et al., 2018; Dineshkumar et al., 2019; Inoue et al., 2020; Khabbazan et al., 2019; Kobayashi & Ide, 2022; Salma & Dodamani, 2021; Sharifi & Hosseingholizadeh, 2020). Integrated use of optical and synthetic aperture radar (SAR) satellite images is effective for precise rice monitoring (Inoue et al., 2020; Soh et al., 2022; Xiao et al., 2021; Zhao et al., 2021). Sentinel-1 SAR and Sentinel-2 Multi-Spectral Instrument (MSI) optical satellites are better optionsfor high resolution and frequent revisit characteristics. Rice is seasonal. There are three types of rice typically cultivated in Bangladesh- Boro, Aus and Aman. Rice is planted in a particular season and harvested. Then the land is ploughed again and another seasonal rice is planted. Majority land is cultivated with Boro and Aman. In order to utilize land in a sustainable way, multi harvest rice variety was introduced by Dr. Abed Chaudhury. These multiharvest varieties called Panchavrihi (PV) or 5 harvest upon planting in Boro can be harvested in Boro, Aus and Aman from the same plants. This paper describes satellite based monitoring of crop productivity of PV comparing and contrasting it with seasonal mono harvest varieties

Materials and Methods
Study Area

In this research, experimental and random plots had been delineated by field work. The GPS point of each corner of all plots has been recorded using the GPS logger app. Those GPS points have been validated using Google Earth (Figure 18.1). Periodic field works had been conducted to record crop phenotyping data, like- growth stage, stem size, grain density etc.

Figure 18.1: GPS Points collected for Boundary Delineation

The studied area is located in Kanihati village of Hajipur union, Kulaura Upazila,Moulvibazar district of Bangladesh. It extends from 91 •56' 15" E to 91 •56' 33" E and from 24 •24' 48" N to 24 •25' 3" N. It receives an average rainfall of 3000-3500 mm and a temperature of 24.5 ◦ -25 ◦ C annually (Shahid, 2010). The soil of this area is Noncalcareous Grey Floodplain (Brammer, 1996). Physio-graphically it is situated in the East Surma-Kusiyara floodplain, and its sediment is alluvial. Valley alluvium and colluvium are deposited on surface geology (Alam et al., 1990).PV rice variety was cultivated in big and small plots here. There were five big plots with PV sub-varieties, named Kanihati 4, Kanihati 5, Kanihati 6, Kanhihati 6 and Kanihati 11. Selected PV sub-varieties were being cultivated in small random blocks, named as Kanihati 1, Kanihati 3, Kanihati 7, Kanihati 12, Kanihati 15 and Kanihati 16. In order to compare with a control variety, one plot was randomly chosen, named as farmer's variety (Figure 18.2).

Figure 18.2: Location of Study Area (Kanihati village, Kulaura Upazila, Bangladesh)

Optical Datasets

Sentinel 2 MSI is advantageous for studying rice among different optical images because of its multispectral bands, high resolution, and frequent revisit (Franch et al., 2021; Misra et al., 2020; Zhao et al., 2021). It is positioned aboard two orbiting satellites (Sentinel-2 A/B), which provide images every five days (Drusch et al., 2012). Sentinel 2 has 13 spectral bands with a resolution from 10 to 60 meters (Szantoi & Strobl, 2019). Sentinel 2 images have been used on specific dates, where the study area was cloud coverage free. Bottom of Atmosphere (BOA) reflectance is used from Sentinel 2 images to study to compute the vegetation indices. Landsat 9 Operational Land Imager (OLI) image has also been used as a substitute for missing data of Sentinel 2 MSI. Landsat 9 has 8 spectral bands (30m) and 1 panchromatic band (15m) (Markham et al., 2020). Landsat Level-2 images have been used for surface reflectance. The pan-sharpening method is applied for Landsat spectral images, where the high-resolution panchromatic band (15m) sharpens the resolution of other spectral bands from 30m to 15m using the Gram-Schmidt pan-sharpening algorithm (Laben & Brower, 1998). Table 18.1 shows the dates of retrieved datasets of Sentinel 2 and Landsat 9.

Table 18.1: Optical Image Acquisition Date

Satellite Date
Sentinel 2  31-Jan, 15-Feb, 20-Feb, 02-Mar, 12-Mar, 17-Mar, 22-Mar, 06-Apr, 21-Apr, 01-May, 10-Jul, 15-Jul, 20-Jul, 14-Aug, 08-Sep, 28-Sep, 18-Oct, 28-Oct, 02-Nov, 07-Nov, 12-Nov, 17-Nov, 22-Nov, 27-Nov, 02-Dec in 2022
Landsat 9 08-May, 24-May, 27-Jul in 2022

NDVI was applied to understand vegetation health and land use remotely (Tarpley et al., 1984). NDVI requires reflectance of near-infrared (NIR) and red bands. NDVI formula is given below (Chen et al., 2006)-

NDVI=NIR-RedNIR+Red

The NDVI has a value between -1 and +1. Values ranging from -1 to 0 denote the presence of dead plants or inorganic objects such as stones, roads, and houses. For live plants, NDVI values range from 0 to 1, with 1 being the healthiest and 0 being the least healthy.

Rice growth consists of three phases vegetative, reproductive, and ripening. During the vegetative and reproductive periods, rice proliferates with canopy closure. As a result, the reflectance of leaf chlorophyll decreases, and soil signals are reduced. NDVI values rapidly escalate throughout this period, reaching a peak after the reproductive phase, when the crop enters the mature stage (Soh et al., 2022)(Figure 18.3). During the mature stage, increased carotenoid content significantly reduces chlorophyll content (Ni et al., 2021). This noticeable decline in NDVI readings shows that the rice plant has reached maturity. This time arises in the final month of the rice growth cycle (Soh et al., 2022)(Figure 18.3).

Figure 18.3: Phases of Rice Growth (Dineshkumar et al., 2019; Kuenzer & Knauer, 2013; Mosleh et al., 2015)

Sentinel-1A (SAR) Data and Preprocessing

The presence of clouds in a tropical region makes obtaining the surface image difficult, the Sentinel-1 SAR images were used in tandem. The SAR wavelength is longer than the wavelength of particles in a cloud, such as droplets, the signal flowing through a cloud is largely unaffected by refraction at the media’s borders. SAR sends a microwave signal to a target, which reflects a portion of the signal to the radar antenna. This reflection is called a Backscatter. The target’s many qualities influence how much it backscatters the signal (Leung et al., 1994). Thermal noise removal, radiometric correction, terrain correction, and speckle noise reduction are all required corrections for the raw SAR imagery.

Sentinel-1 SAR imagery is helpful in studying various crops, particularly rice (Dineshkumar et al., 2019; Khabbazan et al., 2019; Kobayashi & Ide, 2022; Salma & Dodamani, 2021; Singha et al., 2019; Yang et al., 2021). It’s made up of two polar-orbiting satellites (Sentinel-1 A/B) that cover day and night and provide photographs every six days. The Sentinel-1 observation system employs the broad interferometric mode (IW). This mode generates dual-polarised images, i.e., vertical transmit/receive (VV) and vertical transmit/horizontal receive (VH), with a spatial resolution of 10 m (Borgogno-Mondino et al., 2020; Meng et al., 2013).

The Level-1 Ground Range Detected (GRD) was gathered in the ascending orbit products for this study. The Sentinel-1 imagery has performed thermal noise removal, radiometric calibration, and terrain correction (orthorectification). No geometric correction was needed as the study region is in a flat land, and there are no geometric or topography disturbances.

Speckle noise filtering is used to distinguish characteristics of different/heterogeneous land cover over time (Clauss et al., 2018; Minasny et al., 2019). But to analyse homogenous land cover (i.e. crop land, barren land), speckle noise filtering is not mandatory (Kobayashi & Ide, 2022). The only preprocessing of SAR imagery is conversion to a backscattering coefficient (σ0) in decibels (dB).

Table 18.2. Characteristics of Sentinel-1 images (Panetti et al., 2014)

Band Polarisation Mode Wavelength Pixel Size Product Level Product Type
C band Vertical transmit and vertical receive (VV) Interferometric
Wide (IW)
5.6 cm 10 m Level-1 Ground Range
Detected
Vertical transmit and horizontal receive (VH) 5.6 cm 10 m

As Sentinel-1 SAR functions entirely differently than optical images, it has specific parameters which are found to have better accuracy in studying rice. Dineshkumar et al. (2019) reported VV polarisation gives a better interpretation than VH polarisation. Table 18.3 shows the dates of the SAR datasets used in this study.

Table 18.3: SAR Imagery Acquisition Date

Satellite Date
Sentinel 1 10-Feb, 22-Feb, 06-Mar, 18-Mar, 30-Mar, 11-Apr, 23-Apr, 05-May, 17-May, 29-May, 10-Jun, 22-Jun, 10-Jul, 28-Jul, 09-Aug, 21-Aug, 02-Sep, 14-Sep, 26-Sep, 08-Oct, 20-Oct, 01-Nov, 07-Nov in 2022

Dineshkumar et al. (2019) reported that VV backscattering coefficients progressively intensify during the crop vegetative phase or growing period. A peak value of VV backscatter is attained between the beginning of the ripening phase and the end of the reproductive phase,  after which the VV backscatter coefficients again start dipping through the mature stage.

Figure 18.4: Temperature & Precipitation from January to October 2022

Meteorological Data

The rate of crop growth is affected by precipitation and temperature (Seguin et al., 1993). For this study, the precipitation and temperature data were collected from January 2022 to November 2022 (Figure 18.4). Temperature and precipitation have a significant role in influencing the wetness of the surface, which affects the Dielectric constant on the surface. Dielectric constant has also an impact on SAR backscatter.

Figure 18.5: Phenological Phases of the Experimental Rice Variety

Results

Temporal Profiles of NDVI

A multi-harvesting rice variety was cultivated in the study area. As this rice gives multiple harvests, its satellite-based observation curve (both for optical and SAR data) shows a continuous line. Figure 18.5 shows the continuous phenological phases of the experimental rice variety.

The multi-time growth phenology of the rice variety is illustrated on NDVI and SAR graphs. The mean temporal profiles of Sentinel-2 derived NDVI for rice plots were generated during the growing season (Figure 18.6). The profiles include, on each date, the average NDVI of the experimental plots over a control rice variety.

Figure 18.6: Mean temporal profiles of Rice NDVI 

NDVI time series has been plotted over PV and Farmer’s varieties with their different phases. Each PV variety is compared to the farmer’s variety plot by NDVI curve (Figure 18.6). Regarding the PV variety, the vegetative phase starts with the transplant. It has taken place in early February, after raining, to ensure sufficient soil moisture for tillering. The panicle initiation is started at the end of February, which corresponds to the first reproductive phase of phenology (Figure 18.5, 18.6). From this stage, the NDVI values continued to increase, ensuring the development of the plant. Flowering was recorded after 75 days of transplant, the end of the first

reproductive phase. The milk stage initiated the first ripening phase, where NDVI value peaked for the first growth in April. NDVI value was considerably decreasing after the end of the milk stage, and it began to fall due to the ripening of rice grains until the end of the first ripening phase (mature stage). At the end of the first ripening phase, PV varieties had given the first harvest of Boro rice. In this first harvest, plant size was a maximum of 36 inches. At the same time, the random plot had also given random Boro rice (BR 48) harvest, where the NDVI of the farmer’s variety during the ripening phase was below PV varieties.

The first harvest came out in 89 days after the transplant. After the first harvest, second reproductive phase of rice was started from rice remnant. So NDVI curve began to rise. After the first harvest in early June, the flowering stage of the second reproductive stage occurred. At the end of the flowering stage, the second ripening phase started. NDVI value reached the maximum point in the flowering stage, and NDVI began to fall up to the second harvest. Rice ripened for the second harvest of Aus rice at the beginning of July. In the second harvest phase, PV varieties had a length below 36 inches. During this period, the random plot remained fallow. A fall in the NDVI curve is noticed in July, due to the presence of water in this plot.

Third-time growth initiated from the remnant of the second harvest. NDVI curve rose upward and the third ripening phase started by the milk stage at the beginning of September. NDVI curve began to fall at the end of the milk stage. The mature stage occurred at the end of the third ripening phase in early October. Aman rice is produced from the third harvest and the rice plant length is 42 inches during the third harvest. At the same time, the Random plot was ongoing Aman rice (Abed Dhan) cultivation, which had yet to be flowered. Non-ripened Aman rice (Abed Dhan) of the farmer’s plot was visible in the NDVI curve, where the value was higher than ripened PV varieties.

After the third harvest, the fourth growth occurred at the end of October. NDVI curve peaked during the milk stage in the middle of November. It started to decrease due to the fourth ripening stage. The fourth harvest had taken very little time than the early three harvests of PV. Meanwhile, Aman rice (Abed Dhan) from the farmer’s plot was also ripened (Figure 18.6, 18.7, 18.8).

Figure 18.7: Spatio-temporal Changes of Rice NDVI

Figure 18.8: Spatio-temporal Changes of Rice VV Backscatter

3.2 Temporal Profiles of SAR Backscattering in VV

VV polarisation backscatter of SAR is proven effective in studying rice phenological phases. This section evaluated the Sentinel-1 polarisations (VV) for detecting each phenological phase (Figure 18.9). The profiles include, on each date, the average SAR backscatter (σ◦) in VV polarisation of the Kanihati and random rice varieties.

VV backscatter is primarily influenced by changes in soil backscatter caused by surface roughness and dielectric constant. For the PV variety, the SAR backscatter (σ◦) in VV polarisation began to increase at the beginning of the vegetative phase until the flowering stage of the first ripening phase. At the end of the milk stage in late March, the VV backscatter begins to fall. After the end of the first ripening phase, the second reproductive phase initiated. VV backscatter began to increase from the third growth stage until the flowering stage of the second ripening phase. After the first harvest, VV backscatter rose again due to the second growth of the plant. As in previous phases, backscatter increased until the flowering stage and decreased until the last stage of the third ripening phase. 

In February, VV backscatter from random plot rice began to grow during the vegetative phase of Boro rice (BR 48) cultivation. It increased until the flowering stage and kept falling until the mature stage in April. Over the Aus rice season (from April to July), the farmer’s plot remained fallow, but growth in VV backscatter is noticed. It happened due to the development of grasses in the fallow plot, fed by sufficient precipitation (Figure 18.4). At the early August, Aman rice (Abed Dhan) is cultivated in the farmer’s plot. So VV backscatter rose until the reproductive phase in the farmer’s plot

Figure 18.9: Mean temporal profiles of VV backscatter from Sentinel-1 

Temporal Profiles of Random Blocks

A total of 16 rice sub-varieties were cultivated in 48 small plots. Each sub-variety was grown in three random plots. The mean temporal profile of each random block sub-variety is illustrated by both NDVI and VV backscatter (Figure 18.10, 18.11).

In the first harvest (Boro season)-specific sub-varieties of PV had an early harvest in opposition to Boro rice (BR 48) from the farmer’s plot (Figure 18.10, 18.11). During the second harvest period (Aus season), the farmer’s plot remained fallow. PV varieties had a slightly weak harvest in the second growth.

The third growth occurred after the second harvest during the Aman season. Abed dhan was cultivated in farmer’s plot during this period. Random Kanihati plots had a third early harvest of PV rice in October, whereas Abed Dhan from the farmer’s plot was in the flowering stage. PV varieties had fourth growth in late October. NDVI and VV backscatter peaked in the early November (Figure 18.10, 18.11). It began to decrease after the milk stage and ripened at the end of November with Abed Dhan from the farmer’s plot.

Figure 18.10: Mean temporal NDVI profiles of random Block Varieties

Figure 18.11: Mean temporal profiles of VV backscatter for random Blocks from Sentinel-1

Conclusions

For the study, the multi-harvesting PV variety was cultivated with other common rice varieties in larger plots and random blocks. Both optical and SAR datasets can monitor those varieties. Different rice varieties from the farmer’s plot were used as control varieties to compare experimental sub-varieties. A unique rice variety has

been created in Bangladesh using a mix of cloud computing and remote sensing indices. With the use of meteorological data and field survey data, it is discovered that the comprehensive time series of NDVI (optical dataset) and VV (SAR dataset) are able to correctly identify phenological phases, i.e., transplant, blooming, milk, and mature stages. Observations from NDVI and VV backscatter have shown that Kanihati sub-varieties had a strong harvest in the first time (Boro season), then a lower harvest in the second time (Aus season), double harvest (third and fourth time) in Aman season.On the contrary, the control plot (Farmer’s plot) had BR-48 cultivation in Boro season, kept fallow in Aus season, and cultivated Abed Dhan in the Aman season. NDVI is disturbed by cloud coverage to get continuous data. However, SAR is a specialised dataset that can overcome cloud coverage. NDVI curve acts differently in terms of multi-harvesting rice plots, due to the presence of green weeds or grasses at the bottom of stem. This transition phase is distinguishable by SAR. Multi-seasonal rice crop monitoring and remote sensing are now possible in rainy areas. This technical approach could be widely applied to monitoring multi-harvest rice. Many future crop monitoring studies, such as dynamic crop monitoring and physical condition assessment, will benefit from the improved insight of optical and SAR temporal information in real time under diverse agricultural practices and environmental conditions.

References

1) Alam, M., Hasan, A., Khan, M., & Whitney, J. (1990). Geological Map of Bangladesh.
Geological Survey of Bangladesh. US Geological Survey, Dhaka.

2) Borgogno-Mondino, E., de Palma, L., & Novello, V. (2020). Investigating Sentinel 2
Multispectral Imagery Efficiency in Describing Spectral Response of Vineyards
Covered with Plastic Sheets. Agronomy, 10(12), 1909.

3) Brammer, H. (1996). Geography of the Soils of Bangladesh. University Press.

4) Chen, X.-L., Zhao, H.-M., Li, P.-X., & Yin, Z.-Y. (2006). Remote sensing image-based
analysis of the relationship between urban heat island and land use/cover changes.
Remote sensing of environment, 104(2), 133-146.

5) Clauss, K., Ottinger, M., & Künzer, C. (2018). Mapping rice areas with Sentinel-1 time
series and superpixel segmentation. International journal of remote sensing, 39(5),
1399-1420.

6) Dave, V. A., Haldar, D., Dave, R., Misra, A., & Pandey, V. (2017). Cotton crop
biophysical parameter study using hybrid/compact polarimetric RISAT-1 SAR data.
Progress In Electromagnetics Research M, 57, 185-196.

7) Dineshkumar, C., Kumar, J. S., & Nitheshnirmal, S. (2019). Rice monitoring using
sentinel-1 data in the google earth engine platform. Multidisciplinary Digital
Publishing Institute Proceedings, 24(1), 4.

8) Drusch, M., Del Bello, U., Carlier, S., Colin, O., Fernandez, V., Gascon, F., Hoersch,
B., Isola, C., Laberinti, P., & Martimort, P. (2012). Sentinel-2: ESA's optical high-
resolution mission for GMES operational services. Remote sensing of environment,
120, 25-36.

9) FAO. (2009). La agricultura mundial en la perspectiva del año 2050. Cómo alimentar
al mundo en 2050.

10) Franch, B., Bautista, A. S., Fita, D., Rubio, C., Tarrazó-Serrano, D., Sánchez, A.,
Skakun, S., Vermote, E., Becker-Reshef, I., & Uris, A. (2021). Within-field rice yield
estimation based on Sentinel-2 satellite data. Remote Sensing, 13(20), 4095.

11) Haldar, D., Dave, R., & Dave, V. A. (2018). Evaluation of full-polarimetric parameters
for vegetation monitoring in rabi (winter) season. The Egyptian Journal of Remote
Sensing and Space Science, 21, S67-S73.

12) Harfenmeister, K., Itzerott, S., Weltzien, C., & Spengler, D. (2021). Agricultural
monitoring using polarimetric decomposition parameters of sentinel-1 data. Remote
Sensing, 13(4), 575.

13) Inoue, S., Ito, A., & Yonezawa, C. (2020). Mapping Paddy fields in Japan by using a
Sentinel-1 SAR time series supplemented by Sentinel-2 images on Google Earth
Engine. Remote Sensing, 12(10), 1622.

14) Kennedy, D. (2002). The importance of rice. In (Vol. 296, pp. 13-13): American
Association for the Advancement of Science.

15) Khabbazan, S., Vermunt, P., Steele-Dunne, S., Ratering Arntz, L., Marinetti, C., van
der Valk, D., Iannini, L., Molijn, R., Westerdijk, K., & van der Sande, C. (2019). Crop
monitoring using Sentinel-1 data: A case study from The Netherlands. Remote Sensing,
11(16), 1887.

16) Kobayashi, S., & Ide, H. (2022). Rice crop monitoring using Sentinel-1 SAR data: A
case study in Saku, Japan. Remote Sensing, 14(14), 3254.

17) Kuenzer, C., & Knauer, K. (2013). Remote sensing of rice crop areas. International
journal of remote sensing, 34(6), 2101-2139.

18) Laben, C., & Brower, B. (1998). Webster, both of NY process for enhancing the spatial
resolution of multispectral imagery using pan-sharpening. Eastman Kodak Company,
Rochester, NY.

19) Leung, K., Cuddy, D., & Chen, M. (1994). ASF SAR processing system overview.
Proceedings of IGARSS'94-1994 IEEE International Geoscience and Remote Sensing
Symposium,

20) Markham, B., Jenstrom, D., Sauer, B., Pszcolka, S., Dulski, V., Hair, J., McCorkel, J.,
Kvaran, G., Thome, K., & Montanaro, M. (2020). Landsat 9 mission update and status.
Earth Observing Systems XXV,

21) Meng, Q., Cooke, W. H., & Rodgers, J. (2013). Derivation of 16-day time-series NDVI
data for environmental studies using a data assimilation approach. GIScience & Remote
Sensing, 50(5), 500-514.

22) Minasny, B., Shah, R. M., Che Soh, N., Arif, C., & Indra Setiawan, B. (2019).
Automated near-real-time mapping and monitoring of rice extent, cropping patterns,
and growth stages in Southeast Asia using Sentinel-1 time series on a Google Earth
Engine platform. Remote Sensing, 11(14), 1666.

23) Misra, G., Cawkwell, F., & Wingler, A. (2020). Status of phenological research using
Sentinel-2 data: A review. Remote Sensing, 12(17), 2760.

24) Mosleh, M. K., Hassan, Q. K., & Chowdhury, E. H. (2015). Application of remote
sensors in mapping rice area and forecasting its production: A review. Sensors, 15(1),
769-791.

25) Nations, U. (2019). Growing at a slower pace, world population is expected to reach 9.7
billion in 2050 and could peak at nearly 11 billion around 2100 [online] UN DESA
United Nations Department of Economic and Social Affairs. In.

26) Ni, R., Tian, J., Li, X., Yin, D., Li, J., Gong, H., Zhang, J., Zhu, L., & Wu, D. (2021).
An enhanced pixel-based phenological feature for accurate paddy rice mapping with
Sentinel-2 imagery in Google Earth Engine. ISPRS Journal of Photogrammetry and
Remote Sensing, 178, 282-296.

27) Panetti, A., Rostan, F., L'Abbate, M., Bruno, C., Bauleo, A., Catalano, T., Cotogni, M.,
Galvagni, L., Pietropaolo, A., & Taini, G. (2014). Copernicus sentinel-1 satellite and C-
SAR instrument. 2014 IEEE Geoscience and Remote Sensing Symposium,

28) Ramathilagam, A. B., & Haldar, D. (2022). Evaluation of different machine learning
algorithms for pearl millet discrimination using multi-sensor SAR data. Geocarto
International, 37(17), 5116-5132.

29) Salma, S., & Dodamani, B. (2021). Temporal crop monitoring with sentinel-1 sar data.
Trends in Civil Engineering and Challenges for Sustainability, 621-632.

30) Seguin, B., Courault, D., & Guerif, M. (1993). Satellite thermal infrared data
applications in agricultural meteorology. Advances in Space Research, 13(5), 207-217.

31) Shahid, S. (2010). Recent trends in the climate of Bangladesh. Climate Research, 42(3),
185-193.

32) Sharifi, A., & Hosseingholizadeh, M. (2020). Application of sentinel-1 data to estimate
height and biomass of rice crop in Astaneh-ye Ashrafiyeh, Iran. Journal of the Indian
Society of Remote Sensing, 48(1), 11-19.

33) Singha, M., Dong, J., Zhang, G., & Xiao, X. (2019). High resolution paddy rice maps in
cloud-prone Bangladesh and Northeast India using Sentinel-1 data. Scientific data, 6(1),
1-10.

34) Soh, N. C., Shah, R. M., Giap, S. G. E., Setiawan, B. I., & Minasny, B. (2022). High-
Resolution Mapping of Paddy Rice Extent and Growth Stages across Peninsular
Malaysia Using a Fusion of Sentinel-1 and 2 Time Series Data in Google Earth Engine.
Remote Sensing, 14(8), 1875.

35) Soriano-González, J., Angelats, E., Martínez-Eixarch, M., & Alcaraz, C. (2022).
Monitoring rice crop and yield estimation with Sentinel-2 data. Field Crops Research,
281, 108507.

36) Szantoi, Z., & Strobl, P. (2019). Copernicus Sentinel-2 calibration and validation. In
(Vol. 52, pp. 253-255): Taylor & Francis.

37) Tarpley, J., Schneider, S., & Money, R. (1984). Global vegetation indices from the
NOAA-7 meteorological satellite. Journal of Climate and Applied Meteorology, 491-
494.

38) Tayyib, S. (2021). Crops and livestock products. Food and Agriculture Organization.
Retrieved 17th June, 2022 from https://www.fao.org/faostat/en/#data/QCL

39) Xiao, W., Xu, S., & He, T. (2021). Mapping paddy rice with sentinel-1/2 and
phenology-, object-based algorithm—A implementation in Hangjiahu plain in China
using gee platform. Remote Sensing, 13(5), 990.

40) Yang, L., Huang, R., Huang, J., Lin, T., Wang, L., Mijiti, R., Wei, P., Tang, C., Shao,
J., & Li, Q. (2021). Semantic Segmentation Based on Temporal Features: Learning of
Temporal–Spatial Information From Time-Series SAR Images for Paddy Rice
Mapping. IEEE Transactions on Geoscience and Remote Sensing, 60, 1-16.

41) Zhao, R., Li, Y., & Ma, M. (2021). Mapping paddy rice with satellite remote sensing: a
review. Sustainability, 13(2), 503.

LEAVE A REPLY

Please enter your comment!
Please enter your name here