Mapping rice cropping patterns using Sentinel-1 SAR multi-temporal imagery:A case study in the district of Indramayu, West Java.

Gabriel Yarlequé IpanaquéDepartment of International Development, Community and Environment

Clark University

Rice is an important crop for food security and it contributes to theUnited Nations Sustainable Development Goals. Stable and sustainablerice production requires information on where, when and how rice isgrown (Cuerdo et al., 2013). Rice can be grown more than once per yearin the same field, or in a system of alternating crops (FAO, 1996). In Asia,some areas switch between rice and pond aquaculture in the same year.Current monitoring systems estimate the area and production of rice.However, they do not report on the area of rice that is planted in differentcropping patterns. A cropping pattern refers to the seasonal sequenceand spatial arrangement of crops in a particular area for a given year(FAO, 1996). This research explores the use of high spatial resolutionSentinel-1 multi-temporal imagery to understand where and whatcropping patterns exist in the district of Indramayu, located in the islandof Java, Indonesia.

There are two objectives to this research:(1) Explore the temporal signatures of rice-based fields using Sentinel-1

multitemporal imagery.(2) Identify and classify different rice-based cropping patterns in the

study area.

Introduction

Study Area: Indramayu, West Java

Image Processing and Analysis

References

Main Findings

• VH has a larger dynamic range of backscatter than VV through the rice growth cycle.

• The Support Vector Machine classifier was able to identify rice and nonrice areas for the images corresponding to the beginning of each cropping cycle.

• Two cropping patterns were identified:o Rice – Rice:

- Rice 1: Wet Season (Rainfed) : November/December to April. Flooding of fields starts in November in the south.

- Rice 2: Dry Season (Irrigated): May to August. Floding of fields starts in March at some locations

o Rice – Fallow- Areas in the south only produce Rice 1. - Areas in the northwest and southeast only

produce rice 2

Figure 1: Indonesia is the third largest producer of rice in the world (IRRI,2009). Most of rice production takes place in the island of Java, especiallyin West and Central Java. The district of Indramayu, is the main riceproducing area in West Java. It covers 2,041 km2. In September 2018, riceproduction in Indramayu reached 1,392 thousands tones (OfficialWebsite of West Java Province, 2018). The study area has two distinctseasons: Wet Season (October to April), and Dry Season (May to August)(Shean, 2016). Most of the rice production is under irrigation. The mainrice-based cropping patterns in the study are area: rice-rice, rice-fallow,rice-nonrice, fallow-rice and nonrice-rice.

SAR multitemporal images from Sentinel-1A IW Level 1 Ground RangeDetected (GRD) were acquired from October 2017 to September 2018. C-band SAR data with 12-day revisit time including dual polarizationcapability (VV, VH) at 10m spatial resolution were available, whichcovered the seasonal rice crop calendar (rice-rice) and growth stagesover one year. Imagery was accessed and preprocessed through GoogleEarth Engine to derive the backscatter coefficient in each pixel.

Backscatter mechanism

Data

Figure 2: The temporal signal of SAR backscatter changesbased on rice growth. During the planting and vegetativephase, rice fields are inundated. The reflection on thewater surface is the dominant backscatter mechanism,resulting in very low backscatter. During the early growingstages, as the plant emerges through the surface the signalbounces multiple times causing an increase in the temporalSAR backscatter profile. When the plant reaches maturitythe backscatter signal decreases (Nguyen et al., 2016). Thiscan be used to identify rice from other crops.

Figure 3: Sentinel-1 imagesof rice fields on the NorthEast side of Indramayuduring the flooding,reproductive and maturestages.

Different polarizations havedifferent interactions withcrops due to multiplescattering events (McNairm& Shang, 2016)

For Sentinel-1, VH polarized backscatter is more sensitive to rice growth than VV polarized bactscatter, and was selected tomap rice areas. VH showed consistent increase from tillering to heading as opposed to VV. Figure 3, shows the seasonalrice crop calendar and growth stages for the study area. Mean VH backscatter for each month was plotted and to identifythe beginning of each rice cropping pattern associated with flooding of rice fields. The beginning of each cropping patternalso relates to the beginning of the wet or dry season.

RainfedNovemberDecember

85, 818 hectares17, 94867, 870

IrrigatedMarchMay

46, 983 hectares12, 83934, 144

Grand Total 132, 801 hectares

1. Clauss, K., Otinger, M., & Kuenzer, C. (2018). Mapping rice areas with Sentinel-1 time series and superpixelsegmentation. International Journal of Remote Sensing, 39(5), 1399-1420. 2. Cuerdo, G., Villano, L., Gutierrez, M. A., Garcia, C., Laborte, A., Boschetti, M., & Nelson, A. (2013). The when and where of rice. Rice Today, 12(2), 32–33. Retrieved from https://www.scribd.com/document/135034694/RT-Vol-12-No-2-the-When-and-Where-of-Rice#fullscreen=1.3. FAO. (1996). Agro-Ecological Zoning Guidelines. FAO Soils Bulletin 76, 3–5. Retrieved from https://www.mpl.ird.fr/crea/taller-colombia/FAO/AGLL/pdfdocs/aeze.pdf4. McNairn, H., & Shang, J. (2016). A Review of Multitemporal Synthetic Aperture Radar (SAR) for Crop Monitoring. In Y. Ban (Ed.), Multitemporal Remote Sensing (pp. 317–340). Springer International Publishing. https://doi.org/10.1007/978-3-319-47037-5_155. Nguyen, D. B., Gruber, A., & Wagner, W. (2016). Mapping rice extent and cropping scheme in the Mekong Delta using Sentinel-1A data. Remote Sensing Letters, 7(12), 1209–1218. https://doi.org/10.1080/2150704X.2016.12251726. Shean, M. (2016). Indonesia: Rice Production Prospects Reduced by El Nino. USDA Foreign Agricultural Service.

Claus et al., 2018