Evaluating irrigation status in the Mekong Delta through polarimetric L-band SAR data assimilation
Introduction
The objective of this study is to develop a SAR-based monitoring system that provides daily paddy field water level data for irrigation management, required in the simulation of GHG emissions and rice productivity. In Asia, rice is the most important staple food, providing an average 32% of total calorie uptake (Maclean et al., 2002). About 90% of the global paddy area and annual output of rice production are concentrated in monsoonal Asia (FAOSTAT, 2022). Approximately 75% of global rice production is produced in irrigated lowlands encompassing delta basins in Asia, such as the Mekong, Irrawaddy, Chao Phraya, and Bengal Rivers (Barker et al., 1985; Maclean et al., 2002). To meet the increasing food demand derived from global population growth, rice cropping in the region is becoming more intensive with the use of double/triple rice cropping techniques (FAO, 2020). However, rice production requires large amounts of water (3000–5000 L kg−1 rice, IRRI, 2001) and has become a major source of the potent greenhouse gas methane, approximately 11% of anthropogenic methane emissions come from rice paddy submerged soils (IPCC, 2013). Therefore, water-saving irrigation practices that have the potential to mitigate methane emissions by oxidizing the soil environment, such as alternate wetting and drying (AWD), are desirable for dissemination in this region to ensure sustainable water demand while lowering greenhouse gas emissions.
AWD is an intermittent irrigation technique to lower the amount of water used for irrigation without a significant loss of rice productivity (Bouman et al., 2007; Lampayan et al., 2015). This approach consists of temporarily suspending irrigation until the paddy soil becomes dry/oxidized. Initially, irrigation water is applied to obtain flooded conditions (5–10 cm of water above the soil surface). Then, after a certain number of days have passed, the field water level drops below the soil surface. The next irrigation is scheduled before the soil becomes dry. However, the perception of soil dryness and consequently the number of days between 2 irrigations varies widely among farmers and sites (from 1 day to >10 days) depending on the soil texture (clay/silt/sand content), percolation rate, precipitation and evaporation (Bouman et al., 2007, Lampayan et al., 2015). Various soil drought indexes have been used for scheduling irrigation (Bouman et al., 2007); for example, the soil matric potential is used to define the value of the soil drought index as the irrigation criterion. Scientists have also advocated for the use of the farmer-determined water level has also been suggested (hereafter referred to as Dbefore_irrigation, i.e., how deep the field water level drops before the next round of irrigation is initiated). This specific water depth has been widely used by policy makers regarding irrigation activity, as rice farmers can easily monitor this value (Bouman et al., 2007; Rejesus et al., 2011). To define Dbefore_irrigation, several field experiments have been conducted in Asia (e.g., Arai et al., 2021; Belder et al., 2004; Bouman and Tuong, 2001; Lampayan et al., 2015). These studies report that increasing the interval between rounds of irrigations to allow the field water level to drop to deeper soil layers is a promising technique to save water, reduce methane emissions and increase rice yield in the Mekong Delta (Arai, 2022). ‘Irrigation when the field water level has leached 15–20 cm below the soil” is the criterion suggested by the International Rice Research Institute (IRRI). In Vietnam, in particular the Mekong delta, highly intensive cropping system was developed based on the construction and upgrading of a multilevel dike system, allowing double and triple rice crop per year. The AWD practices have also been adopted in certain regions as a water saving regulation.
However, despite the wide adoption of AWD, the values of Dbefore_irrigation differ by location (Ishido et al., 2016; Taminato and Matsubara, 2016). Most importantly, the adoption of AWD by farmers is constrained by several social/infrastructural factors. These factors include distance from the irrigation canal/drainage, density of subcanals inside the dikes, and location of pump ownership (Yamaguchi et al., 2017). At such community scale (i.e., 10-100 m spatial resolution with daily temporal resolution), heterogeneity is caused by nonsynchronized intermittent drainage in adjacent rice paddies, in land surface elevations and in cropping calendars (Dang et al., 2016; Evers and Benedikter, 2009, Miyashita et al., 2016). Other natural factors arise, such as irregular daily field water level dynamics of the irrigation canal due to ebbs and flows of tides (Wassmann et al., 2004).
These heterogeneous water management practices cause difficulties in monitoring/verifying mitigation measures operations (e.g., AWD). Such monitoring is required for irrigation management, such as allocation of irrigation/drainage canals and pumping apparatuses for irrigation/drainage schedule synchronization, opening/closing schedules of water gates attached to each dike surrounding rice paddies, dike heightening, selection of paddies that require field water level observations, etc.
In this context, there is a need to monitor the irrigation status [i.e., field water level dynamics and farmers' irrigation practices (e.g., Dbefore_irrigation)] to evaluate the actual dissemination status of water-saving irrigation activities by each stakeholder and verify the longevity of already implemented mitigation measures. To address this issue, various studies have been conducted using remote sensing techniques to evaluate the soil submersion status. Using optical data, the rice paddy covered with water can be distinguished at the beginning of the season based on the reflectance of the water surface. The method often uses the normalized difference water index (NDWI) and the normalized difference vegetation index (NDVI) from remote sensing images to depict the differences between rice paddies and nonrice paddy areas (Zhang et al., 2020). However, as soon as the rice paddy is covered by vegetation, it is no longer possible to detect flooded and nonflooded paddy fields. Using SAR data with penetration capability, soil submersion/nonsubmersion is expected to be detectable in rice paddies for a longer period during the rice growth stage, depending on the SAR wavelength. Using C-band SAR data, Lam-Dao (2009) reported that soil submersion could be detected up to 20 days after sowing using the HH backscattering intensity of Advanced Synthetic Aperture Radar (ASAR). Using longer-wavelength SAR data with deeper penetration in the vegetation layer, Arai et al. (2018) demonstrated the performance of quadruple/dual-polarimetric L-band SAR (ALOS-2 PALSAR-2: Advanced Land Observing Satellite-2/Phased Array type L-band Synthetic Aperture Radar-2) data to distinguish submerged paddies and nonsubmerged paddies across all rice growth stages. However, the high resolution polarimetric images cover limited area (40 km swath) and irregular temporal frequency. For larger region, ScanSAR data are more adapted. However, the 42-day repeat cycle by which PALSAR-2 ScanSAR data are acquired is not sufficient to monitor the temporal dynamics of paddy field water levels. Future L-band SAR missions, such as ALOS-4, ROSE-L and NISAR, will have 14- and 12-day repeat cycles, respectively, will be still not sufficient for temporal monitoring. Many hydrological simulation studies (e.g., Al-Yaari et al., 2014; Ines et al., 2013; Montzka et al., 2011) applying data assimilation (hereafter, assimilation) have used high temporal resolution passive microwave radiometer data [e.g., Soil Moisture and Ocean Salinity (SMOS), in which the L-band (1.4 GHz) data are collected every 3 days]. However, their spatial resolution of 20–50 km is not adapted to the heterogeneity of the rice paddies in Asia. In this regard, it is very important to develop a technique using high spatial resolution SAR data at low temporal resolution to simulate the temporal dynamics of the irrigation status of rice paddies.
The objective of this study is to develop a system that provides daily field water level data for irrigation management at the community scale. The starting point is the use of ALOS-2 PALSAR-2 data which provide pixel-based information on submersion status of paddy fields every 42 days. We prepared a model simulating field water level dynamics, based on knowledge of the hysteresis of the soil water retention curve and our long-term ground observation data. and we use the SAR data to optimize model parameters. To estimate the values of Dbefore_irrigation at the community scale, we assimilate the backscattering coefficient of the ALOS-2 PALSAR-2 data by preparing a pixel-based observation operator to detect the submersion status of paddy fields. To solve the filter divergence issues that become critical due to the low temporal resolution of satellite observations, we designed a focal statistics-based observation operator to evaluate the temporal submersion percentage and the spatial submersion percentage in a focal analysis window.
Section snippets
Methodology
In field water level simulation at local scale, soil properties and farmers' irrigation practices need to be considered. Unfortunately, most models simulating hydraulic states of agricultural soils require the field water level data as “forcing” input data (e.g., “depth of water retention layer” as input data of DNDC-rice model: Fumoto et al., 2008; Soil, Water, Atmosphere and Plant model: Govindarajan et al., 2008). The difficulty is mainly due to large uncertainties linked to anthropogenic
Results
Comparison with ground-observed inundation status and quadruple polarimetric PALSAR-2 data was carried out to design the PALSAR-2 data assimilation system (i.e., observation operator) to adaptatively estimate the parameters of the irrigation model with high spatiotemporal resolution. Seven decomposition components showed clear consistency with 3 types of rice paddy inundation status [noninundated, semi-inundated paddies and inundated paddies, Fig. 3, Fig. 4, Fig. 5]. The use of all 7 components
Development of the observation operator for SAR data assimilation by developing inundation classification methodology
Inundation was classified with following 3 states as inundated paddies, semi-inundated paddies, and non-inundated paddies regarding the difference of field water level. In the studied clay soil, field water level at 0 cm of the soil surface and − 5 cm below the soil surface were differentiated by the L-band microwave signals. This indicated that L-band microwave can penetrate shallow soil surface to approximately -5 cm below soil surface. Since semi-inundated paddies was characterized by higher
Conclusion
The concept and required techniques of SAR data assimilation with relatively high spatial resolution and low temporal resolution are demonstrated in this study. This study evaluates a soil drought index (i.e., Dbefore_irrigation) that has high potential to inform decision making regarding the sustainable water and carbon resource management of rice paddies. Moreover, this approach can support adaptive simulation with biogeochemistry models considering farmer irrigation practices. By
CRediT authorship contribution statement
Hironori Arai: Conceptualization, Methodology, Software, Data curation, Validation, Writing – original draft, Writing – review & editing, Visualization, Funding acquisition, Project administration. Thuy Le Toan: Supervision, Project administration, Investigation, Writing – review & editing. Wataru Takeuchi: Supervision, Project administration, Resources. Kei Oyoshi: Resources, Project administration. Tamon Fumoto: Formal analysis. Kazuyuki Inubushi: Resources, Investigation, Validation.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
A portion of the field survey activities was supported by the Japan Society for the Promotion of Science (JSPS) Grants-in-aid for Scientific Research & Scholarship received from JSPS as research execution expenses and by JSPS KAKENHI Grant Numbers 15J00001 and 16J02509. The remote sensing study was financially supported by the Japan Aerospace Exploration Agency (JAXA) and by the JSPS KAKENHI Project (Area No. 16J02509) and JSPS Overseas Research Fellowships (No. 201960100). ALOS-2 PALSAR-2 data
References (44)
Surface water maps de-noising and missing-data filling using determinist spatial filters based on several a priori information
Remote Sens. Environ.
(2020)- et al.
Global-scale evaluation of two satellite-based passive microwave soil moisture datasets (SMOS and AMSR-E) with respect to land data assimilation system estimates
Remote Sens. Environ.
(2014) - et al.
Effect of water-saving irrigation on rice yield and water use in typical lowland conditions in Asia
Agr. Water Manag.
(2004) - et al.
Field water management to save water and increase its productivity in irrigated lowland rice
Agr. Water Manag.
(2001) - et al.
Efficient data assimilation for spatiotemporal chaos: a local ensemble transform Kalman filter
Physica D: Nonlinear Phenomena
(2007) - et al.
Assimilation of remotely sensed soil moisture and vegetation with a crop simulation model for maize yield prediction
Remote Sens. Environ.
(2013) - et al.
Water losses through paddy bunds: methods, experimental data, and simulation studies
J. Hydrol.
(2009) - et al.
Adoption and economics of alternate wetting and drying water management for irrigated lowland rice
Field Crop Res.
(2015) - et al.
Hydraulic parameter estimation by remotely-sensed top soil moisture observations with the particle filter
J. Hydrol.
(2011) - et al.
Impact of the alternate wetting and drying (AWD) water-saving irrigation technique: evidence from rice producers in the Philippines
Food Policy
(2011)
A water balance model for rainfed lowland rice fields emphasising lateral water movement within a toposequence
Ecol. Model.
Assessment of the relationship between adoption of a knowledge-intensive water-saving technique and irrigation conditions in the Mekong Delta of Vietnam
Agr. Water Manag.
Increased rice yield and reduced greenhouse gas emissions through alternate wetting and drying in a triple cropped rice field in the Mekong Delta
Sci. Total Environ.
Greenhouse gas emissions derived from rice straw burning and straw-mushroom cultivation in a triple rice cropping system in the Mekong Delta
Soil Sci. Plant Nutr.
Estimation of methane emissions from rice paddies in the Mekong Delta based on land surface dynamics characterization with remote sensing
Remote Sens.
Alternate wetting and drying enhanced the yield of a triple-cropping rice paddy of the Mekong Delta
Soil Sci. Plant Nutr.
Pixel-based evaluation of rice production and related greenhouse gas emissions in the Mekong Delta integrating SAR data and ground observations
The Rice Economy of Asia. Resources for the Future
Water management in irrigated rice: coping with water scarcity
Int. Rice Res. Inst.
Hydrological alterations from water infrastructure development in the Mekong floodplains
Hydrol. Process.
Hydraulic bureaucracy in a modern hydraulic society-strategic group formation in the Mekong Delta, Vietnam
Water Altern.
FAO Statistical Databases
Cited by (6)
Increased rice yield and reduced greenhouse gas emissions through alternate wetting and drying in a triple-cropped rice field in the Mekong Delta
2022, Science of the Total EnvironmentCitation Excerpt :A particularly high inundation probability is observed in the northeastern region of the delta (Arai et al., 2018). In addition, the paddy field water level tends not to be reduced to 15 cm below the soil surface during the cropping period in relatively low-elevation areas in the delta (Arai et al., 2022b). Considering the temporally accumulative property of soil oxidation by AWD and its large potential to reduce regional emissions, it is important to develop a robust methodology to monitor the permanence/continuity of AWD implementation effects by integrating remote sensing technology (Fig. 6).
Impact of different water management and microbe application on yield of rice cultivars under seawater intrusion areas of Indonesia
2023, Chilean Journal of Agricultural ResearchEarth Observation Retrieval and Classification Algorithms for Agriculture
2023, International Geoscience and Remote Sensing Symposium (IGARSS)QUALITY EVALUATION OF WATER RESOURCES FOR IRRIGATION IN SULAIMANI GOVERNORATE, IRAQ
2023, Applied Ecology and Environmental Research