I. INTRODUCTION
Land Surface Models (LSM) provide a quantitative method to simulate the exchanges of water and energy fluxes at the Earth’s surface, which accounts only for soil moisture estimates within the context of natural phenomena. On the other hand, high-resolution satellite images provide soil moisture estimates that incorporate irrigation, a human intervention that contributes to the large imbalances in the natural hydrological cycle. Hence, LSM excludes anthropogenic activities like irrigation, while satellite observations reflect irrigation signals. Comparing soil moisture estimates from the Land Surface Model with satellite data provides information concerning the pattern of when irrigation takes place and later can provide information for further research into the amount of irrigation applied to the command area by calculating the differences between the two values, assuming other water balance factors remain constant.
II. STUDY AREA
The region of interest is South Rangsit, one of the existing irrigation projects in the Pasak River Basin. The projects are located at the bottom of the basin, where water is supplied from the Pasak Jolasid Dam. The project covers an irrigable area of 848 km2 and is located at the central part of the existing irrigation projects.
The country experiences distinct weather patterns influenced by seasonal monsoon winds. The wet season typically extends from May to October, followed by the dry season from November to April. During the dry season, irrigation systems ensure an adequate crop water supply.
III. DATA
In this study, the data used for soil moisture are derived from two main sources: the Land Surface Model (LSM) and Sentinel-1 C-band Synthetic Aperture Radar (SAR). The LSM provides soil moisture estimates without accounting for human interventions, while Sentinel-1 SAR includes S1 Ground Range Detected (GRD) scenes that capture human-induced changes in water balances, specifically irrigation. The data are then converted to soil moisture using an algorithm proposed by Bauer-Marschallinger et al. (2019).
The soil moisture estimates from the LSM are recorded daily in cubic meters per cubic meter (m³/m³), while the data set from Sentinel 1 is not recorded on a regular schedule. Therefore, the analysis considers only those dates for which the LSM and Sentinel-1 data are available. The study covers the period from 2018 to 2021.
Figure 1: The region of interest, which is the South Rangsit, is one of the existing irrigation projects under the Pasak River Basin
IV. METHODS
The datasets from the Land Surface Model (LSM) and Sentinel-1 Soil Moisture were processed using the Google Earth Engine platform to extract time series changes in soil moisture for the region of interest. The soil moisture estimates from the LSM represent daily averages, considering a depth range of 0 to 10 cm to align with the depth recorded by Sentinel-1.
To capture the soil moisture signals from Sentinel-1, the Vertical-Vertical (VV) single co-polarization was utilized, focusing on data from the descending orbit, as irrigation typically occurs during the daytime. This approach enables a detailed analysis of irrigation patterns, particularly reflecting the sudden increases in soil moisture that occur when water is supplied to the fields.
V. RESULTS AND ANALYSES
The study period spans from 2018 to 2021, with a specific focus only on the dry season in the area, which occurs from November to April. The analysis will focus only on this period, as irrigation systems are typically employed to support agricultural practices. It is important to note that soil moisture patterns during the wet season behave differently, which exhibit variations that might be caused by some conditions.
In comparing the soil moisture data from the Land Surface Model (LSM) with precipitation data, it is evident that LSM soil moisture levels correspond closely with rainfall events. Soil moisture increases during and immediately after rainfall and decreases in the absence of precipitation. This pattern remained consistent throughout all four years of the study, with only minor variations noted in certain months.
Figure 2: . The images illustrate the variations in soil moisture between the Land Surface Model (LSM) and Sentinel-1 satellite data from 2018 to 2021. The shaded area represents the dry season when irrigation is supplied to the fields. During this period, Sentinel-1 estimates show higher values than LSM estimates, with sudden spikes indicating potential irrigation events.
Focusing on the dry season from November to April, noticeable differences appear between the soil moisture data from the Land Surface Model (LSM) and Sentinel-1. While the LSM reflects the natural water balance in the region, the signals captured by Sentinel-1 indicate higher soil moisture estimates. This discrepancy suggests that the satellite detects additional soil moisture signals beyond the natural water balance, likely due to irrigation practices.
Furthermore, the Sentinel-1 data reveals sudden spikes in soil moisture estimates, which may indicate irrigation events in the area, particularly as these increases occur even in the absence of rainfall. By examining the differences in soil moisture between the two datasets, it becomes feasible to conduct a more in-depth analysis of irrigation volumes in the region using this approach.
Figure 3: . The images exhibit the differences in soil moisture quantity between the LSM and Sentinel 1, focusing only on the dry season.
Figure 4: . Visualization of the average soil moisture amount captured by the Sentinel 1.
V. CONCLUSION
The Land Surface Model (LSM) demonstrates a reasonable correlation with precipitation data, showing that soil moisture increases in response to rainfall events. However, further analysis is needed to investigate the discrepancies in soil moisture during the wet season, as the patterns between LSM and satellite data exhibit significant variations that require analysis.
In contrast, during the dry season, the soil moisture patterns between LSM and satellite data align as expected, with irrigation likely contributing to the observed variations between the two datasets.
VII. REFERENCES
Bauer-Marschallinger, B., Freeman, V., & Cao, S. P. (2019). Toward Global Soil Moisture Monitoring with Sentinel-1: Harnessing Assets and Overcoming Obstacles. IEEE Transactions on Geoscience and Remote Sensing, 524.
This article is developed by Kysiah Putalan.
ALICE-LAB: Asian Land Information for Climate and Environmental Research Laboratory
The Community Water Model (CWatM) is an open-source hydrology tool that models the water cycle, incorporating both natural processes and human water demands. Used in ALICE-LAB, it supports sustainable water management by distinguishing climate impacts from human activities.
The Noah-MP LSM, developed by NCAR, simulates complex land surface processes like soil moisture, snowpack, and evapotranspiration, vital for hydrology and climate studies. With customizable modules, it supports diverse scales and applications, enabling ALICE-LAB to conduct high-resolution simulations on global water dynamics and long-term climate impacts.