5:15 PM - 6:45 PM
[ACG36-P17] Assessment of AMSR2 Soil Moisture Products in the Mongolian Plateau in 2021
Keywords:Soil Moisture, Mongolian Plateau, AMSR2, Land Surface Temperature, NDVI, SGLI
1. Introduction
In the recent years, global soil moisture observation by satellites equipped with microwave radiometers has progressed, and global soil moisture data sets from GCOM-W (JAXA), SMAP (NASA), SMOS (ESA) and other satellite observation data have been released for research purposes. Therefore, it is important to compare soil moisture products and to understand the characteristics of each product.
Therefore, this study aims to evaluate AMSR2 products by comparing AMSR2 soil moisture products with field observation data for the Mongolian Plateau, and analyzing the relationship with surface temperature and vegetation index.
2. Method and Data
In this study, the year 2021 was used as the target period, and the data were analyzed by comparing the field observation data, AMSR2 observation data, surface temperature and vegetation index data from SGLI in the Mongolian plateau.
i) In Situ data
Fig. 1 shows the target area of this study and the soil moisture observation network on the Mongolian plateau. On the Mongolian plateau, surface soil moisture is measured every one or two hours by three AWS (automatic weather stations) and ten ASSH (automatic stations for soil hydrology). This study used the data at a depth of 3cm.
ii) AMSR2 observation data
AMSR2 is a microwave radiometer onboard GCOM-W and is a sensor to estimate geophysical parameters mainly related to water. This study used Level 2 (L2) soil moisture products (10 km mesh) that contain geophysical quantities calculated from brightness temperatures for each footprint.
iii) SGLI observation data
The SGLI is a multi-wavelength optical radiometer mounted on GCOM-C. It is an optical sensor that observes atmospheric particulates and vegetation activity in the near-ultraviolet to thermal infrared wavelength range. The L2 product (250 m mesh) was used in this study.
3. Results and Discussion
Fig. 2 shows a 1:1 correspondence between In Situ data and AMSR2 soil moisture for both ascending and descending orbits. The ascending and descending orbits are often underestimated, especially in the months of August and September. In terms of accuracy, the correlation coefficient is larger and the RMSE is smaller for the descending orbit than the ascending orbit, indicating that the descending orbit is more accurate than the ascending orbit.
Fig. 3 shows the temporal variation of In Situ data, AMSR2 soil moisture data, SGLI (descending orbit) surface temperature (LST), and normalized difference vegetation index (NDVI). The AMSR2 soil moisture data were averaged over the target area for ascending and descending orbits, respectively, and the SGLI surface temperature (LST) and normalized difference vegetation index (NDVI) were averaged over the target area for cloud cover of 20% or less. Fig. 3 shows that the AMSR2 soil moisture data and In Situ data show similar characteristics, indicating that the vegetation index increases with an increase in soil moisture. From May to July, when the vegetation index is low, the dispersion of surface temperature is large and the AMSR2 soil moisture content is overestimated by about 0.02 relative to the In Situ data, but after the vegetation index increases from August, the dispersion of surface temperature is small and the AMSR2 soil moisture content is underestimated by about 0.05. This indicates that the accuracy of the AMSR2 soil moisture observation changes with changes in surface temperature and vegetation index.
Fig. 4 shows the relationship between the difference between AMSR2 soil moisture and In Situ data and SGLI surface temperature and normalized difference vegetation index during the ascending orbit. These results suggest that the AMSR2 soil moisture data are more strongly influenced by the vegetation index than surface temperature, and that the underestimation of the vegetation index increases with an increase in the vegetation index. This may be due to the fact that the increase in vegetation increases the brightness temperature.
4. Conclusion
This study evaluated AMSR2 soil moisture data by comparing In Situ data and analyzing the relationship between SGLI surface temperature and normalized difference vegetation index. As a result, The AMSR2 soil moisture data were more accurate in the descending orbit than in the ascending orbit. The AMSR2 soil moisture data tended to be underestimated as the vegetation index increased, suggesting that the increase in brightness temperature due to the increase in vegetation was a factor in the underestimation.
Acknowledged
This study was supported by the Japan Aerospace Exploration Agency (JAXA) through a joint research project (JX-PSPC-544613). The authors would like to express their gratitude to JAXA.
In the recent years, global soil moisture observation by satellites equipped with microwave radiometers has progressed, and global soil moisture data sets from GCOM-W (JAXA), SMAP (NASA), SMOS (ESA) and other satellite observation data have been released for research purposes. Therefore, it is important to compare soil moisture products and to understand the characteristics of each product.
Therefore, this study aims to evaluate AMSR2 products by comparing AMSR2 soil moisture products with field observation data for the Mongolian Plateau, and analyzing the relationship with surface temperature and vegetation index.
2. Method and Data
In this study, the year 2021 was used as the target period, and the data were analyzed by comparing the field observation data, AMSR2 observation data, surface temperature and vegetation index data from SGLI in the Mongolian plateau.
i) In Situ data
Fig. 1 shows the target area of this study and the soil moisture observation network on the Mongolian plateau. On the Mongolian plateau, surface soil moisture is measured every one or two hours by three AWS (automatic weather stations) and ten ASSH (automatic stations for soil hydrology). This study used the data at a depth of 3cm.
ii) AMSR2 observation data
AMSR2 is a microwave radiometer onboard GCOM-W and is a sensor to estimate geophysical parameters mainly related to water. This study used Level 2 (L2) soil moisture products (10 km mesh) that contain geophysical quantities calculated from brightness temperatures for each footprint.
iii) SGLI observation data
The SGLI is a multi-wavelength optical radiometer mounted on GCOM-C. It is an optical sensor that observes atmospheric particulates and vegetation activity in the near-ultraviolet to thermal infrared wavelength range. The L2 product (250 m mesh) was used in this study.
3. Results and Discussion
Fig. 2 shows a 1:1 correspondence between In Situ data and AMSR2 soil moisture for both ascending and descending orbits. The ascending and descending orbits are often underestimated, especially in the months of August and September. In terms of accuracy, the correlation coefficient is larger and the RMSE is smaller for the descending orbit than the ascending orbit, indicating that the descending orbit is more accurate than the ascending orbit.
Fig. 3 shows the temporal variation of In Situ data, AMSR2 soil moisture data, SGLI (descending orbit) surface temperature (LST), and normalized difference vegetation index (NDVI). The AMSR2 soil moisture data were averaged over the target area for ascending and descending orbits, respectively, and the SGLI surface temperature (LST) and normalized difference vegetation index (NDVI) were averaged over the target area for cloud cover of 20% or less. Fig. 3 shows that the AMSR2 soil moisture data and In Situ data show similar characteristics, indicating that the vegetation index increases with an increase in soil moisture. From May to July, when the vegetation index is low, the dispersion of surface temperature is large and the AMSR2 soil moisture content is overestimated by about 0.02 relative to the In Situ data, but after the vegetation index increases from August, the dispersion of surface temperature is small and the AMSR2 soil moisture content is underestimated by about 0.05. This indicates that the accuracy of the AMSR2 soil moisture observation changes with changes in surface temperature and vegetation index.
Fig. 4 shows the relationship between the difference between AMSR2 soil moisture and In Situ data and SGLI surface temperature and normalized difference vegetation index during the ascending orbit. These results suggest that the AMSR2 soil moisture data are more strongly influenced by the vegetation index than surface temperature, and that the underestimation of the vegetation index increases with an increase in the vegetation index. This may be due to the fact that the increase in vegetation increases the brightness temperature.
4. Conclusion
This study evaluated AMSR2 soil moisture data by comparing In Situ data and analyzing the relationship between SGLI surface temperature and normalized difference vegetation index. As a result, The AMSR2 soil moisture data were more accurate in the descending orbit than in the ascending orbit. The AMSR2 soil moisture data tended to be underestimated as the vegetation index increased, suggesting that the increase in brightness temperature due to the increase in vegetation was a factor in the underestimation.
Acknowledged
This study was supported by the Japan Aerospace Exploration Agency (JAXA) through a joint research project (JX-PSPC-544613). The authors would like to express their gratitude to JAXA.