Global Soil Moisture Retrievals Fusion Dataset (2015–2019)

Wang, Z. Y.¹  Liu, Y. X. Y.^2,3,*

1. School of Civil and Surveying & Mapping Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, China;

2. Guangzhou Institute of Geography, Guangdong Academy of Sciences, Guangzhou 501170, China;

3. Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou), Guangzhou 511458, China

Abstract: In order to improve the spatial-temporal data quality of satellite based global soil moisture retrievals, the authors developed the Global Soil Moisture Retrievals Fusion Dataset (2015–2019) based on multi-source satellite fusion products and using the Soil Moisture Active Passive retrievals to interpolate the spatio-temporal series of the ECV data, and then reprojected, resampled, and weighted the calculation to produce a Global Soil Moisture Retrievals Fusion Dataset (Global SM). The dataset was validated by comparing the values with ground observations from 134 monitoring stations across eight soil moisture networks in Europe. The data quality was improved by approximately 20%. The dataset is 6.71 GB and consists of 1,737 files; it is archived in .tif format in 1,737 data files with the data size of 6.71 GB.

Keywords: soil moisture; global scale; daily data; 2015–2019

Dataset Availability Statement:

The dataset supporting this paper was published and is accessible through the Digital Journal of Global Change Data Repository at: https://doi.org/10.3974/geodb.2020.08.03.V1.

1 Introduction

Soil moisture—the volume of liquid water contained in a unit volume of soil—is a key physical quantity of global climate change, land surface hydrological processes, and the carbon cycle^[1–4]. It affects vegetation growth by controlling the soil heat capacity, surface evaporation, and vegetation transpiration^[5–7]. Therefore, obtaining accurate soil moisture data is necessary for assessing terrestrial ecosystem successions as well as the carbon, nitrogen, and water cycles; it can also provide early warnings of drought and flood disasters and improve estimations of crop yield^[8–11].

Satellite remote sensing technology can obtain continuous time-series data of land surface soil moisture on the global scale. Massive satellite soil moisture retrievals provide unprecedented opportunities for global climate evolution analysis^[12–13]; however, these datasets have large numbers of null areas because of satellite gaps, radio frequency interference, vegetation optical thickness, and freezing seasons^[14–16]. Different microwave bands (e.g., C-band, X-band, K-band, Ka-band, L-band) have different sensitivities to surface soil moisture^[17]. To improve the time-space sequence integrity and data quality of satellite soil moisture retrievals, the European Space Agency developed the Essential Climate Variable Soil Moisture product (ECV SM) by fusing multi-source satellite remote sensing data of global surface soil moisture since 2010^[18–20]. Compared with single-band satellite soil moisture retrievals, ECV SM has the longest time series, and its spatial sequence integrity and data accuracy have been significantly improved. However, the spatial coverage of this product is relatively low compared with that of other assimilation retrievals. Accordingly, integrating new satellite soil moisture retrievals can effectively improve the spatial integrity and data quality of ECV SM. To improve the integrity and accuracy of ECV SM retrieval data^[21], we integrated ECV SM with the L-band Soil Moisture Active Passive (SMAP) dataset from 2015 to 2019^[22]. After reprojecting, resampling, and interpolating the data, we produced the daily Global Soil Moisture Retrievals Fusion dataset (Global SM)^[23–26]with a resolution of 0.25° from March 31, 2015, to December 31, 2019.

2 Metadata of the Dataset

The metadata summary of the dataset^[27] is summarized in Table 1, which includes the dataset full name, short name, authors, year, temporal resolution, spatial resolution, data format, data size, data files, publisher, and sharing policies, etc.

3 Methods

3.1 Data Sources

ECV SM fuses multi-source active (ERS-1, ERS-2, MetOp-A, ASCAT) and passive microwave retrievals (SMMR, SSM/I, TMI, AMSR-E, AMSR-2, Windsat, SMOS)^[21] to form a global daily soil moisture dataset. SMAP is a global daily soil moisture retrieval dataset from 2015 to 2019; it is based on the L-band passive microwave radiometer inversion, with a spatial resolution of 36 km × 36 km^[22]. Many studies have shown that the sensitivity of the L-band to surface soil moisture is better than that of other microwave bands, and its ground penetration depth is closest to the depth of the soil moisture ground monitoring sensor. NASA has improved the SMAP satellite sensor and its inversion algorithm to enhance its anti-jamming capability of ground man-made radio frequency interference. The verification inferred a higher accuracy and stronger spatio-temporal adaptability of SMAP SM retrievals relative to that of the other satellite soil moisture retrievals in ECV^{[23, 29–31]}.

We verified and evaluated the ECV SM and ascending and descending SMAP SM observations by comparing the data with ground-based measurements from 134 monitoring stations across eight soil moisture measurement networks in Europe. The data of the eight European in-situ networks applied in this study were acquired from the International Soil Moisture Network^[32]. The basic attributes of each in-situ measurement are listed in Table 2.

Table 1  Metadata summary of the “Reprocessing dataset of global soil moisture product (2015–2019)”

Table 2  Information of the eight European in-situ measurement networks

3.2 Data Processing

This study aimed to improve the integrity and accuracy of ECV SM retrieval data. To evaluate the quality of SMAP and ECV SM retrievals, we compared the data with the in-situ observations. Moreover, to ensure the quality and stability of the ground measurements, only ground records covering > 12 hr in one day were considered valid, and the daily in-situ soil moisture was calculated as the arithmetic averages of all sites in every network. In this study, the correlation coefficient (R), bias, and unbiased root mean square error (ubRMSE) were the error parameters used to verify the accuracy of the ECV and SMAP data. Our results suggest that SMAP is a high-quality data source for ECV interpolation due to its higher accuracy. Secondly, to enhance the data coverage percentage, we obtained daily SMAP soil moisture data by calculating the arithmetic averages of the ascending and descending SMAP observations. Moreover, we conducted projection transformation and spatial resampling to enhance the spatial consistency of the ECV SM and SMAP SM datasets. Thirdly, we used Python to read and traverse the ECV daily data to identify the null areas. Finally, we used the SMAP daily SM retrieval consistent with ECV spatial properties to interpolate the ECV SM data and produce the Global SM dataset. Finally, this study evaluated and verified the spatial integrity and accuracy of the Global SM data. The data development process is shown in Figure 1.

Figure 1  Flowchart of the soil moisture retrieval fusional algorithm

4 Data Results and Validation

4.1 Data Composition

The global soil moisture retrievals fusion dataset (2015–2019) includes a total of 1,737 files and consists of daily global coverage data files from March 31, 2015, to December 31, 2019 (named in the form of SM-yyyymmdd.tif). The dataset has a spatial resolution of 0.25° × 0.25° (approximately 25 km × 25 km). The soil moisture unit in the dataset is m³žm^–3, and its value range is [0,1].

4.2 Data Results

Figure 2 compares the ECV SM with the global soil moisture retrievals fusion dataset (2015–2019) for January 1, April 1, July 1, and October 1, 2016. The soil moisture temporal and spatial distribution characteristics were highly consistent with the regional seasonal cycle and ranged from 0 to 0.5 m³žm^–3. The dataset was significant improved in its spatial coverage. In winter and spring, the surface temperature was constantly below 0 °C, with

Figure 2  Comparison Maps of ECV SM data (January (a), April (c), July (e), and October (g)) with the Global SM results (January (b), April (d), July (f), and October (h)) (Unit: m³žm^–3)

frozen soil in the high latitude regions. Greenland and Antarctica were covered with snow and ice throughout the year; however, the soil moisture value in those area was absent, as microwaves can only measure the content of liquid water in soil.

4.3 Data Validation

4.3.1 Integrity Assessment

Figure 3 compares the spatial coverage of the original ECV SM and the Global SM datasets. After data fusion, the spatial coverage of the Global SM improved by approximately 20% compared with the original ECV SM. Moreover, the Global SM filled the ECV SM null values in the Amazon rainforest and the Congo Basin rainforest. The Global SM therefore effectively analyzes the continuity of soil moisture on both temporal and spatial scales. Furthermore, we observed high soil moisture coverage in the middle and low altitudes (at approximately 60°S–60°N) and in areas of low vegetation coverage, as microwaves cannot effectively penetrate ice and vegetation (> 5 kgžm^–2).

Figure 3  Spatial coverage of ECV SM (a) and Global SM (b) data (Unit: %)

4.3.2 Accuracy Evaluation

We verified and evaluated the ECV SM and ascending and descending SMAP SM observations by comparing the data with measurements from 134 ground-based monitoring stations across the eight European soil moisture networks (Figures 4–6). The horizontal lines in the boxplots represent the maximum, upper four quantile, median, lower four quantile, and

Figure 4  Boxplots of correlation coefficients for ECV SM and SMAP SM retrievals

Figure 5  Boxplots of bias for ECV SM and SMAP SM retrievals

Figure 6  Boxplots of RMSE for ECV SM and SMAP SM retrievals

minimum values. The dotted line represents the average of an array, and the red points represent the outliers. The correlation coefficient and bias of SMAP were stronger than those of ECV SM, while the ubRMSE of SMAP was equivalent to that of ECV^[33]. This suggests that the SMAP SM retrievals are reliable, of high quality, and can effectively improve the integrity of the ECV SM data^[34].

We used in-situ data to verify the accuracy of the Global SM, and the results are shown in Table 3. The accuracy of the Global SM was equivalent to that of ECV SM, but the Global SM performed better in the REMEDHUS, FR_Aqui, RSMN, and SMOSMANIA networks. The Global SM can therefore effectively capture the amplitude and temporal variations of soil moisture and accurately fit the in-situ measurements. Overall, the Global SM can precisely reflect the distribution and variability of in-situ measurements.

To further analyze the correlation between the Global SM and in-situ data distributions, the curve of the probability distribution function (PDF) of the in-situ, original ECV SM, and Global SM datasets is shown in Figure 7. The three soil moisture datasets showed a normal distribution, but the ECV SM distribution (red line) was notably clustered and the curve of PDF of the Global SM was more closed to it of the in-situ data.

Table 3  Evaluation results of the Global SM data

Figure 7  Soil moisture PDF curves of ECV SM data, Global SM data, and in-situ measurements

5 Discussion and Conclusion

Using SMAP SM retrievals to fill in null ECV SM data points, we produced a daily global soil moisture dataset with a resolution of 0.25° × 0.25° from March 31, 2015, to December 31, 2019. To ensure data quality, we verified the ECV SM, ascending and descending SMAP SM, and the Global_SM datasets by comparing their values with 134 in-situ measurements across eight European ground-based networks. We found that the Global_SM dataset showed equivalent values and evolutionary spatio-temporal tendencies to that of the in-situ measurements. Moreover, the accuracy and spatial coverage integrity of the Global_SM dataset were significantly improved.

We assumed that the in-situ data represented the “ideal true value”, but its spatial resolution varied from the raster pixels with a resolution of 0.25° × 0.25°—especially in underlying surfaces with complex properties. Therefore, the validation results based on the in-situ data can to some extent prove the quality of the Global_SM dataset but are not entirely equivalent to the dataset’s accuracy.

Satellite soil moisture retrievals commonly have null areas. However, our study demonstrates that the use of high-precision SMAP retrievals to interpolate the null areas of the ECV SM data is an effective method for producing a global soil moisture dataset with high spatial coverage. We suggest that future research attempt to retrieve soil moisture data by mapping the relationship between soil moisture and multi-source surface parameters (such as precipitation, temperature, vegetation index) based on mathematical models.

Author Contributions

Liu, Y. X. Y. designed the dataset algorithms. Wang, Z. Y. contributed to the data processing and analysis and also drafted the manuscript.

References

[1]      Mei, S. Y., Walker, J. P., Rüdiger, C., et al. A comparison of SMOS and AMSR2 soil moisture using representative sites of the OzNet monitoring network [J]. Remote Sensing of Environment, 2017, 195: 297‒312.

[2]      Tang, R., Li, Z. L., Tang, B. An application of the Ts–VI triangle method with enhanced edges determination for evapotranspiration estimation from MODIS data in arid and semi-arid regions: implementation and validation [J]. Remote Sensing of Environment, 2010, 114(3): 540‒551.

[3]      Liu, Y., Yao, L., Jing, W., et al. Comparison of two satellite-based soil moisture reconstruction algorithms: a case study in the state of Oklahoma, USA [J]. Journal of Hydrology, 2020, 10(2): 1‒18.

[4]      Liu, Y., Jing, W., Wang, Q., et al. Generating high-resolution daily soil moisture by using spatial downscaling techniques: a comparison of six machine learning algorithms [J]. Advances in Water Resources, 2020, 141: 103601.

[5]      Dong, J., Crow, W. T., Tobin, K. J., et al. Comparison of microwave remote sensing and land surface modeling for surface soil moisture climatology estimation [J]. Remote Sensing of Environment, 2020, 242: 111756.

[6]      Rigden, A., Mueller, N., Holbrook, N., et al. Combined influence of soil moisture and atmospheric evaporative demand is important for accurately predicting US maize yields [J]. Nature Food, 2020, 1(2): 127‒133.

[7]      Bing, L., Su, H., Shao, Q., et al. Changing Characteristic of Land Surface Evapotranspiration and Soil Moisture in China during the Past 30 Years [J]. Journal of Geo-Information Science, 2012, 14(1): 1‒13.

[8]      Lakshmi, V., Piechota, T., Narayan, U., et al. Soil moisture as an indicator of weather extremes [J]. Geophysical Research Letters, 2004, 31(11): 1‒4.

[9]      Liu, J., Shi, C., Sun, S., et al. Improving land surface hydrological simulations in China using CLDAS meteorological forcing data [J]. Journal of Meteorological Research, 2019, 33(6): 1194‒1206.

[10]   Qin, J., Yang, K., Lu, N., et al. Spatial upscaling of in-situ soil moisture measurements based on MODIS-derived apparent thermal inertia [J]. Remote Sensing of Environment, 2013, 138: 1‒9.

[11]   Liu, S., Xing, B., Yuan, G., et al. Relationship analysis between soil moisture in root zone and top-most layer in China [J]. Chinese Journal of Plant Ecology, 2013, 37(1): 1‒17.

[12]   Dorigo, W., Wagner, W., Albergel, C., et al. ESA CCI soil moisture for improved earth system understanding: state-of-the art and future directions [J]. Remote Sensing of Environment, 2017, 203: 185‒215.

[13]   Liu, Y., Yang, Y., Jing, W., et al. Comparison of different machine learning approaches for monthly satellite-based soil moisture downscaling over northeast China [J]. Remote Sensing, 2018, 10(1): 31.

[14]   Draper, D. W. Radio frequency environment for earth-observing passive microwave imagers [J]. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2018, 11(6): 1913‒1922.

[15]   Piepmeier, J. R., Johnson, J. T., Mohammed, P. N., et al. Radio-Frequency interference mitigation for the Soil Moisture Active Passive Microwave Radiometer [J]. IEEE Transactions on Geoscience & Remote Sensing, 2014, 52(1): 761‒775.

[16]   Zou, X., Zhao, J., Weng, F., et al. Detection of radio-frequency interference signal over land from FY-3B microwave radiation imager (MWRI) [J]. Advances in Meteorological Science & Technology, 2013, 50(12): 4994‒5003.

[17]   Sabaghy, S., Walker, J. P., Renzullo, L. J., et al. Spatially enhanced passive microwave derived soil moisture: capabilities and opportunities [J]. Remote Sensing of Environment, 2018, 209: 551‒580.

[18]   Hollmann, R., Merchant, C. J., Saunders, R., et al. The ESA climate change initiative: satellite data records for essential climate variables [J]. Bulletin of the American Meteorological Society, 2013, 94(10): 1541‒1552.

[19]   Dorigo, W. A., Gruber, A., Jeu, R. A. M. D., et al. Evaluation of the ESA CCI soil moisture product using ground-based observations [J]. Remote Sensing of Environment, 2015, 162(2): 380‒395.

[20]   Chakravorty, A., Chahar, B. R., Sharma, O. P., et al. A regional scale performance evaluation of SMOS and ESA-CCI soil moisture products over India with simulated soil moisture from MERRA-Land [J]. Remote Sensing of Environment, 2016, 186: 514‒527.

[21]   ESA CCI Soil Moisture [DB/OL]. https://www.esa-soilmoisture-cci.org/.

[22]   SMAP Data [DB/OL]. https://nsidc.org/data/smap/smap-data.html.

[23]   Spencer, M., Wheeler, K., White, C., et al. The Soil Moisture Active Passive (SMAP) mission L-Band radar/radiometer instrument [A]. Geoscience and Remote Sensing Symposium, 2010 IEEE International, 2010:  3240–3243.

[24]   O’Neill, P., Entekhabi, D., Njoku, E., et al. The NASA Soil Moisture Active Passive (SMAP) mission: Overview [A]. Geoscience and Remote Sensing Symposium (IGARSS), 2010 IEEE International, 2010: 1210‒1223.

[25]   Ma, H., Zeng, J., Chen, N., et al. Satellite surface soil moisture from SMAP, SMOS, AMSR2 and ESA CCI: a comprehensive assessment using global ground-based observations [J]. Remote Sensing of Environment, 2019, 231: 111215.

[26]   Liu, Y., Yang, Y., Jing, W. Potential applicability of SMAP in ECV soil moisture gap-filling: a case study in Europe [J]. IEEE Access, 2020, 8(133): 114‒127.

[27]   Liu, Y. X. Y. Global soil moisture retrievals fusion dataset (2015–2019) [J/DB/OL]. Digital Journal of Global Change Data Repository, 2020. https://doi.org/10.3974/geodb.2020.08.03.V1.

[28]   GCdataPR Editorial Office. GCdataPR data sharing policy [OL]. https://doi.org/10.3974/dp.policy.2014.05 (Updated 2017).

[29]   Stillman, S., Zeng, X. Evaluation of SMAP soil moisture relative to five other satellite products using the climate reference network measurements over USA [J]. IEEE Transactions on Geoscience & Remote Sensing, 2018, 99: 1‒10.

[30]   Chen, F., Crow, W. T., Bindlish, R., et al. Global-scale evaluation of SMAP, SMOS and ASCAT soil moisture products using triple collocation [J]. Remote Sensing of Environment, 2018, 214: 1‒13.

[31]   Zhang, X., Zhang, T., Zhou, P., et al. Validation analysis of SMAP and AMSR2 soil moisture products over the United States using ground-based measurements[J]. Remote Sensing, 2017, 9(2): 104.

[32]   International Soil Moisture Network. https://ismn.geo.tuwien.ac.at/en/.

[33]   Cui, C., Xu, J., Zeng, J., et al. Soil moisture mapping from satellites: an intercomparison of SMAP, SMOS, FY3B, AMSR2, and ESA CCI over two dense network regions at different spatial scales [J]. Remote Sensing, 2017, 10(2): 33.

[34]   Burgin, M. S., Colliander, A., Njoku, E. G., et al. A comparative study of the SMAP passive soil moisture product with existing satellite-based soil moisture products [J]. IEEE Transactions on Geoscience & Remote Sensing, 2017, 55(5): 2959‒2971.