Table of Contents

Introduction

The validation of the H25 Metop ASCAT DR2015 SSM time series 12.5 km sampling product (hereafter referred to test data set) is summarized in this online Product Validation Document (PVR). The online PVR gives an overview of the data sets and methods used to validate the test data set. The analysis of the test data set follows the guidelines described in the Metop ASCAT Product Validation Report [1]. The committed area and quality benchmarks are defined the Product Requirements Document (PRD) [2]. The committed area represents a restricted geographical region with high confidence in the successful retrieval of surface soil moisture information from Metop ASCAT. The area is limited to low and moderate vegetation regimes, unfrozen and no snow cover, low to moderate topographic variations, as well as no wetlands and coastal areas (see Figure 1).

All quality benchmarks were computed on a global basis and are presented either globally (i.e. all valid results) or masked to the committed product area. The validation framework of the Python Toolbox for the Evaluation of Soil Moisture Observations (pytesmo v0.6.0) has been used to perform the validation.

More information on the soil moisture data records can be found in the Product User Manual (PUM) [3] and Algorithm Theoretical Baseline Document (ATBD) [4].

Figure 1: Committed Metop ASCAT soil moisture area.

Figure 1: Committed Metop ASCAT soil moisture area.

Data

The validation has been performed globally for the time period 2007-01-01 until 2014-12-31 on the WARP 5 grid. As reference data set the ERA-Interim land surface model and the passive CCI soil moisture product (v2.3) were used. The first soil moisture layer (0.00 - 0.07 m) of ERA-Interim was used for the validation. The soil temperature and snow depth information was used for filtering for non-frozen (soil temperature > 4) and snow-free (snow depth = 0) time periods.

Methods

The standard quality benchmark Signal-to-Noise Ratio (SNR) [5] and, in addition, the Pearson correlation coefficient (R) have been computed on a global basis. The Triple Collocation Analysis (TCA) has been performed between the test data set, ERA-Interim and the passive CCI soil moisture product, whereas R was only computed between the test data set and ERA-Interim.

Results

The quality benchmarks have been computed on a global basis, but under certain circumstances (e.g. no valid measurements) no results have been obtained. In addition, locations with a p-value > 0.05 have been discarded.

Boxplot of metrics

The following Boxplot in Figure 2 summarizes the distribution of the quality benchmarks. The whisker indicate the 5th and 95th percentile, whereas the size of the box represents the Inter Quartile Range (IQR). A percentage indicating the number of locations exceeding the threshold/target/optimal requirements is given as well.

Figure 2: The boxplots indicate the distribution of the quality benchmarks globally and just for the committed area. A percentage of locations exceeding each of the three thresholds is indicated as well.

Figure 2: The boxplots indicate the distribution of the quality benchmarks globally and just for the committed area. A percentage of locations exceeding each of the three thresholds is indicated as well.

Signal-to-Noise-Ratio (SNR)

The Signal-to-Noise-Ratio (SNR) is shown in Figure 3 and in Figure 4 for the committed area.

Figure 3: SNR of the test data set.

Figure 3: SNR of the test data set.

Figure 4: SNR of the test data set for the committed area only.

Figure 4: SNR of the test data set for the committed area only.

Pearson correlation coefficient (R)

The Pearson correlation coefficient (R) is illustrated in Figure 5 and in Figure 6 for the committed area.

Figure 5: Pearson R between the test data set and ERA-Interim.

Figure 5: Pearson R between the test data set and ERA-Interim.

Figure 6: Pearson R between the test data set and ERA-Interim for the committed area only.

Figure 6: Pearson R between the test data set and ERA-Interim for the committed area only.

Discussion and Conclusion

The SNR and Pearson R indicate a good performance for the committed product area, except for parts of North America, Northern Europe and Western Australia. On a global scale, a lower performance of the test data set can be found in areas with low soil moisture dynamics (e.g. deserts) or at higher latitudes (see Figure 3 and Figure 5). In the latter case, frozen soil and snow cover make it difficult to retrieve reliable soil moisture information. Therefore, in these regions only summer months can be used for validation.

Looking at the distribution of the results and comparing them against the threshold/target/optimal requirement shows that more than 75% (SNR: 79%, Pearson R: 79%) of the locations are exceeding the minimal threshold and more than 50% (SNR: 62%, Pearson R: 54%) are above the target threshold for the committed area (see Figure 2). Only a small percentage of regions are below the threshold requirement.

In conclusion, the test data set under investigation has successfully been validated and can be reviewed for an official release.

References

[1] “Product Validation Report (PVR) Soil Moisture, Metop ASCAT Soil Moisture,” Doc. No: SAF/HSAF/CDOP2/PVR, v0.3, 2016.

[2] “Product Requirements Document (PRD),” Doc. No: SAF/HSAF/CDOP2/PRD, v1.4, 2016.

[3] “Product User Manual (PUM) Soil Moisture Data Records, Metop ASCAT Soil Moisture Time Series,” Doc. No: SAF/HSAF/CDOP2/PUM, v0.3, 2016.

[4] “Algorithm Theoretical Baseline Document (ATBD) Soil Moisture Data Records, Metop ASCAT Soil Moisture Time Series,” Doc. No: SAF/HSAF/CDOP2/ATBD, v0.3, 2016.

[5] A. Gruber, C.-H. Su, S. Zwieback, W. Crow, W. Dorigo, and W. Wagner, “Recent advances in (soil moisture) triple collocation analysis,” International Journal of Applied Earth Observation and Geoinformation, vol. 45, pp. 200–211, Mar. 2016.