Step 1 - Choose the data you would like to validate - including the dataset name, the version of the dataset, and one of the soil moisture variables provided in the dataset. More details on the supported datasets can be found here.
Step 2 [optional] - Choose the criteria by which you would like to filter this dataset. The filters available depend on the data contained within the chosen dataset. For example, you can filter the C3S data to include only data with no inconsistencies detected (flag = 0). Details of the filter options provided for each dataset are given on the supported datasets page here. You can also hover your mouse pointer over the question mark next to a filter to get a short explanation.
Step 3 [optional] - If you want to intercompare several datasets, you can add more datasets to the validation
using the + button, up to a maximum of five. Configure the settings for the additional datasets by selecting
the respective tab and repeating steps 1 and 2 above.
Intercomparison: The intercomparison mode of QA4SM validates up to five satellite data sets against a common reference data set. For each reference location (e.g. each ISMN station) it finds the nearest observation series in all selected satellite products. All observations series are then scaled (if selected) and temporally matched to the reference series. For validation only the common time stamps (that are available in all satellite products) are used to calculate validation metrics between the reference and each individual satellite product. This way deviations in the metrics due to different temporal coverage are excluded and validation results represent differences in the performance of the compared satellite products.
Step 4 - Choose the reference dataset you would like to use for the validation including the dataset name, the version of the dataset, and the soil moisture variables provided in the dataset. More details on the supported datasets can be found here.
Step 5 [optional] - Choose the criteria by which you would like to filter the reference data prior to running the validation. The filters available depend on the data contained within the chosen dataset. For example, you can filter the ISMN data to include only data points where the soil_moisture_flag is "G" for "good". Details of the filter options provided for each dataset are given on the supported datasets page here. You can also hover your mouse pointer over the question mark next to a filter to get a short explanation.
Step 6 - Choose the date range over which the validation should be performed. Accepted formats are: YYYY*MM*DD or DD*MM*YYYY where * can be any of ".", "/" or "-".
Step 7 - Choose how the data (or reference) will be scaled before metrics calculation. The data can be scaled to the reference (default) or vice versa. Note that in an intercomparision validation (with multiple datasets), only scaling to reference is possible. The scaling method determines how values of one dataset are mapped onto the value range of the other dataset for better comparability.
Step 8 - Optionally name your validation results to make it easier to identify it in the list of all your validations.
Step 9 - Run the validation process. You'll be notified via e-mail once it's finished. You don't need to keep the results window (or even your browser) open for the validation to run. The email will contain a link to your results.
The list shows all your validations, including the currently running ones, sorted by date (latest first). The buttons on the right-hand side of each validation have the following effects:
Once the validation process is finished, you can see short summary of the validation run on the results page.
The following metrics are calculated during the validation process:
|Pearson's r||Pearson correlation coefficient|
|Pearson's r p-value||p-value for pearson correlation coefficient|
|Spearman's rho||Spearman rank correlation coefficient|
|Spearman's rho p-value||p-value for Spearman rank correlation coefficient|
|Root-mean-square deviation||Root-mean-square deviation|
|Bias (difference of means)||Average Error|
|# observations||Number of Observations|
|Unbiased root-mean-square deviation||Unbiased root-mean-square deviation|
|Mean square error||Mean square error|
|Mean square error correlation||Mean square error correlation|
|Mean square error bias||Mean square error bias|
|Mean square error variance||Mean square error variance|
Visualisations of these metrics are displayed in the Result Files section of the page: boxplots and geographical
overview maps. You can select the metric shown with the left drop-down button below the graphs.
For an intercomparison validation, all boxplots are combined into one graph. The dataset displayed in the overview map can be selected with the drop-down button on the right.
You can also download a zipfile of all the plots in png and svg (vector) format by clicking on the Download all graphs button, and the result NetCDF file with all metrics with the Download results in NetCDF button.
If you want to email us to send comments, report errors, or ask questions, you can do so at qa4sm (at) awst.at .Back to top