This guideline provides information about the Small Water Body Mapping (SWBM) application of the Hydrology Thematic Exploitation Platform (HTEP). Simultaneously this document provides a hands-on tutorial showing you how to produce water masks of your area of interest using the application in combination with the features of HTEP.
The Small Water Body Mapping Service developed by TRE ALTAMIRA provides the extent and temporal evolution of permanent and non-permanent water bodies in the form of detailed mapping, time series and evolution of water bodies’ surface in time. The service is based on Sentinel 1 SAR acquisitions and ASAR for historical analysis as well as the possibility of results refinement using Sentinel-2 Optical Data. The usage of EO data means low costs and low labour intensity, easy mapping of large areas and a high spatial resolution. Furthermore near real-time monitoring is possible using satellites with high temporal resolution and water bodies in physically inaccessible areas can be easily mapped.
As input for the SWBM application, there is EO data available of multiple satellite sensors with varying spatial, temporal and spectral resolution. For the SWBM application there is currently data available from Sentinel -1 GRD, Sentinel-2 and Envisat ASA IMP 1P.
Sentinel-1 High Spatial , High Temporal Resolution
The resolution of Sentinel-1 in ground is approximately 20x5m. TRE ALTAMIRAs SWBM application uses Ground Range Detected (GRD) images for processing. These images are already pre-processed; they are corrected and prepared in ground range projection. Once projected with a Digital Elevation Model (DEM) the images can achieve 10x10 meters, as some data is interpolated below the initial resolution. To ease the processing and the volume of data storage, the spatial resolution of the SWBM output products is 20x20 meters. Sentinel-1’s Synthetic Aperture Radar (SAR)-sensor works in the microwave spectrum, specifically in C-band. C band has a wavelength of approximately 5.6 cm and the frequency of the satellites is around 5.405 Ghz. For more info about Sentinel-1, please visit https://sentinel.esa.int/web/sentinel/user-guides/sentinel-1-sar/resolutions. The revisit frequency of Sentinel-1 is, due to ascending and descending and overlap images, approximately 1-3 days depending on the latitude. As such Sentinel-1 has a high temporal resolution.
ENVISAT (ASAR) High Spatial , Monthly Temporal Resolution
The specifications of ENVISAT satellite and its data products are comparable to those of Sentinel-1 as it also uses the Advanced Synthetic Aperture Radar (ASAR)-sensor. ENVISATs sensor therefore works in the microwave spectrum just like Sentinel-1 SAR-sensor. The SWBM app product output using ENVISAT-data products is also 20x20 meters. For more specification of ENVISAT ASAR please visit http://envisat.esa.int/handbooks/asar/toc.html. The main difference with Sentinel-1 is the considerably lower temporal resolution of ENVISAT with a revisit frequency of approximately 35 days. As such ENVISAT has an average temporal resolution. In the context of HTEP, ENVISAT data products can be used for historical analysis of the evolution of (non-)permanent small water bodies. HTEP ENVISAT data is currently only available from 2009 until 2011.
Sentinel-2 High Spatial, Weekly Temporal Resolution
Sentinel-2 works not in the microwave spectrum but in the multi-spectrum: the visible spectrum and infrared spectrum. Each band of Sentinel-2 has its own wavelength and frequency, hence the resolution is dependent on the used band also. Some bands have a resolution of 10x10m whereas others have a resolution of 20x20m. The SWBM application uses 20x20m resolution optical images to refine the water masks obtained from Sentinel-1 and ENVISAT SAR data products. Sentinel-2 has a monthly revisit of approximately five times, meaning a weekly temporal resolution. For more information about Sentinel-2 please visit https://sentinel.esa.int/web/sentinel/user-guides/sentinel-2-msi.
As mentioned in Section 2.6.1.1.1, Sentinel-1 and ENVISAT make use of sensors working in the microwave spectrum. Therefore obstructions such as clouds do not affect the SAR images. SAR is optimal for instance in areas with huge cloud coverage (floods in hurricane storms and in wet season in several areas with large amount of rainfall) where the optical sensors are very limited. As such there is no need to take into account obstructions such as cloud coverage when you pick data products for analysis from either Sentinel-1 or ENVISAT. However, Sentinel-2 data is used to refine the water masks using optical sensors operating in the multispectrum. As those sensors use the backscattering of light within the visible spectrum, wavelengths cannot properly penetrate obstructions such as clouds and haze, meaning those kind of obstructions sometimes result in difficult measuring conditions. The quality of refinement is therefore dependent on the cloud coverage in the Sentinel-2 images.
As the name of the application indicates, the output of this application is a water mask mapping the (non-)permanent water bodies in the analysed area. The output of this application can also be used as input for others applications on HTEP, such as the Water Level application discussed in Quick Start Manual 7. Figure 2 shows the water mask of the river between Viet Tri and Hanoi, in the Red River basin. In this tutorial those results will be reproduced.
This chapter contains a hands-on tutorial how to work with TRE ALTAMIRAs SWBM application on HTEP. The tutorial shows and explains step-by-step the different features of HTEP and the actions to be taken in order to create the small water bodies map of Figure 2. For this tutorial, the area of interest is the part of the Red River between Viet Tri and Hanoi, in the Red River basin.
You do not have an account yet? Then first register on the platform. To register at the platform, it is advised to follow the steps in the Quick Start Manual How to become a user of HTEP, which can be found under the Quick Start-tab in the menu of the HTEP Community Portal.
Once the SWBM application has been accessed, a new tab opens called the Geobrowser. This part of the tutorial will teach you to work with the various features and functions available within the Geobrowser. Currently the default map is of Northern Vietnam and Southern China: the Red River basin. The default map shown upon opening the SWBM app may change in the future.
As discussed in Section 2.6.1.1.1, the SWBM application has currently data available from Sentinel-1, Envisat, and Sentinel-2. The EO data to be selected depends on your requirements and research purposes, as each satellite has its own specifications suiting different requirements. Sentinel-1 and Envisat have relatively similar specifications and are the main input parameters for analysis: Sentinel-1 for monitoring present water bodies and Envisat for historical analysis. Sentinel-2 products are optical data products used to refine the analysis of the main input products.
For this tutorial, Sentinel-1 data is selected.
➢ Search Field (4a): On the top-left of the Geobrowser, you see a search field. In this field, you can do a text search for specific EO data products within the data source chosen in step 3. For now this field is left blank.
➢ Clicking on the magnifying glass (4b) below the search field, opens the Search Panel of Figure 8a: a panel containing multiple additional filters to find your desired data product. For example the productType and a time range filter. Show Other Parameters opens another extensive list of filters, amongst others track and orbit of the image, cloud- and land cover filters and geometry filters for a spatial search.
➢ A big amount of data will originally be shown on the screen and in the list when a polygon is drawn. To limit the amount of images in the search and ease the step of Selecting images, it is recommended to filter the images by the track number. The track is a marker, which is specific of all the images with the same geometry of acquisition, and thus will have a similar footprint, incidence angle and the same pass. The service needs at least two images of the same track to work correctly.Different tracks can be combined, but at least each of the tracks has to contain a pair of acquisitions/dates to produce outputs.
➢ The track number can be obtained by clicking on one of the images on the geobrowser and check the number enclosed in the panel list or in the pop-up window, such as it is highlighted on Figure 8b.
Note
The service needs at least a pair of images of the same track to work correctly.
➢ Once the track is obtained, the search can be filtered by this parameter on the “Show Other parameters” options (see Figure 8a). Additional parameters and options will be disclosed and the user will be able to introduce the number of track in the tag and applying the search such as it is done in the following example of Figure 8c.
➢ Although the Search Panel already provides you the option for a spatial filter based search, you can also apply a spatial search through the tools of 4c. A polygon, rectangle, marker and well-known text (WKT) code can be used to define your area of interest. For this tutorial a spatial filter is applied using a WKT-code. Click on the WKT button: the pop-up of Figure 9 appears.
As you can see a spatial filter can be applied using a WKT-code, but also by simply dragging and dropping a Shapefile or uploading a Shapefile from your computer. For now a WKT-text is used. Copy and paste the following code in the top field: POLYGON((105.437 21.365,105.33 21.326,105.314 21.224,105.407 21.172,105.531 21.115,105.756 21.067,105.807 21.182,105.646 21.238,105.529 21.229,105.437 21.365)) and click on OK. This WKT code is the area of interest, the part of the Red River between Viet Tri and Hanoi: the river should now be boxed by a pink dashed line.
➢ Now also a time filter is applied. The time filter can be applied not only through the Search Panel, but also using the tool of 4d indicated in Figure 7. The slider at the bottom is a time filter that can set by sliding the begin and end date to the desired time range. For now drag the left side of the time filter to 2017-03-01 and the right side of the time filter to 2017-03-31. Alternatively use the Search Panel.
Figure 10 shows the search results from Section 2.6.2.2. Now the data products of interested will be selected and saved in a Data Package.
For this tutorial, select the following three products from the current search results and transfer them to the features basket: S1A GRD IW_DP L1 VV, VH 128 2017-03-20T11:05:53.4115860 (20 March 2017), S1A GRD IW_DP L1 VV, VH 91 2017-03-17T22:50:33.49198502 (17 March 2017) and S1A GRD IW_DP L1 VV, VH 128 2017-03-08T11:06:00.3705280 (8 March 2017). Change the satellite data source from Sentinel-1 to Sentinel-2 to see if there are usable optical data products for result refinement. Three data product appear: click on them for additional information. As you can see, all data products have a cloud coverage of >98%. As such refinement using Sentinel-2 data is impossible for this exercise.
The advantage of a Data Package is that you can easily load your data products of interest at any arbitrary time and you can also easily share it with other hydrologists.
The available options are:
- Do not share: Default setting, meaning your data package is only visible for yourself.
- Share with anyone: Share your data package with all other HTEP users.
- Restricted sharing: Share your data package with a limited number of users, for example only a specific user(s) or with users from the same community.
For now, leave your data package at default settings (Do not share) and Close the pop-up. In the list of public Data Packages there should be a Data Package called SWB_VietTri-Hanoi_March17_RR-Tutorial. This Data Package was created and published for the purpose of this tutorial. Please click on load: the products from this Data Package are loaded. It contains the same products as your own data package, and two additional products: A Sentinel-2 product and another Sentinel-1 product from 2016.
Section 2.6.2.2 and 2.6.2.3 explained how to search for and select your data of interest within the Geobrowser. Having the relevant data selected and saved, it is now time to process this data to obtain the desired product output.
➢ Services: This tab yields a list of available processing services (the different models and algorithms within the application). Currently only the Small water bodies processing service is available, but this number will increase in the future.
➢ Search Field: Once the number of available processing services has increased, the Search Field can be used to filter only those processing services of interest.
➢ Jobs: This tab lists all existing jobs. The jobs shown are the jobs you have created yourself or the jobs who have been published by other HTEP users.
➢ Job title: Give your Job a title, for instance SWB_VietTri-Hanoi_March17_username. Any other name with arbitrary length and symbols is also allowed.
➢ Satellite constellation: Here you define from which satellite you are going to analyse products. For this tutorial this is Sentinel-1.
➢ Area of interest: Define your area of interest. Click on the arrow and a dropdown menu appears. Pick bbox (bounding box) - geometry to define the area used as a spatial filter in Step 4 of Section 2.6.2.1 as the area of interest. The WKT-code will appear in the field.
➢ Images to compute small water bodies: This are the images that will be used to analyse the small water bodies. Drag all Sentinel-1 products from your features basket to this input field.
➢ Sentinel-2 L1C: This is the input field in case you wish to use Sentinel-2 products for refinement of the results. The field can be kept at NONE if no refinement is needed, but for this tutorial please drag the Sentinel-2 product from the features basket to this input field.
➢ Temporal baseline: In this field it should be indicated what maximum time difference is allowed between the input Sentinel-1/ENVISAT data products and the Sentinel-2 data products. If the differences in water bodies between both data products are too large, it is not useful to use one product to refine the results of another product. A suitable baseline depends on the area and the event. The baseline is by default 48 hours, but for this tutorial it is set to 300.
➢ If desired, you can share your processing service on social media with the Share-icon above Job Title.
The previous section showed how to process the data products obtained from Section 2.6.2.2 and 2.6.2.3. Once the process is finished, which may take a considerable amount of time, the results can be visualized and possibly shared with others users and/or your community.