View land cover for selected dates, with an option to select an individual class.
NASA Earth Observatory image modified by SkyTruth
At SkyTruth we felt this important dataset should be made accessible to organizations, volunteer monitors, and activists who may not have the access or training to use the raw dataset. To that end, we’ve added several DW features to our Alerts platform, all available from the Dynamic World sidebar. Here is a summary of the core functions:
We’ve also added these features to help you make the most of DW:
Examples of using Dynamic World in Alerts
Compare land cover from two time periods
After moving the map to your AOI and zooming in:
- Click the Dynamic World tab, check the Land Cover option, and enter your date range.
- Click Find.
- Click the Satellite Imagery/Compare Maps tab and then the Set Map 1 link.
- Click the Dynamic World tab, select the Land Cover option, and enter your second date range.
- Click Find.
- Click the Satellite Imagery/Compare Maps tab and then the Set Map 2 link.
- Select a type of comparison. This example uses Stack Horizontally from the drop down list.
- Click Compare.
To add labels, click the Layers tab and open Map Labels.
Find tree cover for current map view
How we created this map:
- The concept can be applied anywhere, but if you’d like to recreate this particular map view, paste 39.32,-77.82 into the Location search field (upper right on map) and zoom to level 12.
- Click the Dynamic World tab, select the Land Cover option, and enter your date range.
- Check Show Sentinel-2 mosaic for filtered dates.
- From the Land Classifications dropdown menu, select Trees.
- Click Find.
Adjust the Opacity slider to see through the DW layer to the Sentinel-2 imagery.
Identify new oil & gas drilling in the Permian basin
How we created this map:
- The concept can be applied anywhere, but if you’d like to recreate this particular map view, paste 31.53, -102.69 into the Location search field (upper right on map) and zoom to level 14.
- Click the Dynamic World tab and select the Change Detection option.
- Enter both before and after date ranges. This example uses Jan 1 thru April 2 for both 2021 and 2022.
- Check Show Sentinel-2 mosaic for filtered dates.
- Check Select my own filters under Change Detection.
- Complete the filter by selecting Bare as the land cover and 20 as the % increase.
- Click Find.
- The example shows a slider comparison with red areas showing detected change.
- Adjust the opacity on the Dynamic World tab as needed to see the underlying imagery.
Create a time series of land cover in the Las Vegas Bay area
This chart shows the decrease in the area covered by water (blue line) since 2018 for the Las Vegas Bay area.
How we created this map:
- Move around the map and zoom in as needed. For this particular example, I created and saved a rough outline of the Las Vegas bay area by using the Polygon tool available in Explore a New Area from the My Areas sidebar.
- Click the Dynamic World tab and then the Time series of class probabilities button.
- Select Each step = 1 Year from the drop down list.
- Select the before/after month/years.
- Click the Play button (and be patient!).
Identify new built areas
How we created this map:
- The concept can be applied anywhere, but if you’d like to recreate this particular map, paste 23.605,113.0304 in the Location search field (upper right on map) and zoom to level 14.
- Click the Dynamic World tab and select Change Detection.
- Enter both before and after date ranges.
- Check Show Sentinel-2 mosaic for filtered dates.
- Select Select my own filters under Change Detection.
- Complete the filter by selecting Built as the land cover and 20 as the % increase.
- Click Find.
- The above screenshot shows a vertically stacked comparison with red areas showing new built land cover.
- Adjust the Opacity on the Dynamic World tab as needed to see the underlying imagery.
FAQ
How often is Dynamic World updated?
DW is based on Sentinel-2 satellite imagery. Sentinel-2 collects new imagery at least once every 5 days, but Dynamic World only processes images with less than 35% cloud cover. So in cloudy areas, updates could take weeks or months.
Where does Dynamic World perform best?
According to model developers, DW most accurately classifies relatively clear-cut land cover classes, such as:
- Forests
- Water
- Built area
- Ice and snow
Where does Dynamic World struggle?
Model developers note that DW often struggles to accurately distinguish land covers that are more mixed and ambiguous in appearance, such as:
- Bare ground
- Grassland
- Scrub/shrub
- Flooded vegetation
At SkyTruth, we’ve also noted:
- Change Detection is greatly affected by snow-and-ice land cover. As the probability of a scene having snow increases, there’s an equal reduction in the probability of other land covers.
- Given the 10m resolution of Sentinel-2, DW can misclassify smaller footprints, especially when in mixed land cover areas. For example, buildings in densely wooded areas will often show up as part of the tree cover.
- DW can have difficulty classifying rocky areas.
Why was Dynamic World added to SkyTruth Alerts?
- While you might be able to apply DW features to a single date, you’ll usually need a date range of a month or several months to get good minimum cloud coverage data.
- Annual date ranges, while slower to process, can smooth out fluctuations by creating average probabilities over a longer period for each land cover classification.
- You can often get a good result by building your own probability filters for Similarity Search or Change Detection. However, inclusion/exclusion of even a single scene (a scene is for a single date) can change those results. Experimenting with date ranges and times of the year is key when looking at local areas.
- Same for different habitats — you might create a nice set of parameters for viewing built areas in the desert, but when you run it in a hardwood forest you get worse results.
- While DW does filter out cloud cover, it is not always accurate. Sometimes missed clouds will present as snow and ice land cover.
- Running a Time Series can be a big hit on resources, so is currently available only at zoom levels 12 or higher. If the time series fails to build, try selecting a smaller date range, a smaller time cadence (e.g., two months instead of six), or a smaller area.
How accurate is change detection and identifying similar features in Alerts?
DW will not perform as well as a dedicated ML model for similarity search or change detection, but it can be very useful in analyzing trends and identifying areas that need further evaluation.
Where can I read more about the Dynamic World model?
You can read a nice introduction to DW in this article by Tanya Birch, Google Earth Outreach’s Senior Program Manager. For a discussion about the methods used by DW modelers to develop and validate the dataset, this is a good read.
Is there a recommended workflow for using your own probability filtering for Similarity Search and Change Detection?
- Start by entering your date range(s) and checking the Show Sentinel-2 mosaic for filtered dates checkbox
- Add a single filter. For example, the Built land cover for a 30% increase in Change Detection. Click Find to see results.
- Check the Show DW pixel stats on map click checkbox.
- Repeat as needed:
- Click a map pixel to see how DW classifies scenes at that location. (For Change Detection, click on the right map while viewing map comparisons.)
- Use this additional information to adjust the existing filter or add a new filter. For example, if pixel stats show a probability for trees of 60% and you want to filter for trees, you should include the Trees land class with a somewhat high minimum probability.
- Click Find to see results.
What is unique about the way Dynamic World calculates land cover?
Most land classification models assign one land cover classification to an area. A major benefit of the DW approach is that for every incoming Sentinel-2 satellite image and for every pixel in the image, the model estimates the probability that the pixel is tree cover, or grass, or a built area, etc. So instead of assigning Trees as the land class, it might say there is a 72% chance this pixel represents tree cover. These probabilities come into play not only while visualizing DW land cover, but also when determining when change has occurred or when matching on finding similar areas in an image.
