Planetary Response Network: Hurricane Dorian

Why are we marking these features?
To find out more about why we are asking you to identify these particular features, check the About page (tab at the top of the page).

What does Rescue Global do with our classifications?
Once our team aggregates everyone's responses into a map with the consensus about each area, we send those to RG for use on the ground. Their team uses them for "situational awareness", which is a general term but, essentially, more is better! Having better situational awareness helps RG (and other responders) make more informed decisions about how to allocate aid and resources. For specific examples of how volunteers' classifications have been used in the past, check out our Results Page for the 2017 PRN Caribbean Response.

Because of how many people pitch in to help, and because we tend to use more zoomed-out images than, say, an aerial survey from a plane, we can often provide maps covering a larger area than would otherwise be available. Even though that means the details are sometimes harder to see, a coarser map showing the evolving situation regarding large-scale features like floods, landslides across roads, etc. is very valuable.

I need some examples!
Check the Field Guide (tab to the right of the screen). As the project progresses, Talk may also be a good source of examples. The team will usually 'pin' the most helpful/informative posts so they always show up at the top of the page.

I don't see damage, but I could have missed something.
We know that not all images are high-resolution enough to spot all kinds of damage. For instance, in 3-meter resolution images, you can spot major road blockages and catastrophic structural damage, but if there's only partial damage to a roof, it will probably not be visible. When you click "No Damage Detected", you are saying that within the limits of the image quality you do not detect damage. You are not promising that you are 100% sure there is no damage whatsoever. We consider the image resolution and quality when creating damage maps, so that responders on the ground also know what we can't see.

Which image is 'Before' and which is 'After'?
These are usually explicitly labeled, but not always. If there is no label, before is the first image (left button filled) and after is the second image (right button filled).

How should I mark extended/odd shapes?
If a flood, blocked road etc. is larger than the point marker, please just mark its center. Only add a second mark if it looks like a separate flood, blockage etc. Sometimes the shape of a flood (for example) makes the center hard to identify; just do your best to estimate its location. If there are a lot of damaged buildings, please evaluate each one individually.

The image is only sea/ocean
We've tried to remove as many of these as possible, but there are still some ocean-only images. Click the "This image just contains ocean" tickbox and then hit done to flag these images so that our team can make sure they're retired quickly from the active data set.

The image is unclassifiable due to cloud/missing image areas
If the image is completely unclassifiable tick the "This image is unclassifiable (cloud cover, missing image areas, etc.)" box below the marking options and click "Done". If only part of the image is unclassifiable, just classify what you can see.

Note: The imaging software may try to mark clouds with brightly colored regions, and it very rarely marks non-cloudy portions, but it often misses clouds.

My marks are getting in the way of each other.
If there are a lot of things to mark, you can use the "Hide previous marks" option to make the image easier to see. Please be careful not to double-count anything.

These images are very hard to classify. I'm getting frustrated.
If you're getting frustrated or stressed, it's ok to take a break! With many people working together, everyone can step back when they need to and step in when they're ready.

We know some of these datasets are very challenging, but that's exactly why we need your help! Computer algorithms aren't good enough on their own to do this, but people are very good. We know from past projects that our community is very accurate at spotting even subtle signs of changes, and telling those apart from differences that are only due to varying conditions like time of day (shadow length), cloud/haze cover, or how directly overhead the satellite was when it took the picture. Check the examples, ask questions on Talk, and trust your instincts!

Thanks for your help.