This page features troubleshooting questions students have asked for Remote Sensing Foundations Module 4: Image Interpretation: Quantifying, Masking, and Classification
Question: At the end of the file when you are selecting an image for a certain day with Landsat 5, say at the end you see 20000606 in the code we build following the tutorial F1.1. This selects June 6, 2000, but I can’t tell from this example if it is YYYYDDMM or YYYYMMDD. The later makes more sense to me, but I can’t find it online in a clear way, because maybe Google fixed all the images to be more uniform with a certain format?
Answer: Its YYYYMMDD
Question: I worked through the tutorial in F1.1 and I have a question about the visualization parameters described on page 23. In #3 on that page, “min, max” — The text says these numbers (here set at 8000 and 17000) represent the min and max value for band SR_B1. Where can I get that information for a different band or dataset? When I look in the bands information in the GEE catalog for this dataset (at least I think I have the right one) the min and max are 1 and 65455, respectively. Where do 8000 and 17000 come from? Here’s a link to the catalog page.
On the other hand, on pages 35-36, it describes the range for the nighttime lights data as coming from the catalog or ” with other Earth Engine methods. ” When I searched on this dataset in the catalog, it said it was not available, but when I did my own search for using similar terms, I came up with what I think is a match to this set. It’s min and max DO match what is listed on the bands tab in this catalog entry (0 and 63).
So why for the Landsat 5 are the numbers different from what is listed on the catalog page for min and max?
Answer: I think you are confusing image datasets. On pg 35 there is the line:
var lights93 = ee.Image(‘NOAA/DMSP-OLS/NIGHTTIME_LIGHTS/F101993’);
This is where you are importing a different dataset than Landsat 5, as you used earlier in this chapter to make the composites. Now, you are importing a new image to look at the nighttime lights. The range of this is as you say, found in the band information on the EE catalogue page. The next few lines in the tutorial are having you look at different night time images to compare.
Question:I loaded the nighttime lights dataset you reference. And that all works fine and the min and max match what I think is the correct catalog listing for that one. My question is about the Landsat 5 dataset on p. 23(ish) — why are the min and max 8000 and 17000? Those are not the numbers in the band tab in GEE catalog for that dataset: The min and max for that dataset at 1 and 65455 in the catalog, respectively. But we set them at 8000 and 17000 in the tutorial code. Why?
Answer: I’m pretty sure this is done to skim off the minimum/maximum values, so you have more median values. Thus, the values are less linearly stretched and the bands now have “assignments” on how to organize themselves. If you put in 1 – 65, 455; you’d be jumbling up the data and there are no clear assignments for the bands. When you add 8,000 – 17,000; it also means that everything below 8,000 and everything above 17,000 now have an assignment.
Question: Ok but where did the numbers 8000 and 17000 come from? Why those numbers? How can I find such numbers to do this with if they are not in the metadata or catalog (as 8000 and 17000 are not)? Trying to figure out how to swim on my own!
And what do you mean by assignments for the bands? In this case the bands of interest all have the same min and max numbers — either the 1 to 65,455 as listed in the catalog or the 8000 to 17000 that are assigned in the tutorial. So how does that help split things out into bands if they all have the same min and max numbers?
AND what is the assignment that goes to everything above and below 8000 and 17000?
If I’m asking too many questions, please just direct me toward a reference or something. Again, just want to be able to do this on my own and I have no idea why 8000 and 17000 are the numbers that were used in this example. Would like to figure that out so I can do it myself. Thanks.
Answer: I don’t think I can fully answer your first parted question; these are just values that the tutorial decides is best. Most tutorials do this and make you input certain numbers at certain places.
Second part of your question is actually answered in that image you posted last. In step 3, it mentions “The values between minimum and maximum are mapped linearly to grayscale between black and white. Values below 8000 are drawn as black. Values about 17,000 are drawn as white.”
I know you are ambitious to get all of this and swim on your own. It’s great to see your interest in these topics. For now though, we’re simply following the tutorials and getting the fundamentals down which are; concepts, code and familiarization with tools/ideas. Those numbers chosen to set this range won’t come from an experienced user until probably years later.
Question: Thanks! I’ll take a look and glad to hear that we’ll be doing more of this later. That’s really good to know. Right now, I am working on this week’s assignment and having bigtime issues with this — part of why I’m asking. I’m trying to look at MSI data for an area of interest for the assignment and when I add the data to the map to see it, it’s all white! I’m using the min and max used in the tutorial for MSI data on p. 85 (0 to 2000). I’ve been messing around with those numbers, but I can’t seem to change much. And when I do — the numbers I put in there seem to have no relationship to what’s listed for those bands, and all the active farm fields are green instead of red (I’m making a IR false color composite). Totally confused and feel like I’m shooting in the dark to randomly pick these numbers for visualization. I’ll see if this helps. But let me know if you have any clues!! Here’s my code. I put the layers on the map — just to take a look at them — at the end. I’m also having trouble with another question I’ve posted here — how do I know which image is displayed by date in the map when I do this? It would be really helpful to know the date to understand it in terms of the crop year. I’m using the “first” function (I think!).
Thanks. That’s a very helpful perspective. I’m happy to do trial and error (though I still have an all white map!). I just feel like — when I’m guessing — that I don’t know what is ok to do. I could put in random numbers until something looks right, but will that be meaningful when anyone else looks at it? Will it be comparable to other maps made with the same information? I guess I’m asking — are there standard practices (for example, for a false color IR map)? Or is it ok to shoot from the hip in how I visualize it?
Answer: With the white image is to load it into the map and do a 100% stretch on the visualization box…the tutorial has us coding everything thing, hard code, but you can also find the range by manually doing it on the map by clicking on the settings icon of the layer. By doing it manually, you don’t have to hard code anything. Hope this helps!
Question: I think I get it. Do you mean that by clicking on the 100% stretch in the visualization settings — that will fill the range (min and max) into the boxes? And then use those numbers? (Not sure what you mean by hard coding exactly.)
Also — I guess the min/max numbers in the catalog aren’t very useful then? That’s what I thought I should use. What do you think? Ignore them?
Answer: Great! And yes I meant by clicking the 100% stretch on the visualization settings. Hard coding is just using strictly code to get what you want – whereas here, you can use the settings directly built into GEE and use those numbers to fill min/max boxes.
The min/max catalog is only useful depending on what you’re trying to do. In this case, you can ignore them or choose to ignore them based on what you’re trying to perform.
Final Project Ideas, Q & A
Question: Identifying social hiking trails on federal public lands:It seems like remote sensing is a perfect tool to identify the prevalence of social trails (non-sanctioned hiking trails) when paired with a high enough resolution dataset. Would it be possible to apply this to a place like Joshua Tree National Park, CA or Calico Basin Recreation area next to Red Rocks in NV? Would anyone have suggestions for how to conduct this process or further how to measure the mileage of social trails vs. approved trails? Is that even a function in GEE? Potential for Wildfire impacts to recreation resources: Curious if folks have any ideas regarding looking at historical wildfire in a region and comparing that to current vegetation in a landscape? This could be useful or considering active forest management plans and comparing with recreation resources on the ground?
Answer: I may be able to help a little with this one. I was lucky enough to do multiple timber cruises/inventory for state forests here in CT along with other foresters. One thing that is always reoccurring with ideas within Forestry are: invasive species, recreational use, fires and certain tree regeneration capacities. That being said, the idea you have are good but let me know if something really stands out to you or if you can zero in on something more specifically. This way, we can circle back and see what can work. I think these ideas are both doable, especially the wildfire data.
