nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Any help with using the SAR extract water tool?

D_Hornby_ddh_uni

Could anyone in ESRI or the wider community help with this seemingly simple problem of running the extract water tool? Like this user I too get a generic 999999 error.

I have a sentinel 1 dataset, its full name is: S1A_IW_GRDH_1SDV_20200209T110313_20200209T110342_031175_03959E_A378_COG.SAFE

My starting point is to drag the manifest.safe in catalog into the map and it displays as an RGB layer.

I then stepped through the processing sequences as listed in this rather good story map. No errors all seems to work OK.

I then want to run the SAR extract water tool and this is where it all falls apart and there is no guidance as to what the input should be. So when I run the tool it simply errors with 999999. Reading the help file for this tool it states Calibrate the input radar data to gamma nought using the Apply Radiometric Calibration tool.

OK if I restart the workflow as prescribed in the story map and do a gamma nought radiometric calibration then the next step apply radiometric terrain flattening refuses to run demanding input must be beta nought calibrated!

OK so I think maybe water extraction is something you do before terrain correction? So off I go and look for help. One very quickly arrives at this esri youtube about running the extract water tool, excellent yes?...No! So the presenter shows the tool working. The only comment posted which esri have not bothered to respond to (why?) is a very obvious question of what's the actual input into this amazing tool?

Now from that storymap I have learned that the naming convention helps you understand the processing steps that have been applied to the raster. So looking carefully at the video one can see the raster has been beta nought calibrated, something called deburst, terrain corrected then despeckled. My first question is in their example why wasn't the thermal noise not removed as that is the first processing step prescribed in the story map?

Back to help file, no obvious pointers, even asked esri's ai assistant, that was useless. So I stand stranded between having done the preparation steps and the tool.

Does anyone please have any insight into what exactly is the processing sequence needed so that the input raster is valid for the extract water tool. If its something different requiring a different workflow to ensure the data is suitable for the tool why is this not documented in the help? For the record I'm using arcpro 3.6.2.

Najděte více příspěvků označených jako

Image space

Accepted answers

D_Hornby_ddh_uni

Good news got the tool working with help from @Aote_Xin !

So lessons learned. 99999 errors are not very helpful particularly when I (yes it was me who was at fault) made a blunder in my choice of DEM. I'm hoping that esri will improve the error reporting of the tool and the usage advice in the help file so other future users don't make the same mistake as me.

So only when @Aote_Xin provided a DEM that worked were we able to see that basically I had downloaded an inappropriate dataset for the tool. I had downloaded a NASADEM from here. This is not an elevation dataset! After digging around in their user guide table 1 on page 3 I determined the NASADEM_SIM was incorrect and I should have been using their NASADEM_HGT. The NASADEM_SIM is an 8bit 0-255 value raster the sort you might use for a grey scale display. The NASADEM_HGT is an integer raster and appropriate for the Extract water tool.

Also it seems to be best practise but not enforced that the DEM extent should enclose the extent of the SAR dataset.

With a correct DEM and following the sequence:

Apply Radiometric Calibration (Gamma Nought)
Apply Radiometric Terrain Flattening

Extract Water runs successfully.

So where does one source correct DEMS? You could, as I had, go to the primary portal such as EARTHDATA but an alternative worthy of mention is https://opentopography.org/

It's worth mentioning that the YouTube video above gives one the impression that the tool runs super fast, that is definitely not the case for me on an i7, 32GB machine it took about 6 minutes and tinkering around with the minimum area parameter can significantly increase processing time.

When you finally get it to run its pretty awesome! 😁

All comments

D_Hornby_ddh_uni

Good news got the tool working with help from @Aote_Xin !

Also it seems to be best practise but not enforced that the DEM extent should enclose the extent of the SAR dataset.

With a correct DEM and following the sequence:

Apply Radiometric Calibration (Gamma Nought)
Apply Radiometric Terrain Flattening

Extract Water runs successfully.

So where does one source correct DEMS? You could, as I had, go to the primary portal such as EARTHDATA but an alternative worthy of mention is https://opentopography.org/

When you finally get it to run its pretty awesome! 😁

Aote_Xin

Duncan, thanks for sharing the dataset.

I was able to generate detection results from running Extract Water tool in 3.6.2 and 3.6.0 using either NASADEM or COP30 DEM. The steps I took are 1. Apply Radiometric Calibration (Gamma Nought) -> 2. Extract Water.

Then, 1. Apply Radiometric Calibration (Gamma Nought) -> 2. Apply Radiometric Terrain Flattening -> 3. Extract Water also runs through successfully.

Please kindly try the DEM here: S1_duncan_dem.zip. Let me know if you are still seeing the issues.

If you are able to share your DEM, that would be awesome.

My email address is axin@esri.com.

Aote

D_Hornby_ddh_uni

Thanks for this pointer, that google document looks helpful. I'm keen to hear back from @Aote_Xin as I would like to keep the workflow "simple" with a single call to the extract water tool before I explore writing custom code to generate Otsu thresholding. Interesting though.

NavinTonyT

Hi,

I've never used the water extraction tool in ArcGIS Pro, and I don't know how well it performs either. Water detection from SAR usually produces misclassifications a lot.

This is won't solve your ArcGIS issue - but if you want to extract water from SAR, simple way is to run an Otsu thresholding - it does have its disadvantages but it is simple enough. You can find several Python codes, or run in GEE (https://docs.google.com/document/d/12cwzbNXtBQnm5switfL1vlv1stnp5D5gXcQeZMP_iJI/edit?tab=t.0#heading=h.2gd4k88k0hkr).

D_Hornby_ddh_uni

@Aote_Xin

Here is a temporary link to the Source SAR data, I will remove it in a week as its swallowing up my allowance on dropbox!

https://www.dropbox.com/scl/fi/buezphrq19miovxe4pypq/S1A_IW_GRDH_1SDV_20200209T110313_20200209T110342_031175_03959E_A378_COG.SAFE.zip?rlkey=ibfado9ahefvkxoin4ftwf70o&dl=0

For the record the DEM I had been using is the NASADEM which I sourced here.

Also I tried your recommendation take the source SAR (the manifest.safe) and add to map, run that through the Apply Radiometric Calibration (Gamma Nought) then the extract water and it still gives an instant 999999 error.

I remember trying the feature detection wizard, it would step through the pre-processing steps and fail on the extract water from what I remember. So I moved to running the individual geo-processing tools so I could follow the sequence and see if there were any warnings (there were none).

Aote_Xin

Hi Duncan,

I am sorry you are having issues using the Extract Water tool. Can you share your input Sentinel 1 GRD COG_SAFE Dataset?

Within ArcGIS Pro, once you have prepared a Level 1 Sentinel 1 GRD (or SLC) product and a paring DEM, you would be able to run extract the land-water interfaces in the radar scene.

The Level 1 SAR dataset could contain unwanted noise and distortions due to topographic effects. With that said, you can skip most of the SAR processing steps and still get results from Extract Water tool:

1. Run Apply Radiometric Calibration (Gamma Nought)

2. Run Extract Water tool

If you want to remove unwanted noise and distortions before applying Extract Water tool, check out the Feature Detection Wizard. The wizard takes your Level 1 SAR dataset, apply the SAR processing necessary for improving detection results for the selected features. You can also check out the StoryMaps on SAR Analysis which includes a section on Extract Water in Sentinel 1 GRD product.

Let me know if you have other questions,

Aote

Resources:
1. A StoryMaps on SAR Analysis

2. Extract Water tool Help

3. SAR Feature Detection Wizard

D_Hornby_ddh_uni

Hi Laura, Thanks for the reply. My sentinel 1 image is part of the Mekong delta so includes sea and large rivers, plenty of water! With your feed back it makes me think that their tool is broken as the error is almost instant. No time to run out of resources. It strikes me that help file for the tool is woefully under documented, it needs to explicitly state what steps that the input raster has undergone in which sequence and which type of SAR raster it needs to be.

I and I'm sure other users will be very keen to hear what that analyst comes back with. If you get an answer please post it!

Laura

OMG this:

I had the very same issue and showed it to support. They told me my imagery was a problem (got from a place the documentation said to and where the analyst told me to grab from). At that point they refused to help me further. They closed that and opened a new case where they are all stating the downloadable imagery doesn't work well enough for this and my question is then how are we supposed to use it? A new analyst is still looking for a suitable input..

I tried using extract water and a deep learning package i used in object detection.

D_Hornby_ddh_uni

Hi Dan,

safe is the supply format but the very first processing step, remove thermal noise, creates a new output that is the .crf format and from then on the tools create their outputs as .crf so what I try to feed into the extract water tool is a crf file which matches the format you spotted in the python sample code.

danretired

Not used at all, but I couldn't help notice that the tool under the python tab's code examples are all using *.crf as the input data type.

Extract Water (Image Analyst)—ArcGIS Pro | Documentation

is the .SAFE folder an equivalent or are the actual data layers contained within?

Připojte se k diskusi

Přihlášení členové mohou přispívat, sledovat aktualizace a další. Jste tu noví? Zaregistrujte si bezplatný účet.

Rychlé odkazy

Community Help
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.