Hello,
I am fairly new to the ECOSTRESS world, and newer yet to accessing the datasets using AppEEARS. Recently I have been working to download two ECOSTRESS data products "ECO-L2-LSTE-002", and "ECO-L2T-LSTE-002" (Swath and Tiled LSTE, respectively), through the use of AppEEARS. I have successfully downloaded, and begun to explore both datasets, though some questions have come up in doing so.
It appears (no pun intended) that there are some discrepancies (as expected) between the datasets due to their differing levels of pre-processing. I am a bit perplexed by two of the discrepancies that I have noticed, however, and am looking for a bit of help in identifying their cause.
Attached, you will find two figures (A and B). Figure A portrays the full summer growing season of 2023, and figure B, a subsection of this same growing season from July 1st to July 31st. In each figure you may note vertical columns of points, each representing LST readings from ECOSTRESS converted to Celsius for a total of 10 point locations on a given day (hence the columns of points). Red points from "ECO-L2T-LSTE-002" are overlaid with smaller, black points from "ECO-L2-LSTE-002". You will notice that some readings overlap, as expected (meaning that both datasets returned the same value for a given point), while some do not (meaning that each dataset returned a different value for the point).
The three questions that I have pertaining to this are as follows...
1. On certain days (July 10th for example), you will note two distinct groups of black points, one which lines up with the red points, and one far below them. Each cluster contains a reading for all 10 points, meaning that each point produced two readings. After further investigation, it seems that when this occurs, both clusters are from the exact same moment in time, but offer completely different readings for each of the 10 points. Why would this be occurring?
2. While some points match 1:1 between "ECO-L2T-LSTE-002"and "ECO-L2-LSTE-002", some are off by a marginal amount (ex. an LST reading of 12.5*C vs. 13.1*C). In the figures, you will see this as black and red points which are nearly overlapping, though not exactly. Why is this?
3. In several circumstances, "ECO-L2T-LSTE-002" provided readings for a given day, while "ECO-L2-LSTE-002" did not (See red points with no corresponding black points in early-mid August). Why is this?
Thank you for any help that you can provide me!
Swath and tiled datapoints extracted via AppEEARS
Swath and tiled datapoints extracted via AppEEARS
- Attachments
-
- Figure B
- July LST.jpeg (106.09 KiB) Not viewed yet
-
- Figure A
- Full Season LST.jpeg (91.21 KiB) Not viewed yet
Tags:
-
LP DAAC - dgolon
- User Services
- Posts: 296
- Joined: Mon Sep 30, 2019 10:00 am America/New_York
- Has thanked: 18 times
- Been thanked: 2 times
- Contact:
Re: Swath and tiled datapoints extracted via AppEEARS
Hi @ewestey Apologies we missed your question earlier this month. I'll talk to the AppEEARS team to see what is going on and will report back. Thanks.
Subscribe to the LP DAAC listserv by sending a blank email to lpdaac-join@lists.nasa.gov.
Sign up for the Landsat listserv to receive the most up to date information about Landsat data: https://public.govdelivery.com/accounts/USDOIGS/subscriber/new#tab1.
Sign up for the Landsat listserv to receive the most up to date information about Landsat data: https://public.govdelivery.com/accounts/USDOIGS/subscriber/new#tab1.
-
LP DAAC - afriesz
- Subject Matter Expert
- Posts: 59
- Joined: Tue Nov 12, 2019 4:02 pm America/New_York
- Been thanked: 2 times
Re: Swath and tiled datapoints extracted via AppEEARS
@ewestey,
I've tried to address all of your questions below. Let us know if you have any additional questions.
1. On certain days (July 10th for example), you will note two distinct groups of black points, one which lines up with the red points, and one far below them. Each cluster contains a reading for all 10 points, meaning that each point produced two readings. After further investigation, it seems that when this occurs, both clusters are from the exact same moment in time, but offer completely different readings for each of the 10 points. Why would this be occurring?
- The ECOSTRESS tiled data products have a quality threshold applied when producing the tiled scenes. This will result in data observations being present in the swath scenes that are not present or are filtered out in the tiled scenes. Swath scenes processed within the AppEEARS processing pipeline do now have thresholds applied, which is why you’ll see a return from the swath scene and not the tile.
- While I have pulled the data down from your AppEEARS requests, I have not been able to exactly replicate what you are reporting for July 10th. Your request for ECO_L2_LSTE (swath) data for the 10 points does return data values (the black points with LST low values), but I do not see any data returns for the ECO_L2T_LSTE (tiled) data and I do not see a second set of readings for the swath data on July 10th. Can you confirm your analysis? Feel free to send your AppEEARS request name or request ID to lpdaac@usgs.gov.
2. While some points match 1:1 between "ECO-L2T-LSTE-002"and "ECO-L2-LSTE-002", some are off by a marginal amount (ex. an LST reading of 12.5*C vs. 13.1*C). In the figures, you will see this as black and red points which are nearly overlapping, though not exactly. Why is this?
- This is likely due to the separate processing pipeline these two products/outputs go through. Tiled ECOSTRESS products are produced by the ECOSTRESS Science Data Processing System. Their processing projects/resamples the swath scenes to the UTM projection and maps the data to the MGRS grid. When you order an ECOSTRESS swath product from AppEEARS, the processing pipeline maps the swath data to a geographic grid (WGS84), based on the spatial characteristics of the scene, using nearest neighbor resampling. This will cause mismatches in the extracted data values between the two data sources.
3. In several circumstances, "ECO-L2T-LSTE-002" provided readings for a given day, while "ECO-L2-LSTE-002" did not (See red points with no corresponding black points in early-mid August). Why is this?
- This occurs because ECOSTRESS Swath scenes are projected to UTM and mapped to the MGRS Grid. Note, the MGRS/UTM grid is fixed in space and time, but the location of swath scenes between, and within, days is variable and dependent on the orbit of the International Space Station. A tile is created whenever a swath scene with valid data intersects with a MGRS/UTM tile. In many cases the amount of overlap between the swath scene and the tile is small, but an output for the entire spatial area of the tile will be created. Areas within the produced tile will be filled with “no data” where no intersection between the swath scene and the tile boundary occur. So, if a swath scene passes through the far East side of a tile, a tiled file will be created, and those data will be mapped to the tile. If your points intersect with the west side of said tile you will only get a reading/return/file returned for the tiled product, but nothing for the swath product.
I've tried to address all of your questions below. Let us know if you have any additional questions.
1. On certain days (July 10th for example), you will note two distinct groups of black points, one which lines up with the red points, and one far below them. Each cluster contains a reading for all 10 points, meaning that each point produced two readings. After further investigation, it seems that when this occurs, both clusters are from the exact same moment in time, but offer completely different readings for each of the 10 points. Why would this be occurring?
- The ECOSTRESS tiled data products have a quality threshold applied when producing the tiled scenes. This will result in data observations being present in the swath scenes that are not present or are filtered out in the tiled scenes. Swath scenes processed within the AppEEARS processing pipeline do now have thresholds applied, which is why you’ll see a return from the swath scene and not the tile.
- While I have pulled the data down from your AppEEARS requests, I have not been able to exactly replicate what you are reporting for July 10th. Your request for ECO_L2_LSTE (swath) data for the 10 points does return data values (the black points with LST low values), but I do not see any data returns for the ECO_L2T_LSTE (tiled) data and I do not see a second set of readings for the swath data on July 10th. Can you confirm your analysis? Feel free to send your AppEEARS request name or request ID to lpdaac@usgs.gov.
2. While some points match 1:1 between "ECO-L2T-LSTE-002"and "ECO-L2-LSTE-002", some are off by a marginal amount (ex. an LST reading of 12.5*C vs. 13.1*C). In the figures, you will see this as black and red points which are nearly overlapping, though not exactly. Why is this?
- This is likely due to the separate processing pipeline these two products/outputs go through. Tiled ECOSTRESS products are produced by the ECOSTRESS Science Data Processing System. Their processing projects/resamples the swath scenes to the UTM projection and maps the data to the MGRS grid. When you order an ECOSTRESS swath product from AppEEARS, the processing pipeline maps the swath data to a geographic grid (WGS84), based on the spatial characteristics of the scene, using nearest neighbor resampling. This will cause mismatches in the extracted data values between the two data sources.
3. In several circumstances, "ECO-L2T-LSTE-002" provided readings for a given day, while "ECO-L2-LSTE-002" did not (See red points with no corresponding black points in early-mid August). Why is this?
- This occurs because ECOSTRESS Swath scenes are projected to UTM and mapped to the MGRS Grid. Note, the MGRS/UTM grid is fixed in space and time, but the location of swath scenes between, and within, days is variable and dependent on the orbit of the International Space Station. A tile is created whenever a swath scene with valid data intersects with a MGRS/UTM tile. In many cases the amount of overlap between the swath scene and the tile is small, but an output for the entire spatial area of the tile will be created. Areas within the produced tile will be filled with “no data” where no intersection between the swath scene and the tile boundary occur. So, if a swath scene passes through the far East side of a tile, a tiled file will be created, and those data will be mapped to the tile. If your points intersect with the west side of said tile you will only get a reading/return/file returned for the tiled product, but nothing for the swath product.