Subject: Marine Geospatial Ecology Tools (MGET) help
Text archives
| From: | "Jason Roberts" <> |
|---|---|
| To: | "'Haddow, Gregory'" <> |
| Cc: | <> |
| Subject: | RE: [mget-help] connectivity tool help |
| Date: | Fri, 22 Jun 2012 10:27:38 -0400 |
Dear Gregory,
Thanks for your interest in MGET. We hope it will prove useful.
Unfortunately we do not have a connectivity example ready at this time. But
I should be able to help you through the process. The typical procedure is
something like this:
1. Obtain GIS data for habitat locations. If studying coral species, this
are locations of reefs. Often this will come as a shapefile, such as the
UNEP-WCMC Global Distribution of Coral Reefs dataset.
2. Clip reef data to spatial extent of your study area. Make sure there is
some buffer around all of the reefs (do not clip so that reefs are right at
the edge). If you did that, then larvae might drift off the edge of the map
and not be allowed to drift back.
3. Decide on the resolution of the analysis, i.e. the cell size of the
rasters you will create. This is always a complicated decision. On the one
hand, you want to have high resolution because your reef data is often at
high resolution. On the other hand, the ocean currents data is typically at
much lower resolution. If you go with a high resolution, the simulator must
interpolate the currents data up to that resolution and you have no way of
knowing if that interpolation is correct, and therefore how much error it
introduces into the result. Also, higher resolutions require more memory and
much more processing time. It is best, if possible to choose something
intermediate, and not too far away from the currents resolution. For
example, if you are using global HYCOM for currents, which has a resolution
of about 9 km, and your reefs are at about 1 km, use 9 km if possible. If
you really don't like that, maybe split the difference and go with 5 km. Try
not to go all the way to 1 km if possible.
4. Convert the reefs shapefile to a binary raster at the appropriate
resolution. Do it such that non reef cells have NODATA and reef cells have
some value such as 0 or 1 (it does not matter what this value is).
5. Use the ArcGIS Region Group tool to assign the reefs distinct IDs. Now
you'll have distinct patches with IDs 1 ... some higher value.
6. If you do not like these results, you will have to hand edit the raster
using ArcMap, or something similar. Arc is not very good at hand editing
rasters. If you do not have skill here, find an ArcMap expert, explain the
problem, and ask them for help with the editing.
7. Once you are done with these steps, you now have the reef IDs raster.
This raster should have integer values of 1 ... N for reefs and NODATA
everywhere else.
8. Now the reef cover raster. This must have the exact extent and cell size
(and rows and columns) of the reef IDs raster. It must be floating point.
For each reef cell, it gives the proportion (0 to 1) of that cell that is
covered by reef. You can use this to account for the fact that you had to
create a relatively low resolution reef IDs raster (e.g. 9 km resolution)
from a high resolution reef shapefile (e.g. 1 km resolution). You'll need to
use tools such as Block Statistics or Zonal Statistics to accomplish this.
If you don't know how to do it, seek help from an ArcGIS expert in your
department. (Or you can ask me, but this can be hard to describe, so please
try someone local if possible!) Alternatively, if you want to proceed with a
rough simulation without waiting for this tedious step, you could assume
that each cell has 100% covered by reef. To do that, just use Spatial
Analyst tools (e.g. Is Null, Con) to create a floating point raster that has
1 where reefs are, and 0 or NODATA where there are no reefs.
9. Now the water mask raster. Start by obtaining a shoreline polygon
shapefile such as GSHHS. Then clip it and convert it to a raster that
exactly matches the reef IDs raster (same extent and cell size). Set it so
that 0 or NODATA indicates land, and some other value (e.g. 1) indicates
water. Then use your reef IDs raster to change any reef cells that happen to
be land cells into water cells. You'll have to use ArcGIS Spatial Analyst
tools to accomplish these things.
I know it is complicated, but if you have some experience with ArcGIS, it is
not that bad. I hope this helps...
Best,
Jason
-----Original Message-----
From: Haddow, Gregory
[mailto:]
Sent: Friday, June 22, 2012 7:42 AM
To:
Subject: [mget-help] connectivity tool help
Dear MGET-help,
Im a biology MSc student at the University of Exeter in the UK. Im currently
trying to use your MGET connectivity analysis tool in ArcGIS.
I am relatively new to the software and have been struggling to create the
three input raters i.e. reef ID, reef cover and water mask. Do you have any
examples of how to create them like you have examples for species habitat
modeling. I have looked at the ESRI help forums but can't find anything of
use.
Finally i'd like to say thank you for producing this tool, its potential to
add to my research is immense.
Kind regards
Gregory Haddow
School of Biological Sciences
Geoffrey Pope Building
Streatham Campus
Exeter
Devon
UK
Thanks for your interest in MGET. We hope it will prove useful.
Unfortunately we do not have a connectivity example ready at this time. But
I should be able to help you through the process. The typical procedure is
something like this:
1. Obtain GIS data for habitat locations. If studying coral species, this
are locations of reefs. Often this will come as a shapefile, such as the
UNEP-WCMC Global Distribution of Coral Reefs dataset.
2. Clip reef data to spatial extent of your study area. Make sure there is
some buffer around all of the reefs (do not clip so that reefs are right at
the edge). If you did that, then larvae might drift off the edge of the map
and not be allowed to drift back.
3. Decide on the resolution of the analysis, i.e. the cell size of the
rasters you will create. This is always a complicated decision. On the one
hand, you want to have high resolution because your reef data is often at
high resolution. On the other hand, the ocean currents data is typically at
much lower resolution. If you go with a high resolution, the simulator must
interpolate the currents data up to that resolution and you have no way of
knowing if that interpolation is correct, and therefore how much error it
introduces into the result. Also, higher resolutions require more memory and
much more processing time. It is best, if possible to choose something
intermediate, and not too far away from the currents resolution. For
example, if you are using global HYCOM for currents, which has a resolution
of about 9 km, and your reefs are at about 1 km, use 9 km if possible. If
you really don't like that, maybe split the difference and go with 5 km. Try
not to go all the way to 1 km if possible.
4. Convert the reefs shapefile to a binary raster at the appropriate
resolution. Do it such that non reef cells have NODATA and reef cells have
some value such as 0 or 1 (it does not matter what this value is).
5. Use the ArcGIS Region Group tool to assign the reefs distinct IDs. Now
you'll have distinct patches with IDs 1 ... some higher value.
6. If you do not like these results, you will have to hand edit the raster
using ArcMap, or something similar. Arc is not very good at hand editing
rasters. If you do not have skill here, find an ArcMap expert, explain the
problem, and ask them for help with the editing.
7. Once you are done with these steps, you now have the reef IDs raster.
This raster should have integer values of 1 ... N for reefs and NODATA
everywhere else.
8. Now the reef cover raster. This must have the exact extent and cell size
(and rows and columns) of the reef IDs raster. It must be floating point.
For each reef cell, it gives the proportion (0 to 1) of that cell that is
covered by reef. You can use this to account for the fact that you had to
create a relatively low resolution reef IDs raster (e.g. 9 km resolution)
from a high resolution reef shapefile (e.g. 1 km resolution). You'll need to
use tools such as Block Statistics or Zonal Statistics to accomplish this.
If you don't know how to do it, seek help from an ArcGIS expert in your
department. (Or you can ask me, but this can be hard to describe, so please
try someone local if possible!) Alternatively, if you want to proceed with a
rough simulation without waiting for this tedious step, you could assume
that each cell has 100% covered by reef. To do that, just use Spatial
Analyst tools (e.g. Is Null, Con) to create a floating point raster that has
1 where reefs are, and 0 or NODATA where there are no reefs.
9. Now the water mask raster. Start by obtaining a shoreline polygon
shapefile such as GSHHS. Then clip it and convert it to a raster that
exactly matches the reef IDs raster (same extent and cell size). Set it so
that 0 or NODATA indicates land, and some other value (e.g. 1) indicates
water. Then use your reef IDs raster to change any reef cells that happen to
be land cells into water cells. You'll have to use ArcGIS Spatial Analyst
tools to accomplish these things.
I know it is complicated, but if you have some experience with ArcGIS, it is
not that bad. I hope this helps...
Best,
Jason
-----Original Message-----
From: Haddow, Gregory
[mailto:]
Sent: Friday, June 22, 2012 7:42 AM
To:
Subject: [mget-help] connectivity tool help
Dear MGET-help,
Im a biology MSc student at the University of Exeter in the UK. Im currently
trying to use your MGET connectivity analysis tool in ArcGIS.
I am relatively new to the software and have been struggling to create the
three input raters i.e. reef ID, reef cover and water mask. Do you have any
examples of how to create them like you have examples for species habitat
modeling. I have looked at the ESRI help forums but can't find anything of
use.
Finally i'd like to say thank you for producing this tool, its potential to
add to my research is immense.
Kind regards
Gregory Haddow
School of Biological Sciences
Geoffrey Pope Building
Streatham Campus
Exeter
Devon
UK
Archives powered by MHonArc.