ArcMap Help Needed for Student Dissertation

[JamesMills]

Hi all,

I’ll try to keep this as concise as I can. I am a university student currently in the middle of doing my third year project, due to complications with my original project idea I had to devise something new and the project I am now working on involves GIS analysis using ArcMap.

The analyses I’m trying to carry out seem like they should be simple, however I am a complete novice with GIS and have a very limited understanding of ArcMap (had I known that I’d end up doing a project using GIS I would have taken the relevant course module last academic year), so I am in need of help!

My project is examining how flood risk at a number of UK coastal locations (where there are tidal gauges) will change as mean sea level increases, by comparing tidal records with a high resolution digital elevation model. There are numerous stages to this project, so I’ll just keep my post about the GIS stage.

Here’s a bit of information:

I’m attempting to analyse LiDAR (light detection and ranging) raster land elevation data at 6 UK coastal tidal gauge locations. Data was obtained from Environment Agency’s Geomatics group site.

I have downloaded enough LiDAR data for each location to be able to accommodate a study area of 3km radius at each location.

I want to use ArcMap to perform the following processes and analyses on the data:

1) Import my X/Y coordinates (in the form of British National Grid Easting/Northing) for each coastal location to create a set of point features around which I can create Buffer Zones.

2) Define a ‘study area’ around each location point, thereby enclosing a certain number of ‘elevation data’ raster cells in each study area (ideally each study area would be of a semicircle shape with a radius of 3km around each location point, with the semicircle going inland parallel to the coast; if not a circular study area with a smaller radius will probably suffice)

What I know:

I have been advised that creating these study areas will involve creating a ‘Buffer’ around each location point. I have been advised that I will need to assign a unique ID to each Buffer zone in order to perform my subsequent analyses.

Help needed:

- Is it possible to create a semicircle Buffer zone around a point? How?

- How can I create a Buffer around each individual location point?

- How can I give each Buffer Zone a unique ID, so that for example I can get a ‘Zonal Statistics’ table for each Buffer Zone?

3) Find what % (or at least frequency so that I can work out %) of the raster cells in the study area/buffer lie within different ‘elevation bands’ (I can find the range (min/max) of elevation values in the study buffer using Zonal Statistics as Table, but then I want to see how the elevation values are distributed within the buffer zone so that I can create a histogram; by seeing what % of the raster cells in the study area are within different elevation bands such as 0-5m, 5-10m etc.)

What I know:

I have been advised that this will involve ‘Reclassifying’ the data and finding out the ‘Frequency Distribution’ of values in each Buffer Zone.

Help needed:

- How can I find out how many raster cells overall are enclosed by each study area buffer zone?

- What specific ‘Reclassify’ process do I need to perform in order to find out how the buffer zone elevation data raster cell values are distributed between different elevation bands (the frequency of values in each band)? How do I do this?

(If I know how many raster cells are in each buffer zone overall, and then I can find out what frequency of raster cells fall within different elevation bands, then I can calculate the % values easily).

4) Determine an elevation value below which, for example, 5% of the study area raster cells lie in terms of their data value

What I know: I have been told that this again involves examination of the ‘Frequency Distribution’ of values

Help needed: How do I do this?

5) For any given elevation value (i.e. a calculated elevation value for an average storm tide event at present and then at 10 year intervals in the future), determine what % of the raster cells in each study area lie below or at that value

What I know: I have been told that this again involves examination of the ‘Frequency Distribution’ of values

Help needed: How do I do this?

Ultimately these GIS processes are an attempt to obtain the following information for each location/study area:

- Range of elevation values in the study area, and % values denoting what proportion of study area raster cells in each elevation band (so I can create a basic ‘local land elevation’ Histogram)

- Elevation value (in meters) below which 5% of study area raster cells lie (the lowest 5% of cells; I’ll use this elevation value to look at the Probability of any given monthly extreme tide surpassing this elevation value, seeing if probability increases as tide values are adjusted by expected sea level increase etc.)

- What % of study area raster cells lie below any given elevation value (I’ll calculate the height of an average storm tide event at present, and then at 10/20/30… years in future etc., and then I want to see what % (or frequency, at least) of the study area’s raster cells lie at or below each of these elevation values, so that I can determine how the % of land at risk from flooding will change as mean sea level increases)

So that’s all I’m really trying to find out here!

If anyone can spare 10 minutes to give me guidance on how to do any or all of these analyses, you would be helping me out big time!

As mentioned, I’m basically a beginner with all of this ArcMap stuff, so the more detail you can give the better!

Thank you,

James

[XanderBakker · Accepted solution]

Hi James,

Although you provided a lot of information, I still miss some details on your software to know what functionality you have available:

- what version of ArcGIS are you using?

- what license level do you have (arcview-basic, arceditor-standard or arcinfo-advanced)?

- which extensions (Spatial Analyst, 3D Analyst) do you have available?

1) The import depends on your data format. Did you get .TXT files with XYZ values or did you download LAS files (this would be easiest)

2) To create a buffer you can read this help topic:

ArcGIS Help (10.2, 10.2.1, and 10.2.2) 

To create a semi circle, you will probably have to edit the buffers and split them.

ArcGIS Help (10.2, 10.2.1, and 10.2.2)

Normally your buffers will have a field OBJECTID which has a unique value for each feature.

You can use that as the unique identifier or add a field and fill them with the unique values you want.

3) The zonal statistics will provide information of a statistic per zone. This will not create the histogram.

When a Raster is of type integer, you will have a histogram, but it will contain a histogram for all the study areas in your raster.

To convert a floating raster to integer, use this tool:

ArcGIS Help (10.2, 10.2.1, and 10.2.2)

You can use the combine tool to create a raster holding the combination of zones and integer values from your LiDAR:

ArcGIS Help (10.2, 10.2.1, and 10.2.2)

The result is a table that allows you to calculate the stats your looking for.

Instead of translating the raster to integer you could reclassify the raster in the ranges you want.

This simplifies the result.

ArcGIS Help (10.2, 10.2.1, and 10.2.2)

I leave it for the moment to this, since I gtg...

Kind regards, Xander

[JamesMills]

Thank you very much for your kind advice Xander, it's a great help!

In answer to your queries:
- I am using ArcGIS 10.2.2

- The copy of ArcGIS I am using is owned by my university, and I'm pretty sure their license is as comprehensive as possible (so I can only assume arcinfo-advanced)

- I am not able to access the software right this moment, but from memory a wide range of Extensions are available (there is definitely Spatial Analyst, 3D Analyst, Geostatistical Analyst, and several others). Again, I'm pretty sure the university's version is comprehensive in its capabilities.

1) Don't worry too much about the coordinates import as I'm pretty sure I understand how to do this at least; basically I've just got the coordinates for each tidal gauge in an Excel spreadsheet and I wanted to make sure I knew how to import this to create a point for each. If you're referring to the LiDAR data, it came in the form of ASCII files which I then had to convert to raster

2) Thanks for the Buffer advice! I'll do my best to create the desired semicircle study areas

3) I'm not actually trying to create the histogram within ArcMap, merely attempting to extract the relevant information so that I can create a Histogram in Excel for example. Why do I need to convert the floating raster data to integer? Wouldn’t this simply reduce the precision of the elevation data values contained in the raster? The histogram I am attempting to create would have ‘Elevation band’ (i.e. 0-5m, 5-10m) as the X-axis and ‘% of Study area’ (or frequency of values in the band) as the Y-axis. I am not looking to create a traditional histogram showing absolute frequencies of all possible
elevation values. This is why it is my understanding that Reclassification would be a better approach.

Thank you again for your advice Xander, it is very much appreciated.

James

[XanderBakker]

Hi James,

Great to hear that you can advance with the previous explanation, but to complete it, I'll continue with...

4) In case you have done a zonal statistics for the semi circle, you will have a count columns which holds the number of cells inside each semi circle.

With your "histogram" (frequency distribution), you can sort the data on height (ascending order) and calculate the percentage for each record (remember: area = count * cellsize^2). Next calculate the cumulative area (or percentage) and determine the value that corresponds to the percentage you're looking for (e.g. the 5% you were mentioning).

How to do this with some python code, please refer to this thread:

https://community.esri.com/message/391944#392135

5) for determining the flood plain you should not take the amount of pixels that are beneath the flood level. Only those pixels that can be reached should be taken into account.

An example thread were this is discussed you can find here:

Re: Determine Floodplain: Based on known flood level

Good luck with your analysis.

Kind regards, Xander