Do all your shapefiles have to be a projected coordinate system in order to calculate geometry?

[MelissaTraverso]

Hi,

I am trying to calculate burn area of brazil using MODIS burn area product shapefiles for northern and central South America. These were monthly shapfiles from April - Decemeber of 2015. The goal was to merge/dissolve these shapefiles of burn area for this time frame and calculate the percentage burn out of total area of brazil. The area of the burn polygons were not included within the attributes so I had to add a field and calculate geometry to do this. The steps I went through are below. Perhaps I made a mistake somewhere because if you see the attached PDF you will see that my calculated percentage burned of 1.56% is a lot smaller than it appears in the imagery. Does the PDF look like 1.56% burned to you?  

STEPS

1. Burn area shapefiles had an unknown projections so I projected all of them with WGS 1984.

2. I clipped the shapefiles to a shapefile of Brazil because the MODIS data was for northern and central south America.

3. For each shapefile I went to "add field", titlted in "Area" selected "double" and right clicked the field and went to "calculate geometry".

4. I ran into an issue here because my shapefiles were in a geographic coordinate system. I read online how to get around this so I made my data frame a projected coordinate system "WGS 1984 world Mercator". I was able to calculate area using the projected coordinate system of the data frame but keeping my shapefiles as GCS of WGS 1984 .

5. I then merged all monthly shapefiles together and then used the dissolve tool to merge multiple polygons into one polygon.

6. When I finally dissolved the merged monthly shapefiles so that the overlapping polygons would not double count the area my total area burned for brazil was 130,942 sq. km as opposed to 190,785 sq km from the merge. This was found by right clicking the attribute field "area" in the dissolved shapefile and selecting "statistics" and reading the number under "sum". The total area of brazil is 8,358,140 sq km. This gives you a percent of 1.56%

Is there something I could have done in this process to get the areas miscalculated? Potentially at step 4?  Why does the imagery appear to be 20-30% burned but my calculation is much smaller?

Any helpful tips or advice would be appreciated!

Thanks,

M

[DanPatterson_Retired]

To see what you have, 

rename the *.prj to *.prX  (it then won't be used in the next step)

open a new dataframe (Insert, Data frame)

add just the one shapefile.

right-click on the layer in the TOC and report the extent

do the left, right, top and bottom look like decimal degrees? (ie in the range -180/180 to -90/90)

If they are then the coordinates are geographic (ie decimal degree values), now you only need to figure out what the datum is... what is used by the people the published the 'manual' ?  a WGS84 datum? (don't confuse that with a coordinate system, people use that loosely and shouldn't)  if the datum is a WGS84 datum, then you have a Geographic coordinate system ... aka GCS WGS84

Report specifics from the extent properties and your 'manual'

[MelissaTraverso]

Hi Dan,

I did this and yes the extent appear to be in decimal degrees in the ranges you specified. I am fairly positive that the datum is WGS84 especially from reading the prj file in a text editor.

Again, this is what was in the prj file. How do I be sure that the datum is WGS84?

GEOGCS["unknown",DATUM["D_unknown",SPHEROID["unretrievable_using_WGS84",6378137,298.257223563]],PRIMEM["Greenwich",0],UNIT["Degree",0.017453292519943295]]

The manual reads:

"A user-friendly GeoTIFF version of the MCD64 product is derived from the standard MCD64A1 HDF

version by University ofMaryland. The GeoTIFF files are reprojected in Plate-Carr´ee projection and cover a

set of sub-continental windows (Figure 2). A table containing the regions covered and bounding coordinates

of the 24 windows.

Shapefiles of the MCD64A1 Burn Date layer are derived from the monthly GeoTIFF files by the University

of Maryland. The shapefiles are available with the same projection (Plate-Carr´ee) and geographic extent

used for the GeoTIFF sub-continental windows (Figure 2)."

Windows 5 and 6 are the regions in south America that I want to look at.

Thanks,

Melissa

[DanPatterson_Retired]

yes... with the caveat ... unretrievable_using_WGS84... but you have what it is supposed to be and selected your study area, so that projection is pretty useless, but define it as such, then move on after you define it, to projecting it to produce a new file in your desired projection (previous suggestions should be fine).  But make sure you use the Project tool to project shapefiles and equivalent tools for raster in the Data Management, Projections... blah blah.  Do not skip the step of defining your data first.  there must be a million posts where people skip that step and end up in a mess.  I wish esri would stop allowing people to use data which are not defined correctly at all.  Good luck

[MelissaTraverso]

Hi Dan,

When you say " yes... with the caveat ...unretrievable_using_WGS84... but you have what it is supposed to be and selected your study area, so that projection is pretty useless, but define it as such" you mean to say use the tool "define projection" and define the projection as GCS "WGS 1984"?

Thanks,

Melissa

[MelissaTraverso]

If I have already used the Project tool on all the raw shapefiles and projected them to GCS WGS 1984, is this the same as if I were to define their projection as WGS 1984? This will save me a lot of work if I can already use the shapefiles where I projected to GCS WGS 1984 assuming defining projection and project would do the same thing if I were to use the same GCS.

Moving forward, would I then just need to re-project these GCS shapefiles to PCS South American Albers Equal Area Conic and then re-calculate the geometry based on the PCS of the data source?

Thanks,

Melissa

[DanPatterson_Retired]

If you did the check on the file 'extents' of the coordinates as I suggested and they all appear to be in a Geographic coordinate system ( perhaps GCS WGS84) then the Project tool will project them into a PCS specified by you.  Define Projection... produces no new file   Project Tool ... makes a new file in a different coordinate system.

[RebeccaStrauch__GISP]

ah...That might be because you already tried to project it.

[MelissaTraverso]

Hi Rebecca,

I downloaded a raw shapefile again to double check it wasn't something I already made a change to and the prj file reads the same thing

GEOGCS["unknown",DATUM["D_unknown",SPHEROID["unretrievable_using_WGS84",6378137,298.257223563]],PRIMEM["Greenwich",0],UNIT["Degree",0.017453292519943295]]

[DeborahHuber]

Melissa,

Not 100% sure on a simple solution to your particular issue in South America, but I can share my experience on a recent project calculating areas in the USA. It was a housing inventory spanning the country. Building envelopes were laser measured by field staff and drafted in CAD - so I knew the exact gross square footage. Those footprints were pulled into GIS and snapped to GPS points.

My contract required product delivery in Web Mercator (since the end user views ALL of the houses), but the calculations I performed in Web Mercator were WAY off of what the CAD measurements were telling me they should be.  I had to break my calculations up into State Plane sections. For example, California: I queried my data by all of the footprints that fell inside Nad83 State Plane California zone IV, changed my data frame to that projection, performed the Calculation.  Then I moved on to query just the footprints that fell in California zone 5, Calculated geometry for those records, and so on. 

My calculations matched the CAD calculations +/-1 sq. ft. - the slight difference in measurement was due to the footprints being rotated to snap to the GPS points - when I performed the calcs BEFORE snapping and turning them, they nailed the measurement with the CAD value every time.  Maybe you'd need to do the same?  Is there a smaller, more localized projection you could use for individual sections of SA?  Then you could just calculate sections at a time using the Data Frame projection, but your overall data does not change its native projection.