Subscribe to this thread
Home - General / All posts - continuation of M9 exported .asc location error
orerockon

377 post(s)
#06-Sep-19 15:27

I had composed a rather lengthy post but for some reason when I try to attach a file it's puking. So the short form:

oeaulong said: "I am afraid we have hijacked Colin's thread. Manifold 9 opened this up without a hitch as a .TIF

Manifold 8 opened it up as a surface as well from the binary. No need for .asc, I don't know why you are getting errors. There would need to be a transform to prettify it in Display Options, I selected the negative values and made them invisible, then made the color classification."

I am not getting enough information on why you are getting tripped up by it. If you want to pursue this further, please start a new Thread as we are covering topics that have wandered from the subject."

(Win 10, restarted, nothing else open), importing the .tif as a surface in (8.0.30.0) I get a white country on a black background. It imports in a millisecond and the the XYZ values look reasonable. Shading I see nonsense numbers and all it does is change the background color. If this is a problem specific to my M8 installation , it might explain more than a few headaches I've had importing surfaces for a week now.

oeaulong

213 post(s)
#06-Sep-19 16:00

Mfd 8: I saw the same thing. The values for 'no data' is like BIG negative number (-16999999999999999..

0000000000000000000000000000000000000000000000000000000000000000000000000000000..

0000000000000000000000000000000000000000000000000000000000000000000000000000000..

0000000000000000000000000000000000000000000000000000000000000000000000000000000..

0000000000000000000000000000000000000000000000000000000.00) !! This is trash, junk, got to remove it. I did this a quick way by using the pixel selection tool "Select Touch" and holding the Shift down clicked on the black part. FIRST: You FIRST need to change the range of selection. Open up the Tool Properties pane, Alt+Shift+T. Then change the Value tolerance to 1.

Now you can perform the Shift_click on the black part. The Shift-click tells to select the pixels of this value range across the entire image. So, what this does is grab non-contiguous pixel values as well, i.e. The Great Lakes. Once it is selected, hit delete to turn those values into manifolds Invisible values. Still there, just invisible. <attachment 1>

Once that is done you can open up the Display Options under View menu and then choose a palette to display. I chose Precipitation for the demo, but you should choose what best displays the meaning. Click the downArrow "Apply" button and the classification table should populate, this time with reasonable numbers. It doesn't factor the invisible pixels. I give a shadow to provide some contrast, click OK and done. <attachment 2>

Attachments:
DisplayOptionsPreciptiationPalette.PNG
unwantedPixelSelect.PNG

orerockon

377 post(s)
#06-Sep-19 17:19

OK so that makes it visible and the values look reasonable! One more (kinda) related quesion. I'm looking at this (historical average summer air temp) on a couple scales. It says the resolution is a 4 km grid. I can transfer heights (they say in mm) to watershed polygons (100 to 3000 sq. km) to get an average, min, max (I don't think the min & max are particularly informative), that also look reasonable. If I use PRISM data I get very different answers that are waaaay too high and I have no clue why, could be my fault, the reason I am using this climate-change-modeled temp instead. If I chopped up my watersheds into 4 km2 blocks and get an average do you think they would be roughly equivalent? I think I can handle that.

oeaulong

213 post(s)
#06-Sep-19 18:28

I'm looking at this (historical average summer air temp) on a couple scales. It says the resolution is a 4 km grid.

Ok. is this from the same source? presented in Lat Long, it wouldn't give you a projected grid size. If I took the above discussed precipitation layer and projected in Lambert Conformal Conic, Mfd calculates a 3.2km grid size.

I can transfer heights (they say in mm) to watershed polygons (100 to 3000 sq. km) to get an average, min, max (I don't think the min & max are particularly informative), that also look reasonable.

Reasonable for a coarse nationwide presentation? Sure, but in cooking it down to smaller watersheds, I am afraid that the granularity would get in the way of the analysis. Then you bring in PRISM (I assume from OSU), a 30 second resolution (projected to ~800m?) as a completely different model & dataset and then what are you trying to accomplish? I don't have the strongest statistical background and education so I cannot be any real source of knowledge here.

If I chopped up my watersheds into 4 km2 blocks and get an average do you think they would be roughly equivalent? I think I can handle that.

It seems to me though that a. you are trusting the underlying data too much. & b. mashing them together you run the risk of complications due to inherent anomalies that hid inside the data itself from dissimilar sources. To then try to bludgeon the analysis into place by shrinking the aggregating collection vector layers seems to my under-educated eyeballs to be folly. I can only say, "Try it". If it works, I don't think it told you much, if it doesn't then the same applies.

There are, many folks that hang out here with more rigorous analytical toolboxes than I have. I just found that there are many ways to get data into Manifold of any version.

Good luck.

orerockon

377 post(s)
#08-Sep-19 03:52

I chatted with a guy who has been doing stream temperature modeling that I plan on using and it turns out he used the PRISM data to model historical air temperature and precipitation for the entire NW US. Problem solved but thanks for the lesson!

Manifold User Community Use Agreement Copyright (C) 2007-2017 Manifold Software Limited. All rights reserved.