Marine Data Literacy 2.0

Providing instruction for managing, converting, analyzing and displaying oceanographic station data, marine meteorological data, GIS-compatible marine and coastal data or model simulations, and mapped remote sensing imagery

 

 

 

 

Home > 5. Gridded Data > 5.13 Grid Stats

5.13 Using Grid Statistics and Calculations for Ecological Assessments in Saga

  • Exercise Title:  Using Grid Statistics and Calculations for Ecological Assessments in Saga

  • Abstract:  In this exercise you'll see how to make and interpret higher-level grids that are based on statistics and calculations involving existing grids of a selected parameter (sea temperature in this case).  The general methods shown here can be used in countless other situations and for other parameters of interest.  "Marine physiological ecology" is the science of habitat suitability for marine organisms, most of which have adaptability or resistance mechanisms to deal with changes in environmental parameters in that habitat.  Temperature, oxygen, salinity and pressure are parameters often viewed by specialists in this field, with an interest in stresses on the biota caused by seasonal to inter-annual changes.  Products that directly visualize these parameter changes are produced below.

  • Preliminary Reading (in OceanTeacher, unless otherwise indicated):

  • Required Software:

  • Other Resources: 

    • 12 (or 4) monthly SST grids for Liberia made in Exercise 2.22 [The cited exercise is no longer available; this exercise (5.13) is being revised; sorry for the issue.]

  • Author:  Murray Brown

  • Version:  7-17-2014

1.  Use the methods in 2.22 Importing "Scaled" HDF Satellite SST Climatologies into Marine GIS: Color Web to create 12 monthly SST grids for Liberia, using a 0.05-degree target grid.  This target grid is necessary to keep the original resolution.  Because they are so close (4 km versus 0.05 degrees) the resampling method is the basic NEAREST NEIGHBOR algorithm.  [You might only have 4 monthly grids if you are as lazy as the instructor; that's OK.]
2.  All 12 (or 4) monthly grids must be loaded into Saga.  Proof of their identical structure (a requirement of this method) is demonstrated by their placement under one grid system.
3.  Select TOOLS > SPATIAL & GEOSTATISTICS-GRIDS > STATISTICS FOR GRIDS.

This module allows you to calculate any or all of the 7 statistical measures shown here, for example the ARITHMETIC MEAN.  These measures, when they are SET, are calculated from all 12 monthly values of each pixel, and a new grid is created consisting of these statistical pixels.

4.  For GRID SYSTEM select the system of the monthly SST grids.

Then click on the ellipsis (...) to the right of NO OBJECTS.

5.  This grid object selection window appears.
6.  Use the > control to move the data grids to the right side, as you see here.  [Leave the dummy grid behind, if you still have it loaded.]

Then click OK.

7.  Now you can select which statistics to create.  Change these settings to CREATE:
  • ARITHMETIC MEAN
  • MINIMUM
  • MAXIMUM
  • RANGE

Then click OK.

8.  These new grids appear in the Saga data list.
9.  Re-name the new analysis grids appropriately, perhaps as you see here.
10.  At this milestone, you might want to save any new shape(s) or grid(s) to make sure you don't lose anything in case of problems later on.  Just name them with the above filenames (or equally descriptive ones), and use a logical folder location (such as PRODUCTS > SAGA > GRIDS > COLORWEB).
11.  You might also save the entire project at this stage, with FILE > PROJECT > SAVE PROJECT AS, and use the folder PRODUCTS > SAGA > PROJECTS, perhaps with this project name:
  • sst_stats_4mos_11mu_day_liberia_modisa_colorweb_0.05deg
12.  Now let's take a look at the products.  Here's the MEAN or AVERAGE distribution.  This just gives you a good idea of the general characteristics of the area, but it does not show extreme values at all.
13.  Here's the MINIMUM distribution.  Based on your biological knowledge, you might see values so low that they could cause death or injury to some organisms.  So based on this map, you could search for the areas where these conditions occur.
14.  Here's the MAXIMUM distribution.  So similar to the above, you might identify lethally high values, and then use maps to find them. 

For example we know that corals start showing impacts at temperatures in the low 30's.  The range indicates that there are cells in that range, but obviously very few of them.  Where are they?

15.  One of the principles of physiological ecology is that the annual range of environmental values (e.g. temperature) can be just as significant as the minimum or maximum for organism survival.  If you have any target organisms for your research, what upper limit of change can they endure, and is it exceeded in these data?  Where?
16.  Now we have an idea of what we're looking for in the data maps.  So based on the above, then a "universal" SST range for Liberia would appear to be approximately 22-30, ignoring the extremes.  So let's see what the SST maps look like, all using the same range for the palette.
17.  Here, the VALUE RANGE has been set to the same range for all figures (22-30 degrees).  Now the figures can confidently be compared visually, and they "make sense" scientifically.  You can simply check the single, unified color palette on the top right to estimate SST values for any figure.  As a biologist, you could now use these grids to assess the highest and lowest temperatures that usually occur here, and where they occur.  This is a very basic ecological concept that is much easier to do with these statistical images, than by visually scanning 4 to 12 monthly images.

18.  What if you were asked to "map" the area where annual temperature changes are 4 degrees or more, for example.  By "map" we mean to create an objectively calculated map that unambiguously shows exactly where that area is, and the user is not required to make guesses based on color shades.  How do you do that?
19.  One way to do it is by "classifying" the data.  Select TOOLS > GRID-TOOLS > CHANGE GRID VALUES.
20.  Make these selection:
  • GRID SYSTEM - The SST images system
  • GRID - The annual range grid
  • CHANGED GRID - Set to CREATE
  • REPLACE CONDITION - Set to LOW VALUE <= GRID VALUE < HIGH VALUE

Then click on the ellipsis (...) to the right of LOOKUP TABLE.

21.  The initial table has these 3 columns, but only 2 rows.  We need a total of 6 rows.  You can use a right-click on any table row to ADD RECORD or INSERT RECORD, as needed.

22.  Edit the table to have these values.  The REPLACE WITH values are codes for temperature ranges, not SST values themselves. 

  • Code 102 contains all the lowest values, from 0.0 up to 2.0, so nothing is missed
  • Codes 103-106 contain SST values that step up in increments of 1.0, from 2.0 to 6.0
  • Code 107 contains values above 6.0 to the maximum possible valid SST, so nothing is missed

Then click OK.  And click OK again.

23.  You now have a new grid named CHANGED GRID that does not contain SST values or SST difference values at all.  It contains only numerical codes.  Each pixel has been assigned a code to show where it is found in the above table.
24.  Here is the histogram for CHANGED GRID, showing that it only has 4 different values.  Apparently there were no values in the lowest or highest bins..
25.  And here is the changed grid, a "classified grid", showing the different code zones.
26.  You might want to save this grid now, in the folder PRODUCTS > SAGA > GRIDS with the filename sst_range_classes_11mu_day_modisa_colorweb_0.05deg
27.  If you want to visualize only a portion of these zones, you can manipulate the NO DATA fields of the OPTIONS panel.  This is possible, because the temperature range codes are sequential (an important thing to remember, if you do this sort of work).

Here the codes 102, 103 and 104 (covering temperature ranges from 0.0 to 4.0 degrees) are "masked" by declaring them to be NO DATA values.  Click SETTINGS > APPLY to see the result.

28.  This is a map of the areas where the annual temperature range exceeds 4.0 degrees.  [The land area of Liberia is also shown, because it has a value of -99999, the native NO DATA value in Saga.]

Maps like this are much more useful for decision-making and presentation, if you want to emphasize a particular area, rather than dazzle the viewer with colors.  A standard graphics editor (IrfanView) has been used to add informative labels.

29.  We hope you saw many possibilities in the above exercise for other statistical analyses that you can do with Saga.  Once you have the library of basic grids (which is usually 95 % of the work), then the above steps are easy and fast.  Good luck.