Accepted Solutions
mmmm, perhaps Combine is not going to be the solution, since it supports only up to 20 rasters (would require more than 50 steps). The result is a raster with an attribute table holding all the species that occur (as you are looking for)...
The aim of this analysis is to have each cell show the mean wing length, body length & thorax width of all species which occur in the pixel. So in the end, I hope to have three rasters which show me all of the three measurements, and therefore show me the spatial distribution of the average body size (either wing length, body length or thorax width).
Yes I have 1000 rasters each with a attribute table like this:
All rasters are classified as 1 (equal presence of the species).
Each single cell of the rasters overlap exactly and have the same cell size. But the raster themselves have different extensions.
One cell is 100 * 100 km. The study area is the whole "old world." (so theoretically there would be a maximal of 205 columns and 141 rows possible)
Many thanks!
I would probably use a Python script to "batch" process the data.
What I also saw in your data is that the fields TW_SMUNDI_ and FL_SMUNDI_ are text field. The values in these fields should be converted (in the script) to double (float) to be able to calculate some statistics.
I just created a new featureclass and added a rectangle polygon that covers the entire area and did an intersect between the rectangle and the provided polygons. The 12 polygons converted to 97 polygons.
This intersect featureclass I used to do a simple test on the field BL_SMUNDI_:
- I add two fields:
- XB_Keep (will be filled with value 1 for those polygons I want to keep) and
- Mean_BL that will hold the mean value
- I loop through the polygons and create two dictionaries:
- dct_geo that will hold the JSON geometry (to check if a polygon already exists in the dictionary) plus a list of oid's that have the same geometry
- dct_att that will hold the oid versus the value from the field BL_SMUNDI_
- In the next loop I calculate the mean value for each unique geometry, using the list of oids and the related dictionary with oid and values
- the result is written to a dictionary dct_res that holds the first oid of the unique geometry and the mean value calculated
- The last cursor is to write those values back to the input featureclass
import arcpy
fc_in = r"C:\Forum\SpecieRasters\data.gdb\intersect"
# fc_out = r"C:\Forum\SpecieRasters\data.gdb\result01"
# fld1 = "TW_SMUNDI_" # text field!
fld2 = "BL_SMUNDI_"
# fld3 = "FL_SMUNDI_" # text field!
fld_keep = "XB_Keep"
fld_res = "Mean_BL"
if len(arcpy.ListFields(fc_in, wild_card=fld_keep)) == 0:
arcpy.AddField_management(fc_in, fld_keep, "LONG")
if len(arcpy.ListFields(fc_in, wild_card=fld_res)) == 0:
arcpy.AddField_management(fc_in, fld_res, "Double")
dct_geo = {}
dct_att = {}
flds = ("OID@", "SHAPE@", fld2)
with arcpy.da.SearchCursor(fc_in, flds) as curs:
for row in curs:
oid = row[0]
polygon = row[1]
val = row[2]
dct_att[oid] = val
j = polygon.JSON
if j in dct_geo:
lst_oid = dct_geo = lst_oid
else:
dct_geo = [oid]
del row, curs
dct_res = {}
for j, lst_oid in dct_geo.items():
lst_val = [dct_att[oid] for oid in lst_oid]
tot = sum(lst_val)
cnt = len(lst_val)
mean = tot / float(cnt)
dct_res[lst_oid[0]] = mean
flds = ("OID@", fld_keep, fld_res)
with arcpy.da.UpdateCursor(fc_in, flds) as curs:
for row in curs:
if row[0] in dct_res:
row[1] = 1
row[2] = dct_res[row[0]]
curs.updateRow(row)
del row, cursWhen you query the output featureclass using XB_Keep = 1 you and display on Mean_BL the result looks like this:
as opposed to drawing the overlapping input polygons on BL_SMUNDI_:
I will adapt the code to make it a bit friendlier and to correct for the string input fields that hold values.
Kind regards, Xander
