AREACONNECT

ANA

16.5.9: AREACONNECT

AREACONNECT(data1, data2 [, compact] [, /RAW])

Determines connections between areas in data data1 and data2, based on spatial overlap. The data must be arrays of type LONG. Each element of data1 must contain a number that identifies the area or class that the element is a part of, for example determined by AREA, and likewise for data2. A data value equal to zero is regarded as background, and data elements with zeros in both data1 and data2 are ignored altogether.

The data elements with any particular area number need not form a contiguous set; a particular data element from data1 is only compared to the corresponding data element from data2, and to no others. The numbers in data1 are independent of the numbers in data2: No special relationship is inferred if some area in data1 has the same ID number as some area in data2.

The return value of this function is a two-dimensional LONG array with 3 elements in its second dimension. If we call that array R, then R(*,0) contains area numbers from data1, R(*,1) contains area numbers from data2, and R(*,2) contains the number of data elements that have that combination of data1 and data2 numbers. Data elements that have a value of zero in both data arrays are ignored. The results are sorted fi st in ascending order of data1 values, and then, for entries with equal data1 values, in ascending order of data2 values.

If /RAW is specified, then nothing more is done. Otherwise, the areas are divided into topological classes and relevant information is returned in global variables, as listed in the following table. If there are no entries in any particular topological class, then a scalar -1 is returned in the corresponding global variable. Otherwise, a LONG array is returned.

$APP: Those areas that have non-zero data2 values but no non-zero data1 values are appearing. Their data2 area numbers are returned in ascending order in $APP.
$DISAPP: Those areas that have non-zero data1 values but no non-zero data2 values are disappearing. Their data1 area numbers are returned in ascending order in $DISAPP.
$SPLIT1, $SPLIT2, $SPLIT2_INDEX: Those data1 areas that have overlap with more than one data2 area are splitting. Their data1 (i.e., pre-split) area numbers are returned in ascending order in $SPLIT1. The data2 (i.e., post-split) area numbers corresponding to pre-split data1 area number $SPLIT1(I) are given, in ascending order, by $SPLIT2 ( $SPLIT2_INDEX(I) : $SPLIT2_INDEX(I + 1) - 1).
$MERGE1, $MERGE2, $MERGE1_INDEX: Those data2 areas that have overlap with more than one data1 area are the product of merging. Their data2 (i.e., post-merge) area numbers are returned in ascending order in $MERGE2. The data1 (i.e., pre-merge) area numbers corresponding to post-merge data2 area number $MERGE2(I) are given, in ascending order, by $MERGE1( $MERGE1_INDEX(I) : $MERGE1_INDEX(I + 1) - 1).
$STAY1, $STAY2: Those areas that neither appear, disappear, merge, nor split, are staying topologically the same. Their data1 area numbers are returned in ascending order in $STAY1, and the corresponding data2 area numbers in $STAY2.

At this point, the area numbers in data2 are not related to the area numbers in data1, so even areas that remain topologically the same from data1 to data2 do not have the same area number in both data sets. Such a relation can be requested by specifying an appropriate value for compact, in which case the values in data2 and related global variables are modified to best express their relation to data1. The numbers in data1 and related global variables are not modified.

If compact is unspecified, negative, or equal to 0, then nothing is modified. Otherwise, the data2 areas in the class of stayers receive the corresponding number from data1 (so that $STAY2 becomes equal to $STAY1). For splitters, the first of the data2 split products inherits the area number of the data1 source area, so that $SPLIT2($SPLIT2_INDEX) becomes equal to $SPLIT1. For mergers, the data1 area number of the first merger component is propagated to the data2 merger product, unless that area number was already used earlier, in which case the remaining merger components are inspected until an unused area number is found or the last component was dismissed.

After this process (if compact is positive), there may still be some data2 areas that have not yet received a new number. If compact is equal to 1, then those areas receive successive numbers starting at one greater than the greatest number in data1. If compact is equal to 2, then those areas receive numbers that are as small as possible but not yet otherwise used in data2 and not less than 2. If compact is greater than 2, then those areas receive successive numbers starting at one greater than the value of compact or the greatest number in data1, whichever is greater.