OPERATOR(S) NAME:Victor H. Rivera-Monroy, Robert Twilley, Michael Kemp

DATE: April 30, 2001

Create as much space as needed within the form.

DATABASE INFORMATION

Geographic Zones Selected (leave blank if done by range):

Geographic Range Selected (leave blank if done by zones):

LATIN AMERICA

Cell types selected from database (Terrestrial, Coastal, OceanicI, OceanicII, OceanicIII):

COASTAL

Variables and ranges selected from database (if filtered – if not filtered, no entry is needed):

FROST_DAYS_TOTAL NUMBER <10

SEA TEMPERATURE _MINIMUM_MONTH: >15.5

SEA TEMPERATURE _INTER_ANNUAL_STDEV: <1

Variables added (if any):

What do you expect to get out of clustering these variables or what are you trying to achieve?

Test if these variables can be used as proxies to map mangrove distribution in Latin America

Results are in following files:

Exp1.htm (Table)

Exp1map.gif (Map)

LOICZVIEW INFORMATION

Save the Cluster summary – if will contain much of the information needed!

(Instructions for summary and data saving are at the end of this document)

Cluster Run Number:_3_

Variables selected in cluster (include any weights if not set at 1): Frost days, Sea Temperature, , Sea temperature _inter_annual_stdev

Number of Clusters: 3

Distance Method: [ X ]Average [ ]Maximum

COMMENTS OR DESCRIPTION ON CLUSTER OUTPUT:

The cluster follows very closely the actual distribution of mangroves. However, in the Pacif coast the boundary between Peru and Ecuador extends into colder regions where mangrove forests are not actually present.

Additional Comments on problems or improvements needed:

**************************************************************************************

-We suggest to include in the the Cluster display an overlapping map with actual lat/long data. This feature can help to locate specific points to delimit boundaries among areas/zones and to identifiy particular points where the user wants to acquire more descriptive information from another sites or archived data.

-We strongly recommend to keep the value in the filter window for further revision inside the metadata variable selection window—currently it is deleted when the user goes back to try to change it. Also if the user wants to confirm that the value in the filter window is correct the program add this “review” to the filtered variables table.