Areas calculated on Saiku

Dear Imem,

As we speak we are working on a different way of calculating areas, but for older versions of Collect Earth surveys the area calculations are rather straight forward, here is an attempt to explain it:

Inside a Collect Earth project file there is a file called region_areas.csv. In the case of Tunisia this file looks like this :

Province,Plot File,Area(HA),Weight
ariana,,45000,1
ben_arous,,66200,1
bizerte,,355000,1
beja,,358700,1
gabes,,747000,1
gafsa,,758100,1
jendouba,,329400,1
kairouan,,645100,1
kasserine,,825000,1
... and so on

The first column is the name of the region, the second column is the name of the Plot File that contained the plot information (it can be empty if the region name was collected), the third column is the total area of the Region in Hectares and finally the last column is the weight of the plot (by default it is 1 ).

To calculate the expansion factor of a plot (the area in hectares that the plot represent) Collect Earth divides the Area of the region (provided as we saw in the region_areas.csv) by the number of plots within the region. As we are using systematic sampling designs the expansion factor would be the same for all of the plots within a region. When Saiku performs calculations using the Area measure what is does is to add the expansion factors of the plots in the results.

The weight of a plot is quite simple to calculate too, if you have defined that all the regions have a sampling design (grid) of 10 by 10 kilometers and the region of Jendouba (because is desertic or some other reason ) has a sampling design with plots that are separated by 20 by 20 kilometers this means that the weight of the plots in Jendouba are four times as much as the weights of the plots of the other provinces --> There are four times less plots in the province of Jendouba ( per square kilometer ) as in the other provinces, thus the results for plots inside this province have to weight four times as much. The resulting region_areas.csv file should then be :

Province,Plot File,Area(HA),Weight
ariana,,45000,1
ben_arous,,66200,1
bizerte,,355000,1
beja,,358700,1
gabes,,747000,1
gafsa,,758100,1
jendouba,,329400,4
kairouan,,645100,1
kasserine,,825000,1
... and so on

Notice how now all the provinces have a weight of 1 except Jendouba that has a weight of 4.

This approach to calculate weights and areas assumes that your survey has a "region" attribute. Most of the surveys that we have worked with have it. Usually the actual value of the region per each plot comes directly from the sampling design ( check the CSV/CED files where the plot locations are defined and you will notice a region column) so that the operators don't need to fill this information and it is handled directly by Collect Earth.

In the future we will have a similar mechanism to calculate the expansion factors and measures that does not force you to use the "region" attribute but instead let you calculate these expansion factors based in other attributes such as country or ecologic zone.

Did this answer your question?