I agree this is kind of confusing! I checked with the science team to get an answer to this, and I’ll update the user’s guide to explain it better.
If you look in the
model_data directory in the Crop Production sample data you’ll find:
observed_yield has a global map of actual observed crop yield for each crop in the year 2000
climate_bin_maps has a global climate map for each crop, based on growing degree days (GDD).
climate_percentile_yield_tables contains a table for each crop, showing the percentile yields for each climate bin for that crop across the world. For example, the 95th percentile value for wheat in
climate bin 1 is 3.763889. So, 3.763889 is the 95th percentile for wheat growing in
climate bin 1 areas across the world. In other words, 95% of areas that grow wheat in
climate bin 1 produce less than 3.763889 tons/hectare.
The model maps the percentile yield values from the tables in
climate_percentile_yield_tables to the maps in
climate_bin_maps covering the area of interest. The resulting percentile yield maps give an idea of the possible range of yields that you could expect growing that crop in that area. For instance, the 50th percentile map would be an “average” yield, and the 95th percentile map would be a near-optimal yield attained by improving farming practices etc.
There are also
crop_observed_production.tif outputs, which are derived from actual yield data from the
observed_yield directory. You can use the observed production data to know the “current” (year 2000) actual crop production is. The percentile yield outputs are useful if the crop isn’t currently grown in the area of interest, or to get an idea of how much the yield could increase by more efficient farming practices etc.