I’m currently working with the Urban Nature Access model and had a question regarding the interpretation of demand values.
In my results, I’m seeing a demand value of approximately 25 m² in a single 10x10 meter pixel. In this pixel, the population density is about 1,000 people, and I’ve set the per capita standard (g_cap) to 9 m² per person.
According to the demand formula:
〖demand〗_i=P_i*g_cap
I would expect the total demand in that pixel to be around 9,000 m² (i.e., 1,000 people × 9 m²/person).
However, I’m unsure how to interpret a result of just 25 m² in that pixel. I understand that access is not limited to the pixel itself and is distributed spatially based on proximity and decay functions. But from the outputs alone, how can I determine whether the demand is high or low without calculating it manually from population × g_cap?
Could you clarify how the model reports and scales demand values per pixel, and whether those outputs are normalized, redistributed, or represent raw totals?
First, I’d like to double check something about your population raster. You said, “In this pixel, the population density is about 1,000 people.” The model expects values in this raster to be a count of people per pixel, so I want to make sure that your 1000 value is strictly a count of people and not, say, people/m^2.
The resolution of your population raster versus the resolution of your LULC raster could also be contributing to perceived discrepancies. Prior to calculating demand, the model resamples the population raster to match the resolution of the LULC raster (and does so in a way designed not to change overall population totals). This allows us to compare pixels directly, so we can answer questions like “how many people reside within N meters of this urban nature pixel?”
In the Intermediate directory of the model results, there should be a raster called aligned_population.tif, which reflects the population counts per pixel after the resampling step. These are the pixel values that are used in the demand calculation. Taking a look at this raster (and how it differs from your input population raster) may help make things clearer.
Let us know if that helps, or if you still have additional questions!