Hello!
I’m currently using the InVEST Nutrient Delivery Ratio (NDR) model with input DEM and land use data at a 30m × 30m resolution. The rainfall (used as the runoff proxy) is at a coarser resolution of 1000m × 1000m. The model outputs for nitrogen and phosphorus have the following characteristics:
Nitrogen: max = 1.079, mean = 0.13
Phosphorus: max = 0.14, mean = 0.02
Since my input and output raster resolutions are 30m × 30m, I understand that the output units are in kg per pixel, which corresponds to kg per 900 m².
To convert these results to kg per ha, I divide the values by 0.09 (since 900 m² = 0.09 ha). However, after doing this conversion, the nutrient values become quite large (e.g., N max becomes ~12 kg/ha), which feels too high and inconsistent with expected nutrient export levels in similar studies.
I’m wondering:
Are these output values (before conversion) reasonable and already normalized per hectare in some way?
Is dividing by 0.09 the correct approach, or is there another recommended method to convert to kg/ha?
Could the coarse resolution of the rainfall proxy (1 km) be affecting the nutrient export magnitude?
Any clarification on how to properly interpret and convert the output units would be greatly appreciated. Thank you!
Which version of Workbench are you using? If you are using version 3.15 or 3.16, the output values are already in kg/hectare, as the User Guide notes. If you are using a version older than 3.15, then the values are in kg/pixel, and you would need to make the conversion to kg/hectare.
Thank you very much for your reply!
I’m currently using InVEST version 3.14.3.
May I ask if the output results still need to be converted in terms of units?
If so, should I simply divide the output values by 0.09 to convert them to kg/ha?
I really appreciate your time and guidance!
Yes, if you’re using version 3.14.3, the results are in kg/pixel, and if you want results in kg/hectare you need to divide by 0.09.
Could the coarse resolution of the rainfall proxy (1 km) be affecting the nutrient export magnitude?
The resolution itself shouldn’t affect the magnitude (especially since this layer is normalized by the model), but it will probably affect the pattern in the resulting maps, you may be able to see the coarse pattern visually.
I don’t know what a reasonable nutrient export value is for your study area, but remember that this is an annual average value. Also, double check the units and values of your inputs to understand how they are contributing to the results. Finally, in order to feel confident in the absolute values produced by the model, we recommend calibrating it with real-world data.