We have developed a half meter resolution landcover map from a pansharpened image Pleiades image. We want to resample the landcover raster to higher resolutions (15 and 30m). What would be the most recommended resampling technique to do this?
This is a landcover map of area not larger than 200 square kilometers. We want it to be used as an input in a hydrologic model. As the remaining of our data sets (DEM and soil raster) are coarser resolution the model will automatically resample our half meter resolution landcover raster to a coarser resolution raster using one of its default resampling methods. We want to avoid this and instead apply a recommended resampling technique. Any experience regarding an ideal method to do the resampling would be much appreciated.
The sampling scheme you choose will probably affect your simulation results. We have investigated this issue in the following paper:
Zhao, S. and Liu, S., 2014. Scale criticality in estimating ecosystem carbon dynamics. Global change biology, 20(7), pp.2240-2251.
The best solution might be using the composition of land cover rather than any individual class in your model. However, most models are not able to deal with multiple land cover classes in a mapping unit or grid cell. So the next option might be the majority resampling approach. Of course, you can derive a set of different land cover maps from the original data using different resampling schemes, and then investigate their impacts on hydrologic modeling to assess the uncertainty derived from land cover resampling.
Thematic raster aggregation in Erdas...
P.S. "We want to resample the landcover raster to higher resolutions (15 and 30m)." not higher, but lower...
Right Mikhail, I used "higher" keeping spatial resolution in mind. I should have used the appropriate word. Thanks for the recommended method, we will consider applying it.
Hi Prerna, sorry for a late reply.
The tool which I'll be using requires 3 data sets: DEM, Soil and land cover. The DEM and the soil data are 15m and 1km respectively. My modeling tool will reduce the resolution of the 0.5m landcover data which I'll use as input in anyway. So instead of the tool resample it to one of its default resampling techniques I want to supply it with a resampled one so that the process can be controlled interms of uncertainty.
As the model undergoes a cell by cell analysis it resamples the input data to the same resolution. The most inflential input will be the DEM. Hence the remaining two input (landcover and soil) will be aggregated (in case of landcover from 0.5m to 15m) or resampled (in the case of soil from 1km to 15m).
Dear, what is the importance of resampling? for instance for my study, the DEM has a 12.5 by 12.5 m resolution, while the soil has 150 m and the land use is 30 by 30 m resolution. is it a must to resample?
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude