search

AreteY / post-wildfire-recovery

An Earth Lab Certificate project studying post-wildfire recovery.
MIT License
0 stars 1 forks source link

Evaluate field data for NEON GRSM site #3

Closed AreteY closed 2 years ago

AreteY commented 2 years ago

Look at pre- and post-Chimney Tops 2 fire vegetation type and cover for Great Smoky Mountains National Park.

Chathu84 commented 2 years ago

Please add NEON field data and their citation as well to data sources.

AreteY commented 2 years ago

Hi @Chathu84 - I've downloaded the shapefiles for all of the NEON Terrestrial Observation Plots and plotted them within the fire boundary. My next step is get the tiles and vegetation that correspond with the plots.

Chathu84 commented 2 years ago

Hi @AreteY Yes, That is great. Then we can extract spectral signatures from those plots (for trees and other veg classes if NEON have specifically collected information on) to build our library.

AreteY commented 2 years ago

Hi @Chathu84 I've confirmed that the same TOS plots are reviewed in 2016 (pre-fire) and 2017 (post-fire). There are 10 TOS plots within the fire perimeter that are in 9 tiles. The field data classify for NLCD class, scientific name (lowest taxonomic rank), and scientific family. Which classification should we use? Also should I get the pixel spectra from the dNBR spectra?

Chathu84 commented 2 years ago

Hi @AreteY Lets use the NLCD class. The coarser one. However, if all the plots data are categorized into one NLCD class (because that is a very coarse classification scheme) then we would need to go to family. Yes, please get the DNBR spectra from the dNBR spectra.

AreteY commented 2 years ago

Hi @Chathu84 All of the plots are categorized as the deciduous forest NLCD Class. The family classification is complicated (can be more than 20 unique family classes per plot), but we can identify with most frequent by count (ie. plot 001 Rubiaceae, plots 007 and 008 Smilacaceae, etc.). There is one more column called 'otherVariables', which contains things like wood, moss, lichen, litter, soil, and standing dead. Plot 059 has the most standing dead classifications (count=5), then plot 008 with 2, and plot 055 with 1.

AreteY commented 2 years ago

Hi @Chathu84 I've looked at the plots again and found that for each plot ID, there are 8 subplots that are 1-m^2 and we can get the coords for these subplots using the API. It's strange when you look at the percent cover for the 'otherVariables' (non-vegetation), the total percent cover is usually around 99. But the sum of the vegetation (vegetation entries that are assigned a family) is a smaller number. For example, for plot 55, the sum of the vegetation percent cover varies from 13.5 to 34.5 for the different subplots. Does this mean, for example, that we can say for subplot 40.1.1, that the total vegetation is 13.5% and the rest 86.5% is non-vegetation (and we know the partitioning by the otherVariables column)? Does this seem reasonable?

Chathu84 commented 2 years ago

Hi @AreteY Yes, for that subplot, 13.5% is the veg percentage and the rest is the others. It's great we have more subplots (within big plots) and have their coordinates.

AreteY commented 2 years ago

Closing field data issue #3