Can we see the impacts of early season soil treatment?

Here I am trying to look at whether we can observe differences in GCC_90 based on early season soil management practices. I created labels for four classes of early season ground cover:

Burnt

Cleared

Thin Crop Bunches Two classes are made for “bunches”. These site look like they are harvested by hand. We should figure out what exactly they are. “Thin” bunches tend to be more intermittent, shorter, and often lay flat against the soil.

Thick Crop Bunches Thick “bunches” tend to stand upright and have much higher density of standing material.

Ground Litter

Differences between groups

First step is to see if there are differences in the distribution of GCC for each group

unique(in_data$Soil_Updated)
## [1] NA                   "Cleared"            "Growth, Not Sorted"
## [4] "Ground Litter"      "Crop Bunches Thick" "Burnt"             
## [7] "Crop Bunches Thin"
"./Data/sorted/Soil_Updated/Burnt/128940-5778e4db-3b35-4839-b6ce-4f0971c70bb2-1479277942199.jpg"
## [1] "./Data/sorted/Soil_Updated/Burnt/128940-5778e4db-3b35-4839-b6ce-4f0971c70bb2-1479277942199.jpg"
# assign labels to all sites where soil label is assigned 
for(label in c( "Ground Litter","Crop Bunches Thick","Crop Bunches Thin" ,"Burnt","Cleared")){
  in_data$final_soil_class[in_data$userfield %in% unique(in_data[in_data$Soil_Updated==label,'userfield'])] =label
}

# number of observations per class
table( in_data$final_soil_class)
## 
##              Burnt            Cleared Crop Bunches Thick 
##                 95               1734                112 
##  Crop Bunches Thin      Ground Litter 
##                127                335
ggplot() +geom_boxplot(data=in_data[in_data$final_soil_class!='NA',],aes(x=final_soil_class, y=gcc_90,fill=factor(final_soil_class)))  + 
    theme_minimal() +theme(axis.text.x = element_text(angle = 90, hjust = 1))+xlab('')

Test

## 
##  Welch Two Sample t-test
## 
## data:  litter and cleared
## t = 8.0699, df = 479.12, p-value = 5.711e-15
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.03119988 0.05128367
## sample estimates:
## mean of x mean of y 
## 0.5593798 0.5181380

Although not that surprising given the graph we can see that there is a statistically significant difference between GCC values for cleared and ground with litter.

Maximum GCC values by soil treatment

Now let’s see if there are differences in the maximum GCC-90 value by soil class (going forward this should be run on smoothed data).

max_data = in_data %>% group_by(final_soil_class,userfield) %>%   summarise(`Max GCC_90` = median(gcc_90, na.rm = TRUE))

ggplot() +geom_boxplot(data=max_data[max_data$final_soil_class!='NA',],aes(x=final_soil_class, y=`Max GCC_90`,fill=factor(final_soil_class)))  + 
    theme_minimal() +theme(axis.text.x = element_text(angle = 90, hjust = 1))+xlab('')

Summary

Looks like we can use this data to evaluate the impact of climate smart data practices on GCC or even NDVI. Clearly more work is needed here but hoping this gets a conversation going.