addEEtoSpOcc_multi: Add Google Earth Engine covariate data to a multi-season...

In AfricaBirdData/ABAP: Access to African Bird Atlas Project data from R

Add Google Earth Engine covariate data to a multi-season spOccupancy data list

Description

Google Earth Engine (GEE) data for each pentad can be extracted to a data frame using ABAP's addVarEEcollection and addVarEEimage functions. addEEtoSpOcc_multi can then be used to add these types of data to a list in order to fit single-species multi-season occupancy models using stPGOcc (spatial) or tPGOcc (non-spatial) from the spOccupancy package.

Usage

addEEtoSpOcc_multi(
  spOcc,
  ee_data,
  ee_assign,
  process = c("occ", "det"),
  seasons = NULL
)

Arguments

spOcc	a list containing ABAP detection/non-detection data returned by abapToSpOcc_multi.
ee_data	a data frame with GEE data extracted using addVarEEcollection or addVarEEimage. The data frame needs to contain a column called pentads with the pentad ID from data extracted using the GEE functions. The remaining columns should only contain the GEE covariate values that are intended to be added to the spOccupancy list. For ease of use in occupancy model formula notation it's recommended that the variable names in the data frame are concise and informative, and don't contain spaces. See the example below for how to create this data frame after extracting GEE data.
ee_assign	"site" if the covariate only varies between sites or "site-season" if the covariate varies across sites and between seasons.
process	either 'occ' or 'det' specifying whether the covariate is affecting occupancy or detection processes, respectively.
seasons	string indicating the name of the variable in ee_data that should be used to define seasons in multi-season models.

Details

If ee_assign = "site-year" ee_data must be a data frame-like object (sf could work) with each covariate in one column, the pentad identifier in another column named pentad, and the season identifier in another column (with the same name specified in season). if ee_assign = "site" ee_data must be a data frame-like object with each covariate in one column, and the pentad identifier in another column named 'pentad'.

Value

The data list spOcc with the additional covariates.

Note

The numeric ranges of various GEE data can be vastly different so it is advised that you scale your covariate data before running an occupancy model.

Author(s)

Dominic Henry [e-mail]
Pachi Cervantes

See Also

abapToSpOcc_multi, stPGOcc, tPGOcc, addVarEEimage, addVarEEcollection

Examples

## Not run: 
 ## rgee and ABDtools are required to annotate data with Google Earth Engine
library(rgee)
library(ABDtools)
library(dplyr)

## Extract ABAP pentad data
abap_pentads <- getRegionPentads(.region_type = "province",
                                 .region = "Eastern Cape")

## Extract multi-season ABAP bird data
abap_multi <- getAbapData(.spp = 212,
                          .region_type = "province",
                          .region = "Eastern Cape",
                          .years = c(2009,2010,2011,2012))

## We will use years as occupancy reference seasons
abap_multi$year <- format(abap_multi$StartDate, "%Y")

## Create unmarked frame (with X & Y coords as site covariates)
spOcc_multi <- abapToSpOcc_multi(abap_multi, abap_pentads, seasons = "year")
str(spOcc_multi)

## Start up GEE
ee_check()
ee_Initialize(drive = TRUE)

## Create assetId for pentads of interest
assetId <- file.path(ee_get_assethome(), 'EC_pentads')

## Upload to pentads to GEE (only run this once per asset) and
## load the remote asset into R session
pentads <- uploadFeaturesToEE(feats = abap_pentads,
                              asset_id = assetId,
                              load = TRUE)

### ADD SITE-YEAR VARIABLES ###

## Create a multi-band image with mean NDVI for each year
ndvi_multiband <- EEcollectionToMultiband(collection = "MODIS/006/MOD13A2",
                                          dates = c("2009-01-01", "2012-12-31"),
                                          band = "NDVI",
                                          group_type = "year",
                                          groups = 2009:2012,
                                          reducer = "mean",
                                          unmask = FALSE)

## Find mean and sd NDVI value for each pentad and year from the multi-band image
ndvi_mean <- addVarEEimage(pentads, ndvi_multiband, "mean")
ndvi_sd <- addVarEEimage(pentads, ndvi_multiband, "stdDev")

## Format the data to include the pentad column and GEE values for each year
ndvi_mean <- ndvi_mean %>%
    select(pentad, paste0("NDVI_", as.character(2009:2012)))

ndvi_sd <- ndvi_sd %>%
    select(pentad, paste0("NDVI_", as.character(2009:2012)))

## Create a data frame with covariate columns, pentad and season.
ee_siteyear <- ndvi_mean %>%
    as.data.frame() %>%
    dplyr::select(-geometry) %>%
    tidyr::pivot_longer(-pentad, names_to = "year", values_to = "ndvi_mean")

## Add GEE site-year data to spOccupancy data list. We will add these covariates
## to the detection process.
spOcc_multi_ee1 <- addEEtoSpOcc_multi(spOcc_multi,
                                      ee_data = ee_siteyear,
                                      ee_assign = "site-season",
                                      process = "det",
                                      seasons = "year")

str(spOcc_multi_ee1)

### ADD SITE VARIABLES ###

## Annotate data with mean elevation per pentad
elev_mean <- addVarEEimage(ee_feats = pentads,
                           image = "MERIT/DEM/v1_0_3",
                           reducer = "mean",
                           bands = "dem",
                           unmask = FALSE)

## Keep only the columns that we need
ee_site <- elev_mean %>%
    as.data.frame() %>%
    dplyr::select(pentad, dem_mean)

## Add GEE site data to spOccupancy data list. We will now add it to the
## occupancy process.
spOcc_multi_ee2 <- addEEtoSpOcc_multi(spOcc_multi,
                                      ee_data = ee_site,
                                      ee_assign = "site",
                                      process = "occ")

## End(Not run)

AfricaBirdData/ABAP documentation built on Aug. 4, 2024, 4:41 p.m.