The model outputs can be found nested within the out/ directory (see Structure). STILT outputs two files for later use. Particle trajectories are saved to a _traj.rds file and gridded footprints saved to a _foot.nc file.
Particle trajectories and simulation parameter information are packaged and saved in a compressed .rds file (serialized single R object) with the naming convention with the naming convention YYYYMMDDHH_LONG_LATI_ZAGL_traj.rds. This allows regridding of the footprints without recalculating particle trajectories.
This object is can be loaded with readRDS() and is structured as
out <- readRDS('YYYYMMDDHH_LONG_LATI_ZAGL_foot.rds')
str(out)
# List of 4
# $ runtime : POSIXct[1:1], format: "2015-06-16"
# $ file : chr "/uufs/chpc.utah.edu/common/home/u0791983/stilt-sims/test/out/2015061600_-111.847672_40.766189_10/2015061600_-111.847672_40.7661"| __truncated__
# $ receptor:Classes ‘tbl_df’, ‘tbl’ and 'data.frame': 1 obs. of 4 variables:
# ..$ run_time: POSIXct[1:1], format: "2015-06-16"
# ..$ lati : num 40.8
# ..$ long : num -112
# ..$ zagl : num 10
# $ particle:Classes ‘tbl_df’, ‘tbl’ and 'data.frame': 128441 obs. of 26 variables:
# ..$ time: num [1:128441] -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 ...
# ..$ indx: num [1:128441] 1 2 3 4 5 6 7 8 9 10 ...
# ..$ long: num [1:128441] -112 -112 -112 -112 -112 ...
# ..$ lati: num [1:128441] 40.8 40.8 40.8 40.8 40.8 ...
# ..$ zagl: num [1:128441] 61.3 100.6 87.3 98.3 88.6 ...
# ..$ sigw: num [1:128441] 1.15 1.15 1.15 1.15 1.15 ...
# ..$ tlgr: num [1:128441] 8.05 8.05 8.05 8.05 8.05 ...
# ..$ zsfc: num [1:128441] 1532 1532 1532 1532 1532 ...
# ..$ icdx: num [1:128441] 2 2 2 2 2 2 2 2 2 2 ...
# ..$ temp: num [1:128441] 301 301 301 301 301 ...
# ..$ samt: num [1:128441] 2 2 2 2 2 ...
# ..$ foot: num [1:128441] 0.0219 0.0219 0.0219 0.0219 0.0219 ...
# ..$ shtf: num [1:128441] 0 0 0 0 0 0 0 0 0 0 ...
# ..$ tcld: num [1:128441] 87.9 87.9 87.9 87.9 87.9 ...
# ..$ dmas: num [1:128441] 1.104 1.104 1.104 0.991 1.119 ...
# ..$ dens: num [1:128441] 0.968 0.965 0.966 0.965 0.966 ...
# ..$ rhfr: num [1:128441] 0.277 0.282 0.28 0.282 0.28 ...
# ..$ sphu: num [1:128441] 0.00222 0.00222 0.00222 0.00222 0.00222 ...
# ..$ solw: num [1:128441] -99 -99 -99 -99 -99 -99 -99 -99 -99 -99 ...
# ..$ lcld: num [1:128441] -99 -99 -99 -99 -99 -99 -99 -99 -99 -99 ...
# ..$ zloc: num [1:128441] -999 -999 -999 -999 -999 -999 -999 -999 -999 -999 ...
# ..$ dswf: num [1:128441] 25.2 25.2 25.2 25.2 25.2 25.2 25.2 25.2 25.2 25.2 ...
# ..$ wout: num [1:128441] 0.722 0.722 0.722 0.722 0.722 ...
# ..$ mlht: num [1:128441] 329 329 329 329 329 ...
# ..$ rain: num [1:128441] 4.02e-07 4.02e-07 4.02e-07 4.02e-07 4.02e-07 ...
# ..$ crai: num [1:128441] -0.9 -0.9 -0.9 -0.9 -0.9 -0.9 -0.9 -0.9 -0.9 -0.9 ...
Particle trajectory data is stored in a data frame with columns corresponding with varsiwant and can be accessed with out$particle.
Footprints are packaged and saved in a compressed .nc file conforming to the Climate and Forecast (CF) metadata convention with the naming convention YYYYMMDDHH_LONG_LATI_ZAGL_foot.nc. This object contains information about the model domain, the grid resolution, and footprint values. This object is typically a three dimensional array with dimensions ordered (x, y, z). However, the object will only have dimensions (x, y) for time integrated footprints.
ncdump -h 2015061822_-111.980323_40.782561_5_foot.nc
netcdf 2015061822_-111.980323_40.782561_5_foot.nc {
dimensions:
lon = 550 ;
lat = 500 ;
time = 24 ;
variables:
double lon(lon) ;
lon:units = "degrees_east" ;
lon:standard_name = "longitude" ;
lon:long_name = "longitude at cell center" ;
double lat(lat) ;
lat:units = "degrees_north" ;
lat:standard_name = "latitude" ;
lat:long_name = "latitude at cell center" ;
double time(time) ;
time:units = "seconds since 1970-01-01 00:00:00Z" ;
time:standard_name = "time" ;
time:long_name = "utc time" ;
time:calendar = "standard" ;
float foot(time, lat, lon) ;
foot:units = "ppm (umol-1 m2 s)" ;
foot:_FillValue = -1.f ;
foot:standard_name = "footprint" ;
foot:long_name = "stilt surface influence footprint" ;
// global attributes:
:crs = "+proj=longlat" ;
:crs_format = "PROJ.4" ;
:documentation = "github.com/uataq/stilt" ;
:title = "STILT Footprint" ;
:time_created = "2018-05-14 19:49:09" ;
}
For those familiar with raster operations, the default output adheres to the CF-1.4 metadata conventions which is inherently compatible compatible with the R raster package. For more information about raster manipulation, the Raster R package is a good place to start.
If time_integrate = TRUE, the footprint .nc files can be loaded directly with raster(). If time_integrate = FALSE, the data can be loaded with brick(), which is a three dimensional version of a standard raster. The POSIX time (UTC seconds since 1970-01-01) is stored in the Z dimension and can be easily accessed with getZ().
Using the Raster R package, the data can be loaded with brick() and is structured as
library(raster)
out <- brick('2015070211_-111.835_40.763_7_foot.nc')
out
# class : RasterBrick
# dimensions : 280, 390, 109200, 5 (nrow, ncol, ncell, nlayers)
# resolution : 0.002, 0.002 (x, y)
# extent : -112.3, -111.52, 40.39, 40.95 (xmin, xmax, ymin, ymax)
# ...
time <- as.POSIXct(getZ(out), tz = 'UTC', origin = '1970-01-01')
str(time)
# POSIXct[1:5], format: "2015-07-02 05:00:00" "2015-07-02 06:00:00" ...
Alternatively, the data can be loaded using standard netCDF methods.
library(ncdf4)
nc <- nc_open('2015070211_-111.835_40.763_7_foot.nc')
nc
# File 2015070211_-111.835_40.763_7_foot.nc (NC_FORMAT_CLASSIC):
# 1 variables (excluding dimension variables):
# float Footprint[longitude_center,latitude_center,time]
# units: ppm (umol-1 m2 s)
# _FillValue: -1
# 3 dimensions:
# longitude_center Size:390
# units: degrees_east
# long_name: longitude_center
# position: cell_center
# latitude_center Size:280
# units: degrees_north
# long_name: latitude_center
# position: cell_center
# time Size:5
# units: seconds since 1970-01-01
# long_name: time
# timezone: UTC
# 7 global attributes:
# crs: +proj=longlat +ellpsWGS84
# crs_format: PROJ.4
# Conventions: CF-1.4
# Title: STILT Footprint Output
# Compatibility: raster::raster() and raster::brick()
# Documentation: benfasoli.github.io/stilt
# Author: Ben Fasoli
To get your feet wet, try one of the tutorials such as simulating carbon dioxide for the William Browning Building at the University of Utah.