GeoAI_2021_unit_04_EX_Create Masks

[Baldl]

No description provided.

[seegerkonstantin]

R is unable the find the "AllEqual" function/command in the above created "remove_files" function. I looked on the internet if this is a potential typo or I forgot to load one of the packages but I could not find the "AllEqual" function and the "all.equal" argument by base R doesnot work.

[gisma]

@seegerkonstantin ALLEqual is a function of the package greenbrown . So pls try to reinstall the package and try greenbrown::AllEqual

[FloFranz]

I have the same error in applying the remove_files function. Then I updated my R version, but it doesn`t helped. Error message: Error in loadNamespace(x) : es gibt kein Paket namens ‘greenbrown’ . I tried to install the package greenbrown from source with install.packages("greenbrown", repos="http://R-Forge.R-project.org"), but the same error occurs. I also tried to download a *.tar file with an older version of the greenbrown package as described here...not helped. Any other idea how it can be fixed?

[seegerkonstantin]

@FloFranz I solved this problem by installing greenbrown with the following command: "install.packages("greenbrown", repos="http://R-Forge.R-project.org")" as described on this side https://greenbrown.r-forge.r-project.org/install.php .

[FloFranz]

@seegerkonstantin Confusing! I first had to install bfast, phenopix and Kendall. Then it also worked to install greenbrown as mentioned above.

[gisma]

@FloFranz sorry for confusing. It is often helpful to have a look in the Description file of the package.There you find the dependencies in this case Imports: Kendall, bfast, phenopix, quantreg, parallel, plyr. In common this should work with dependencies = TRUE, however it sometimes fails than you can install this list manually.

[gisma]

As far as I see it you need to install the requested packages as you tried.

1. SDMTools'
2. phenopix
   However you were not able to install SDMTools for your R version. So what you can try next is:

• install it from source
  install.packages("SDMTools",,"http://rforge.net/",type="source")
• download the package and install it manually local via the MEnu in Rstudio

[gisma]

@SedaBekar

1. start the System Control Panel applet (Start - Settings - Control Panel - System)
2. select the Advanced tab
3. choose Environment Variables
4. at System Variables, select Path, then click Edit and add the path to Rtools e.g. D:\RTools

[gisma]

@SedaBekar this is fine - pretty good solution ! It would be the best du put it in your setup script or in the .Renviron file of the project.

[Muenchj4]

@gisma
I cannot install greenbrown. It says there were non-zero exit status. Is there a further way to install this package apart from install.packages and tar?

[Muenchj4]

@gisma I have tried to follow the way of Florian Franz, but I had still to install stucchange manually to be able to install greenbrown.

[gisma]

pleasse follow the above instructions. I have no other hints.

[Muenchj4]

@gisma Now another problem occurred when I tried to set the extent of the raster image. It says is.image_collection is not true. Why? And how can I change this?

[gisma]

Again pls Upload your complete and actual Scripts to a repo . Otherwise can not answer your questions.

[Muenchj4]

@gisma

#ex create masks
require(envimaR)
source(file.path(envimaR::alternativeEnvi(root_folder = "C:/Users/jomue/edu/geoAI",

•                                       alt_env_id = "COMPUTERNAME",
  

•                                       alt_env_value = "PCRZP",
  

•                                       alt_env_root_folder = "F:/BEN/edu/geoAI"),
  

•              "src/000-rspatial-setup.R"))
  

variable alt_env_id is NOT defined
'COMPUTERNAME' is set by defaultvariable alt_env_value is NOT defined
'PCRZP' is set by defaultvariable alt_env_root_folder is NOT defined
'F:/BEN/edu' is set by defaultvariable root_folder is NOT defined by calling script... 'root_folder' is set as rootDIR
variable path_prefix is NOT defined by calling script...
'path_' is set as default

variable fcts_folder is NOT defined by calling script...
'src/functions/' is set as default

read data

ras <- raster::stack(file.path(envrmt$path_data, "/marburg_dop.tif"))

subset to three channels

ras <- subset(ras, c("red", "green", "blue"))

download and crop the OSM building data to the extent of the raster data

buildings = opq(bbox = "marburg de") %>%

• add_osm_feature(key = "building") %>%
• osmdata_sf()

buildings = buildings$osm_polygons
buildings = sf::st_transform(buildings, crs(ras))
ras_extent <- extent(ras)
Error in extent(ras) : is.image_collection(x) ist nicht TRUE

[Muenchj4]

Here my setup script:
#------------------------------------------------------------------------------

basic rspatial course setup script

Type: script

Name: 000-rspatial-setup.R

Author: Chris Reudenbach, creuden@gmail.com

Description: create/read project folder structure and returns pathes as list

load all necessary packages

sources all functions in a defined function folder

Dependencies: devtools::install_github("envima/envimaR")

Output: list containing the folder strings as shortcuts

Copyright: Reudenbach/Nauss 2015-2021, GPL (>= 3)

Details: Please note that the queries on standard arguments have been set for

pragmatic reasons for a typical course setting. Of course, it is

intended to assign the arguments directly in the createEnvi function

Date: 2021-12-12

#------------------------------------------------------------------------------
require(envimaR)

1 - source files

#-----------------

source(file.path(envimaR::alternativeEnvi(root_folder = "C:/Users/jomue/edu/geoAI",
alt_env_id = "COMPUTERNAME",
alt_env_value = "PCRZP",
alt_env_root_folder = "F:/BEN/edu/geoAI"),
"src/geoAI_setup.R"))

rspatial packages

packagesToLoad = c("rgdal","png","tensorflow","keras","reticulate","osmdata","greenbrown","rsample","tfdatasets","purrr","magick","rprojroot","sen2r","patchwork","ggplot2","rpart","rpart.plot","rasterVis", "forcats","e1071","mapview", "mapedit", "tmap", "raster", "terra", "stars", "gdalcubes", "sf","dplyr","tidyverse","RStoolbox",
"randomForest", "ranger", "e1071", "caret", "link2GI","gdalUtils","raster","exactextractr", "doParallel", "CAST","ranger")

mandantory folder structure

projectDirList = c("data/",
"data/modelling/",
"data/modelling/model_training_data/",
"data/modelling/model_training_data/dop/",
"data/modelling/model_training_data/bui/",
"data/modelling/models/",
"data/modelling/prediction/",
"data/modelling/validation/",
"data/modelling/model_testing_data/",
"data/modelling/model_testing_data/dop/",
"data/modelling/model_testing_data/bui/",
"docs/",
"run/",
"tmp",
"src/",
"src/functions/"
)

append additional packages if defined by calling script

if (exists("appendProjectDirList") && appendProjectDirList[[1]] != "")
{
projectDirList = append(projectDirList,appendProjectDirList)
}

if (exists("appendpackagesToLoad") && appendpackagesToLoad[[1]] != "")
{
packagesToLoad = append(packagesToLoad,appendpackagesToLoad)
}

Now create/read root direcory, folder structure and load packages

NOTE root_folder MUST be defined in calling script

if (!exists("root_folder")) {
message("variable root_folder is NOT defined by calling script...\ 'root_folder' is set as rootDIR")
root_folder=rootDIR
}
if (!exists("alt_env_id")) {
message("variable alt_env_id is NOT defined by calling script...\n 'COMPUTERNAME' is set as default\n")
alt_env_id = "COMPUTERNAME"
}
if (!exists("alt_env_value")) {
message("variable alt_env_value is NOT defined by calling script...\n 'PCRZP' is set as default\n")
alt_env_value = "PCRZP"
}
if (!exists("alt_env_root_folder")) {
message("variable alt_env_root_folder is NOT defined by calling script...\n 'F:/BEN/edu' is set as default\n")
alt_env_root_folder = "F:/BEN/edu"
}

if (!exists("path_prefix")) {
message("variable path_prefix is NOT defined by calling script...\n 'path_' is set as default\n")
path_prefix = "path_"
}

root_folder = envimaR::alternativeEnvi(root_folder = root_folder,
alt_env_id = alt_env_id,
alt_env_value = alt_env_value,
alt_env_root_folder = alt_env_root_folder)

if (!exists("fcts_folder")) {
message("variable fcts_folder is NOT defined by calling script...\n 'src/functions/' is set as default\n")
fcts_folder = paste0(root_folder,"/src/functions/")
}

append additional folders if defined by calling script

if (exists("appendProjectDirList") && appendProjectDirList[[1]] != "")
{
projectDirList = append(projectDirList,appendProjectDirList)
}

call central function

envrmt = envimaR::createEnvi(root_folder = root_folder,
folders = projectDirList,
fcts_folder = fcts_folder,
path_prefix = path_prefix,
libs = packagesToLoad,
setup_script = "000-rspatial-setup.R",
source_functions = TRUE,
alt_env_id = alt_env_id,
alt_env_value = alt_env_value,
alt_env_root_folder = alt_env_root_folder)

set temp path to speed up raster package operations

raster::rasterOptions(tmpdir = envrmt$path_tmp)

suppres gdal warnings

rgdal::set_thin_PROJ6_warnings(TRUE)

define some color palettes

mvTop = mapview::mapviewPalette("mapviewTopoColors")
mvSpec = mapview::mapviewPalette("mapviewSpectralColors")
mvVec = mapview::mapviewPalette("mapviewVectorColors")
mvRas = mapview::mapviewPalette("mapviewRasterColors")

[gisma]

Die to possible dependencies in your functions and also due to a convenience etc. I rather orefer a complete GitHub repository.

[Muenchj4]

gh repo clone Muenchj4/geoAI

[Muenchj4]

In this repository you can find the scripts. I am still trying to follow up the dependencies.

[Muenchj4]

I have now tried to print the extent of the raster file mannualy like this:
"ras_extent <- c(xmin=480000,xmax=484000,ymin=5626000,ymax=5628000)"
But than I wanted to crop the extent like this:
" e_test <- c(xmin=483000,xmax=484000, ymin=5626000, ymax=5628000)
e_train <- c(xmin=480000, xmax=483000, ymin=5626000, ymax=5628000)
marburg_mask_train <- crop(rasterized_vector, e_train)"
The message occured:
"Error in crop(rasterized_vector, e_train) : is.cube(cube) ist nicht TRUE"
The same error occurred printing the cube of rasterized vector manually like this:
"marburg_mask_train <- crop(cube=c(xmin=480000,xmax=484000,ymin=5626000,ymax=5628000), extent=e_train)"
Error in crop(cube = c(xmin = 480000, xmax = 484000, ymin = 5626000, ymax = 5628000), :
is.cube(cube) ist nicht TRUE
Is there another way to crop the rasterized vector? Or is there a way to tackle this problem from the beginning?

[gisma]

I will Check IT asap

[gisma]

Have an Look at the Setup File .probably you will find some predefined extents . In Addition Check the crs are they identical?

[Muenchj4]

In the setup file I cannot find any predefined extent. I have tried to use proj4string instead of crs, but I still gets the message is.cube is not TRUE. What could be the reason?

[gisma]

@Muenchj4 the above script starting with #ex create masks is not included in this repository.

[Muenchj4]

@gisma now I have included the script in the it is runned until the error message occurs.

[Muenchj4]

@gisma Have you allready chequed my scripts? Can you imagine why the extent function does not work in my script?

[Muenchj4]

@gisma Do you know why rasterized_vector has no content:

mapview(rasterized_vector)
Error in res[i] <- readBin(x@file@con, what = dtype, n = 1, size = dsize, :
replacement has length zero
In addition: Warning message:
In rasterCheckSize(x, maxpixels = maxpixels) :
maximum number of pixels for Raster* viewing is 5e+05 ;
the supplied Raster* has 2e+08
... decreasing Raster* resolution to 5e+05 pixels
to view full resolution set 'maxpixels = 2e+08 '

[konstantinengelmayer]

png::writePNG(
        terra::as.array(subs),
        target = file.path(path, paste0(targetname, i, ".png"))
      )

Die Funktion in dem lapply- Loop am Ende der subset_ds Funktion erstellt Falschfarbenkomposite der DOP´s. Ich hab nicht wirklich heraus gefunden warum, aber einen funktionierenden Workaround gefunden.

if(nlyr(subs) > 1){
      png(file.path(path, paste0(targetname, i, ".png")), height=nrow(subs), width=ncol(subs)) 
      plotRGB(subs)
      dev.off()
    }else{
      png::writePNG(
        terra::as.array(subs),
        target = file.path(path, paste0(targetname, i, ".png"))
      )
    }

Wenn das Raster subs mehr als einen Kanal besitzt, werden die PNG´s mit dev.off() gespeichert, wenn nicht werden die Masken wie gehabt gespeichert.