In rhabel/modLWFB90: R-package for applying LWF-Brook90 directly from FVA-BW servers

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

This vignette shall serve as an introduction to the modLWFB90 package. It shows how to use its essential functions and how to specify certain aspects if desired.

---

Introduction

The hydrological model LWFBrook90 is increasingly used at the FVA-BW in various projects and applications. It is a dynamic 1-D model that, through a large number of settings, allows for a detailed representation of soil, climate, and vegetation conditions. Hence, it is pretty useful for soil water modeling in the forest background we're working in.

Paul Schmidt-Walter at the NW-FVA developed an R-package that serves as an R-wrapper for LWFBrook90.exe. LWFBrook90R makes it possible and relatively easy to use and work with LWFBrook90 from within Rstudio.

The package in front of you, modLWFB90, even goes one step further: It uses the whole functionality of LWFBrook90R, but adds a link to soil data and climate data available on the FVA-BW servers. This way users can do hydrological modelling by just giving the desired methods as arguments. All you need is one tidy data frame with the points you want to model in the following form:

• ID_custom - a unique ID that all intermediate products as well as the results will be assigned to
• easting and
• northing - coordinates in UTM 32N EPSG:32632

Then, all you need to specify is the time period, a number of settings (the origin of the soil data, the PTF to be used, the type of forest you want to model, etc.), and the way you want the output to be structured.

---

Available Settings

Vegetation

In the course of the project WHH-KW, parametrisation settings for the main tree species (beech - Buche, spruce - Fichte, oak - Eiche, pine - Kiefer) were agreed upon and can be selected in modLWFB90.

Climate data

For WHH-KW the UHH Hamburg created a new and improved climate data set containing regionalised 250m-raster climate data for the historic time period 1961-2018, as well as the following climate scenarios:

Soil data

The following soil options can be selected:

• STOK - Data from the Standortskartierung (add description...)
• BZE - regionlised BZE-II data (add description...)
• OWN - you can provide your own soil data

There's also a function to compare the three soil data outputs.

---

Basic Usage

Getting the packages up and running

First of all, get release 0.4.5 of LWFBrook90R. Newer versions have changes that are incompatible with modLWFB90. github:pschmidtwalter/LWFBrook90R Then, install modLWFB90, together with the dependency packages, which are required and will be installed automatically. The easiest way to install from github is with the following line of code:

if(!require("remotes")) install.packages("remotes")
# HTTPS not supported when installing from github:  
Sys.setenv(R_REMOTES_NO_ERRORS_FROM_WARNINGS=TRUE)
Sys.setenv(HTTPS_PROXY=Sys.getenv("HTTP_PROXY")) 

remotes::install_github(repo="pschmidtwalter/LWFBrook90R@v0.4.5") 
remotes::install_github(repo="rhabel/modLWFB90")

library(modLWFB90)

As https is not supported from FVA-servers, this first has to be overwritten You might encounter issues with updates that can't be overwritten and other R-package-problems, in which case the following function might help you install the missing packages quicker (no install.packages(x); library(x) over and over again):

check_packages <- function(pkg){
  new.pkg <- pkg[!(pkg %in% installed.packages()[, "Package"])]
  if (length(new.pkg))
    install.packages(new.pkg, dependencies = TRUE)
  sapply(pkg, require, character.only = TRUE)

}

# example: 
check_packages(c("LWFBrook90R", "parallel"))

Input opjects

Essential for hydrological modelling are three datasets: Soil, Climate, and Parameters. This is how to get them the simplest possible way. First, we need sample data. test.ids.bds includes five points in the forest near Lake Constance (Bodensee).

test.ids.bds

Soil data

The function to create soil data is called fnc_get_soil and returns a list of soil data frames, so best assign it to a list object like this:

ls.soil <- fnc_get_soil(df.ids = test.ids.bds, 
                        soil_option = "STOK", 
                        PTF_to_use = "WESSOLEK", 
                        limit_MvG = T)
ls.soil[[1]]

There's a number of options to choose from:

• df.ids - your data frame with ID_custom, easting and northing
• soil_option - either STOK, BZE, (STOK_BZE for or OWN, if own you need to specify own soil data through df.soils.
• PTF_to_use - either HYPRES, PTFPUH2 or WESSOLEK
• for further settings (e.g. max soil depth or roots distribution) see ?fnc_get_soil

---

There's also the option to compare the soil data created by the different options. Plus, the option to compare the hydraulic parameters created by the differnt PTFs. See ?fnc_compare_soil for more information.

---

Climate data

The function to create climate data is called fnc_get_climate and works the same way:

ls.clim <- fnc_get_clim(df.ids = test.ids.bds, 
                        mindate = as.Date("2010-01-01"), 
                        maxdate = as.Date("2011-12-31"))

Here you can only specify the time period for which you need climate data. The additional arguments path_std and path_climdb lead to the current position of the climate databases on the FVA-Servers. They may have to be adjusted at some point, but for now they are up to date and shouldn't be changed.

If you model many points, large climate data files might use up all your RAM. In this case you should have a look at fnc_read_climdb and fnc_write_climdb.

---

Parameters

The list of parameters you can pass to LWFBrook90 is pretty exhaustive. See LWFBrook90R::setparam_LWFB90 for a complete list. In the project WHH-KW, a set of parameters was agreed upon for the most relevant tree species (Buche - beech, Eiche - oak, Fichte - spruce, Kiefer - pine). They can be used by specifying the argument tree_species either as one value for all points, or as a vector of values for each individual point. Additional information on the modelling points can be passed with the argument df.ind.info. See ?fnc_get_params to find out how to do this.

The default setting for spruce would look like this:

ls.parms <- fnc_get_params(df.ids = test.ids.bds, 
                           tree_species = "spruce")

---

Options

Modelling options are set with the function set_optionsLWFB90 from the LWFBrook90R package. If you use modLWFB90 the only settings you need to adjust are

• budburst.method and leaffall.method - if you want something other than fixed
• startdate and enddate - as desired, but should match your settings in fnc_get_clim
• root.method

check ?set_optionsLWFB90 for more information.

---

Adjust Output

Finally, we need to define how our output should look like. LWFBrook90 creates a lot of output files. In order to keep data storage to a minimum, both run_LWFB90 and run_multisite_LWFB90 provide an output_fun - argument that can be used to reduce the output and directly write it. This is what the functions fnc_write (daily values) and fnc_write_agg (options to aggregate over year, vegetation period, or months) function are made for. They write temporary files with the desired output for each point. fnc_write_to_sql can be used to convert these files into a SQLite-DB. This "step-in-between" is necessary because SQLite does not support parallel writing. If you model few enough points that your memory can handle them, you can set dir_name to NA and get the results loaded into your environment. See the help pages of the write-functions on how to do this.

---

Example

# create ID-dataframe
df.ids = test.ids.bds

# set options
options.b90 <- set_optionsLWFB90(budburst.method = "Menzel",
                                 leaffall.method = "vonWilpert",
                                 startdate = as.Date("2010-01-01"),
                                 enddate = as.Date("2011-12-31"),
                                 root.method = "betamodel")
# set df.output
df.output <- set_outputLWFB90()
df.output[,] <- 0L
df.output[c("Evap", "Swat", "Flow"), c("Day")] <- 1
df.output[c("Flow"), c("Mon")] <- 1

output_path <- paste0(getwd(), "/")


# create data
ls.parms <- fnc_get_params(df.ids = test.ids.bds,
                           tree_species = "spruce")
ls.clim <- fnc_get_clim(df.ids = test.ids.bds)
ls.soil <- fnc_get_soil(df.ids = test.ids.bds,
                           soil_option = "STOK",
                           testgebiet = "BDS",
                           PTF_to_use = "WESSOLEK")

# this chunk of code removes non-mapped STOK-points, if there are any and 
# the BZE-filling option was not selected. Otherwise it doesn't do anything.
ls.clim <- ls.clim[which(!unlist(lapply(ls.soil, is.null))==T)]
ls.parms <- ls.parms[which(!unlist(lapply(ls.soil, is.null))==T)]
ls.soil <- ls.soil[which(!unlist(lapply(ls.soil, is.null))==T)]


# Run: 
  res <- run_multisite_LWFB90(options_b90 = options.b90,
                              param_b90 = ls.parms,
                              soil = ls.soil,

                              climate = ls.clim,

                              output = df.output,

                              all_combinations = F,
                              cores = parallel::detectCores(),
                              # if you have more than one core available, this will speed things up significantly

                              output_fun = fnc_write_agg,
                              aggr_tp = "vegper",
                              dir_name=paste0(output_path, "tmp/"))

  # save soils 
  names(ls.soil) <- NULL
  soils <- bind_rows(ls.soil)

  # create dirs
  if(!dir.exists(paste0(output_path, "tmp/soils/"))){
    dir.create(paste0(output_path, "tmp/soils/"), recursive = T)
  }
  save(soils, file = paste0(output_path, "tmp/soils/soils.RData"))

  # write to db --------------- ####
  fnc_write_to_sql(db_name = paste0(output_path, "test_db.sqlite"),
                   dir_tmp = paste0(output_path, "tmp"))

rhabel/modLWFB90 documentation built on Nov. 21, 2024, 3:28 a.m.