global.R

# # Installing and loading required packages (https://gist.github.com/stevenworthington/3178163)
# ipak <- function(pkg){
#   new.pkg <- pkg[!(pkg %in% installed.packages()[, "Package"])]
#   if (length(new.pkg))
#     install.packages(new.pkg, dependencies = TRUE)
#   sapply(pkg, library, character.only = TRUE)
# }
# 
# 
# packages <- c("shiny", "tidyverse", "sp", "plotly", "DT", "leaflet.extras", "leaflet")
# ipak(packages)
# # sp: For the point.in.polygon function

## Useful when installing packages on the linux server
# chooseCRANmirror(ind=89)  
# library('devtools', lib = "/usr/local/lib/R/site-library")  

#####################################################################

library(shiny)
library(tidyverse)
library(sp)
library(plotly)
library(DT)
library(leaflet)
library(leaflet.extras)

library(rmarkdown)

if (names(dev.cur()) != "null device") dev.off()
pdf(NULL)

## My modules: either load package or source modules from this directory. 
## Note that the package version of those app functions is not finished for operational use. Only for documnetation purposes
source('R/map_modules.R')
source('R/table_modules.R')
source('R/plot_modules.R')
source('R/plotting_functions.R')
source('R/mapping_functions.R')

hbv_catchments <- readLines("data/hbv_catchments.json") %>% paste(collapse = "\n")

# Load the Rdata files that were prepared with the NVEDATA package.
# This creates the global variable

load("data/update_time.RData")

load("data/HBV_2014.RData")
load("data/HBV_2016.RData")
load("data/DDD.RData")
load("data/ODM.RData")
load("data/flomtabell.RData")
load("data/HBV_past_year.RData")
load("data/meta_data.rda")

meta_data <- dplyr::filter(meta_data, br23_HBV == "Y" | br9_Flomvarsling == "Y")


station_numbers <- as.character(unique(HBV_2014$regine.main))  # All of the HBV_2016 and DD stations are in HBV_2014
station_names <- as.character(unique(HBV_2014$station.name))  # May not be optimal (if 2 stations had same name), but it works
station_nbname <- as.character(unique(HBV_2014$nbname))

## Metadata organized as below is needed for the maps.
# Maybe we can streamline with the rest later
station_indices <- which(meta_data$regine_main %in% station_numbers)  # 1 station in HBV_2014 is not in the metadata
stations <- lapply(meta_data, function(x) {x[station_indices]})
# stations$nbname_SPECIALCHAR <- paste(stations$regine_main, "-", stations$name, sep ="")
stations$nbname <- paste(stations$regine_main, "-", station_names[match(stations$regine_main, station_numbers)], sep ="")


OBS <- dplyr::filter(DDD, Type == "Runoff", Variable == "Obs")
OBS$time<- as.Date(OBS$time)
DDD <- dplyr::filter(DDD, Variable != "Obs")

# Calculation of a stations$flood_warning indicator for the forecast period
# I want to have it under the "stations" list for the moment as this list is used by the map functions
## WARNING: this first implementation has potential bugs and requires decisions on which variables o use!

HBV_2014_SimCorr <- dplyr::filter(HBV_2014, Type == "Runoff" & Variable == "HBV.UM.korr")
HBV_2014_SimCorr_maxed <- group_by(HBV_2014_SimCorr, nbname, regine.main) %>% dplyr::summarise(maxed = max(na.omit(Values)))

flom_obs_mean <- dplyr::filter(flomtabell, Type == "Obs" & Variable == "mean") 

index_HBV <- match(stations$regine_main, HBV_2014_SimCorr_maxed$regine.main)
index_flomtabell <- match(HBV_2014_SimCorr_maxed$regine.main, flom_obs_mean$regine.main)

stations$flood_warning <- HBV_2014_SimCorr_maxed$maxed[index_HBV] / flom_obs_mean$Values[index_flomtabell[index_HBV]]


HBV_2014_SimH90 <- dplyr::filter(HBV_2014, Type == "Runoff" & Variable == "Hi90")
HBV_2014_SimL90 <- dplyr::filter(HBV_2014, Type == "Runoff" & Variable == "Lo90")
HBV_2014_diff <- HBV_2014_SimH90
HBV_2014_diff$Values <- HBV_2014_SimH90$Values - HBV_2014_SimL90$Values

HBV_2014_diff_maxed <- group_by(HBV_2014_diff, nbname, regine.main) %>% dplyr::summarise(maxed = max(Values, na.rm = TRUE))
stations$uncertainty <- HBV_2014_diff_maxed$maxed[index_HBV] / flom_obs_mean$Values[index_flomtabell[index_HBV]]