AWESOME_validation.Rmd

---
title: "SoilErosionDB"
author: "Jinshi"
date: "8/12/2020"
output: html_document
---

# Load package
```{r load package, message=FALSE, include=FALSE, echo=FALSE}
# https://stackoverflow.com/questions/4090169/elegant-way-to-check-for-missing-packages-and-install-them/19873732

package_list <- c("cowplot","data.table","dplyr","ggplot2", "lubridate", "leaflet"
                  ,"kableExtra","knitr","ggmap","maps","mapdata","tidyr","sp","ggpubr"
                  ,"readxl")
package_new <- package_list[!(package_list %in% installed.packages()[,"Package"])]
if(length(package_new)) install.packages(package_new)

library(cowplot)
library(data.table)
library(dplyr)
library(ggplot2)
theme_set(theme_bw())
library(lubridate)
library(kableExtra)
library(knitr)
library("ggpubr")
library(tidyr)
library("leaflet")
library(sp)
library(readxl)
```


```{r functions}
country_lat_long_check <- function(sdata){
  ggplot(data = counties) + 
    geom_polygon(aes(x = long, y = lat, group = group),
                 color = "gray", fill = 'white', alpha = 0.5, size = 0.25) + 
    guides(fill="none") +
    theme(panel.grid.major = element_blank(),
          panel.grid.minor = element_blank()) +
    geom_rect(data = sdata,
              mapping=aes(xmin=Lon_min, xmax=Lon_max, ymin=Lat_min, ymax=Lat_max),
              color="red", alpha = 0.5) +
    labs(title=sdata$Country,
         x ="Latitude", y = "Longitude") ->
    p_check
  print(p_check)}


qc_background <- function (sdata) { 
  ggplot(sub_data, aes(x=Value)) + 
    geom_histogram(color="black", fill="gray", bins = 30) + 
    theme_bw() +
    xlab (var_num_col[i]) ->
    p_hist
  print(p_hist)
}
```

# Setup
```{r preliminaries, message=FALSE, include=FALSE, echo=FALSE, cache=TRUE}
# Set chunks defaults, these options will be applied to all subsequent chunks
knitr::opts_chunk$set(message = TRUE, include = TRUE, echo = FALSE,
                      fig.height = 4, fig.width = 8)

```

# load data
```{r load data}
wos_summary <- drake::readd(wos_summary)
# SEDB_del <- drake::readd(SEDB_del)
# SEDB_del <- read.csv('data/SoilErosionDB_v2.csv')
counties <- drake::readd(counties)
GlobalMATMAP <- drake::readd(GlobalMATMAP)
IGBP_MODIS <- drake::readd(IGBP_MODIS)

## join and get climate, vegetation information
SEDB_del %>% mutate(Latitude2 = round(Latitude*2)/2,
                 Longitude2 = round(Longitude*2)/2,
                 Lat_dif = ifelse(Latitude2 - Latitude >=0, 0.25, -0.25),
                 Lon_dif = ifelse(Longitude2 - Longitude >=0, 0.25, -0.25),
                 Latitude2 = Latitude2 - Lat_dif,
                 Longitude2 = Longitude2 - Lon_dif) %>% 
  dplyr::select(-Lat_dif, -Lon_dif) -> 
  SEDB_del

# Get Ecosystem class, MAT and MAP for srdb data
left_join(SEDB_del, IGBP_MODIS, by=c("Latitude2"="Latitude", "Longitude2"="Longitude")) ->
  SEDB_del

# Get MAT and MAP
left_join(SEDB_del, GlobalMATMAP, by=c("Latitude2"="Latitude", "Longitude2"="Longitude")) ->
  SEDB_del

SEDB_del %>% 
  dplyr::select(Study_midyear, Latitude, Longitude, MAT, MAP, Study_temp, Study_precip) 
# Change Latitude and Longitude to 0.5 resolution for IGBP

```


```{r check potential latitude and longitude input error}
varCountry <- sort(unique(SEDB_del$Country)) # data from XXX countries, all data has country information
sum_country <- data.frame()
for (i in seq(varCountry)) {
  subdata <- SEDB_del %>% filter(Country == varCountry[i] & !is.na(Country))
  lat_try <-  subdata %>% dplyr::select(Latitude) %>% na.omit()
  lon_try <- subdata %>% dplyr::select(Longitude) %>% na.omit() 
  
  min_lat <- ifelse(nrow(lat_try) == 0, NA, min(lat_try))
  max_lat <- ifelse(nrow(lat_try) == 0, NA, max(lat_try))
  min_lon <- ifelse(nrow(lon_try) == 0, NA, min(lon_try))
  max_lon <- ifelse(nrow(lon_try) == 0, NA, max(lon_try))
  
  Country_name <- varCountry[i]
    
  n_obs <- nrow(subdata) # number of observations in this country

  output <- bind_cols(i, Country_name, min_lat, max_lat, min_lon, max_lon, n_obs)
  sum_country <- bind_rows(sum_country, output)
  
  print(paste0("-----", i))
}

colnames(sum_country) <- c("ID", "Country", "Lat_min", "Lat_max", "Lon_min", "Lon_max", "obs")

sum_country  # view and check potential latitude and longitude input error

for(i in seq(nrow(sum_country))) {
  sdata = sum_country[i,]
  country_lat_long_check(sdata)
  print(paste0("*****", i))
}

```



```{r histgram of numeric records information}
colnames(SEDB_del)
var_num_col <- colnames(SEDB_del) 
# create a function to 

for(i in seq(var_num_col)){
  sub_data = SEDB_del[, which(colnames(SEDB_del) == var_num_col[i])] %>% as.data.frame()
  colnames(sub_data) = "Value"
  sub_data %>% na.omit() -> sub_data
  if(var_num_col[i] %in% c("Unique_ID", "barren_yn", "Study_number")) {next}
  else if(is.numeric(sub_data$Value)) {qc_background (SEDB_del)} 
  else {next}
  print(i)
}

```


```{r}
sdata
SEDB_del %>% count(Country)
SEDB_del %>% count(Latitude, Longitude)
```