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

#

# CreateImageDiffPlots

# 

# R script generates collection of nine plots that display the distributions of  populations of 

# three elevation pixel pairs in proximity to the border between ASTER and STRM data in mosaic 

# images# under development for the Environment and Organisms project.

# 

#

# Inputs: 

#   1) Comma Separated Value (CSV) table containing randomly-sampled elevation value pairs,

#      extracted from ASTER/CGIAR mosaic and CDEM images, created by the R script: makeImagePairTable.r

#      Note: At present, be sure that this file has been sorted by column 1 (ColumnID) in Excel

#      before using in this program.

#

#   2) sampling factor: integer (range=1 to number of recors in input table. Determines

#      the size of randomly-selected sampe of total table record 'triplets' (north, border,

#      and south) rows of each column sample) to be displayed in the plots: 

#           sampling factor = 1  : plot every table record

#                             10 : plot a random sample containing 1/10th of records

#                            100 : plot random sample containing 1/100th of records.

# 

# To run:

#

#  1) place this file and the input file "tableForMark4000_5_8_SortColID.csv" in folder

#  2) start R

#  3) > source("CreateImageDiffPlotsOverlay.r")

#  4) > CreateImageDiffPlots(sampling factor)

#     < press <cr> key to view plots sequentially

#  

# TODO: add symbol legend to each plot.

# Author: Rick Reeves, NCEAS

# May 14, 2011

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

CreateImageDiffPlotsNew <- function(plotSampFact = 100)

{

# Check plotSampleFact range  

   if (plotSampFact < 1)

      plotSampFact = 1

   if (plotSampFact > 100)

      plotSampFact = 100

# Read input table that was sorted OFFLINE in Excel on the first column (ColumnID)

   pointTable <-read.csv("tableForMark4000_5_8_SortColID.csv")

# Table created by randomly sampling from two superimposed 

# image: 

#  1) DOM mosaic image comprised of ASTER and SRTM components.

#  2) 'Baseline' Canadian DEM (CDEM) image.

# 

# The input table contains three rows for each randomly-selected

# pixel pair from both images. Each row contains two pixel pairs,

# the first pair drawn from the image mosaic, the second pair

# drawn from the CDEM image: 

#   First pair: North pixel, South pixel (ASTER/SRTM mosaic)

#   Second pair: North pixel, South pixel (CDEM)

#

#  The first row of each 'triplet' contains pixel pairs North of border,

#  The second row contains pixel pairs spanning  border,

#  The third row contains pixel pairs South of border,

#

# This script generates a series of plots that display

# differences between:

#    1) The mosaic and CDEM images

#    2) Image pixels on and away from the ASTER / SRTM boundary.

# 

   northRowIndexes = seq(from=1, to=(nrow(pointTable) - 3),by=3)

   borderRowIndexes = seq(from=2, to=(nrow(pointTable) - 1),by=3)   

   southRowIndexes = seq(from=3, to=(nrow(pointTable)),by=3)

#

# calculate and append the difference between elevations

# and CDEM

# these lines create the inputs for differnce image plots for each of three

# pixel pair subsets: North of border (All Aster), border (combo Aster/Srtm),

#                     South of border (all Srtm)

# 

# First, add the 'difference columns' to the entire table

#

   pointTable <-cbind(pointTable,(pointTable$elevNorth - pointTable$elevSouth))

   pointTable <-cbind(pointTable,(pointTable$cdemNorth - pointTable$cdemSouth))

   pointTable <-cbind(pointTable,(pointTable$elevNorth - pointTable$cdemNorth))

   pointTable <-cbind(pointTable,(pointTable$elevSouth - pointTable$cdemSouth))

#   

   colnames(pointTable)[6] <- "diffMosaicNorthSouth"

   colnames(pointTable)[7] <- "diffCDEMNorthSouth"

   colnames(pointTable)[8] <- "diffNorthMosaicCDEM"

   colnames(pointTable)[9] <- "diffSouthMosaicCDEM"   

# add a placeholder for the 'boundary' value, across each table

#  pointTable <-cbind(pointTable,(-1))

#  colnames(pointTable)[10] <- "deltaAcrossBorder"

# Difference between Mosaic (ASTER or CGIAR or border) 

# and CDEM elevation as pertentage of the mosaic elevation

   pointTable <-cbind(pointTable,(pointTable$diffNorthMosaicCDEM/pointTable$elevNorth * 100)) 

   pointTable <-cbind(pointTable,(pointTable$diffSouthMosaicCDEM/pointTable$elevSouth * 100)) 

   colnames(pointTable)[10] <- "magDiffMosaicCDEMNorthPct"

   colnames(pointTable)[11] <- "magDiffMosaicCDEMSouthPct"   

# For the plots, subdivide the table into three segments: 

#     rows north of border

#     rows crossing border

#     rows south of border

   northRowTblAll = pointTable[northRowIndexes,]

   borderRowTblAll = pointTable[borderRowIndexes,]

   southRowTblAll = pointTable[southRowIndexes,]

   subset <- 1:nrow(northRowTblAll)

   randSub <- sample(subset,as.integer(length(subset) / plotSampFact)) 

   northRowTbl <- northRowTblAll[randSub,]

   borderRowTbl <- borderRowTblAll[randSub,]

   southRowTbl <- southRowTblAll[randSub,]   

message("hit key to create each plot...")

browser()

# Three plotting characters used

   plotCh1 <- 17 # 'north' (aster) points

   plotCh2 <- 18 # 'border' (aster+srtm) points 

   plotCh3 <- 20 # 'south' (srtm) points

# NEW: Three plots: The difference between Mosaic and CDEM for pixels along

# each of three border edges: North (ASTER), Border, South (SRTM)

# for now, north, border, south have separate plots

# We need to tailor the plot 'triplet' X and Y axes to the dynamic ranges of all three data sets.

   par(mfrow=c(1,3)) # Create a three column multi-plot

# Create values for the 'delta across pixel boundaries' plot

   deltaBoundaryASTER <- northRowTbl$diffNorthMosaicCDEM - northRowTbl$diffSouthMosaicCDEM

   deltaBoundaryBorder <- borderRowTbl$diffNorthMosaicCDEM - borderRowTbl$diffSouthMosaicCDEM  

   deltaBoundarySRTM <- southRowTbl$diffNorthMosaicCDEM - southRowTbl$diffSouthMosaicCDEM   

   commonXAxis <- c(0,max(pointTable$ColumnID))   

   commonYAxis <- range(c(deltaBoundarySRTM,deltaBoundaryBorder,deltaBoundaryASTER),na.rm=TRUE)   

   xAxisLbl <- sprintf("M/SD: ASTER: %.2f / %.2f",mean(deltaBoundaryASTER,na.rm=TRUE),sd(deltaBoundaryASTER,na.rm=TRUE))

   plot(northRowTbl$ColumnID,deltaBoundaryASTER,xlim=commonXAxis, ylim=commonYAxis,main="Difference along Row Boundary: ASTER",xlab=xAxisLbl,col="red",pch=plotCh1)

   xAxisLbl <- sprintf("M/SD: BORDER: %.2f / %.2f",mean(deltaBoundaryBorder,na.rm=TRUE),sd(deltaBoundaryBorder,na.rm=TRUE))

   plot(northRowTbl$ColumnID,deltaBoundaryBorder,xlim=commonXAxis, ylim=commonYAxis,main="Difference along Row Boundary: Border",xlab=xAxisLbl,col="darkgreen",pch=plotCh2)

   xAxisLbl <- sprintf("M/SD: SRTM: %.2f / %.2f",mean(deltaBoundarySRTM,na.rm=TRUE),sd(deltaBoundarySRTM,na.rm=TRUE))

   plot(northRowTbl$ColumnID,deltaBoundarySRTM,main="Difference along Row Boundary: SRTM",xlim=commonXAxis, ylim=commonYAxis,xlab=xAxisLbl,col="blue",pch=plotCh3)

message("Three-Plot done")

browser()

dev.new()

   deltaBoundary <- southRowTbl$diffNorthMosaicCDEM - southRowTbl$diffSouthMosaicCDEM

   xAxisLbl <- sprintf("M/SD: SRTM: %.2f / %.2f",mean(deltaBoundary,na.rm=TRUE),sd(deltaBoundary,na.rm=TRUE))

   plot(northRowTbl$ColumnID,deltaBoundary,main="Difference along Row Boundary: Border",xlab=xAxisLbl,col="red",pch=plotCh1)

browser()

   xAxisLbl <- sprintf("M/SD: AST: %.2f / %.2f BRD: %.2f / %.2f STM: %.2f / %.2f",mean(northRowTbl$elevSouth,na.rm=TRUE),sd(northRowTbl$elevSouth,na.rm=TRUE),

                                                                            mean(borderRowTbl$elevSouth,na.rm=TRUE),sd(borderRowTbl$elevSouth,na.rm=TRUE),

                                                                            mean(southRowTbl$elevSouth,na.rm=TRUE),sd(southRowTbl$elevSouth,na.rm=TRUE))

   plot(northRowTbl$ColumnID,northRowTbl$elevSouth,main="South Mosaic Pixel Elev: ASTER (red) Border (green) SRTM (blue)",xlab=xAxisLbl,col="red",pch=plotCh1)

   points(borderRowTbl$ColumnID,borderRowTbl$elevSouth,col="darkgreen",pch=plotCh2)

   points(southRowTbl$ColumnID,southRowTbl$elevSouth,col="blue",pch=plotCh3)

browser()

dev.new()

# Second plot pair: North/South CDEM pixel elevation for North/South subsets 'distance East from Western edge'.

   xAxisLbl <- sprintf("M/SD: AST: %.2f / %.2f BRD: %.2f / %.2f STM: %.2f / %.2f",mean(northRowTbl$cdemNorth,na.rm=TRUE),sd(northRowTbl$cdemNorth,na.rm=TRUE),

                                                                            mean(borderRowTbl$cdemNorth,na.rm=TRUE),sd(borderRowTbl$cdemNorth,na.rm=TRUE),

                                                                            mean(southRowTbl$cdemNorth,na.rm=TRUE),sd(southRowTbl$cdemNorth,na.rm=TRUE))

   plot(northRowTbl$ColumnID,northRowTbl$cdemNorth,main="North CDEM Pixel Elev ASTER (r) Border (g) SRTM (b)",xlab=xAxisLbl,col="red",pch=plotCh1)

   points(borderRowTbl$ColumnID,borderRowTbl$cdemNorth,col="darkgreen",pch=plotCh2)

   points(southRowTbl$ColumnID,southRowTbl$cdemNorth,col="blue",pch=plotCh3)

browser()   

dev.new()

   xAxisLbl <- sprintf("M/SD: AST: %.2f / %.2f BRD: %.2f / %.2f STM: %.2f / %.2f",mean(northRowTbl$cdemSouth,na.rm=TRUE),sd(northRowTbl$cdemSouth,na.rm=TRUE),

                                                                            mean(borderRowTbl$cdemSouth,na.rm=TRUE),sd(borderRowTbl$cdemSouth,na.rm=TRUE),

                                                                            mean(southRowTbl$cdemSouth,na.rm=TRUE),sd(southRowTbl$cdemSouth,na.rm=TRUE))   

   plot(northRowTbl$ColumnID,northRowTbl$cdemSouth,main="South CDEM Pixel Elev ASTER (r) Border (g) SRTM (b)",xlab=xAxisLbl,col="red",pch=plotCh1)

   points(borderRowTbl$ColumnID,borderRowTbl$cdemSouth,col="darkgreen",pch=plotCh2)

   points(southRowTbl$ColumnID,southRowTbl$cdemSouth,col="blue",pch=plotCh3)

browser()

dev.new()

message("All plots created - hit key to delete them...")

browser()

graphics.off()

}