Comparison of different satellite rainfall data

We have done a comparison study on different satellite rainfall data in Upper Blue Nile, you can find it here. Here narrowing the choice of products to three, based on their performances,  the long term (2002-2012) mean daily  spatial distribution, is depicted in figure 1.

fig1: the long term mean daily rainfall data in UBN basin

And, similarly, the long term mean annual comparison is shown in figure 2.

fig 2: The long term annual mean rainfall of UBN basin. 

Note on spatial pattern of the rainfall and the difference between these products. I will come back another time for detail discussion on this.

satellite rainfall estimation products in Upper Blue Nile Basin

This is stub for the complimentary material for our paper: Comparative evaluation of  satellite rainfall estimation products and bias correction in Upper Blue Nile Basin. I will provide the data, R scripts and some supplementary results soon....

• The Satellite rainfall dataset 
• The ground-gauge dataset 
• The DEM 
• R scripts 

All will be filled soon


Table: Bias correction function for different SREs in UBN basin

Remote sensing rainfall products: part two: SM2RAIN

SM2RAIN is satellite rainfall etimation algorithm developed by Luca Brocca and his research team. The idea is that it is possible to invert the soil moisture imagary to reanalyse the rainfall. References about it can be found here, here, here and references therein. I am using this rainfall product to estimate spatial rainfall over upper Ble Nile basin. Lets have qiuck look to the data.
As usual necessary library

library(raster)
## Loading required package: sp
library(rasterVis)
## Loading required package: lattice
## Loading required package: latticeExtra
## Loading required package: RColorBrewer
## Loading required package: hexbin
library(raster)
library(rgdal)
## rgdal: version: 0.9-1, (SVN revision 518)
## Geospatial Data Abstraction Library extensions to R successfully loaded
## Loaded GDAL runtime: GDAL 1.9.2, released 2012/10/08
## Path to GDAL shared files: /Library/Frameworks/R.framework/Versions/3.1/Resources/library/rgdal/gdal
## Loaded PROJ.4 runtime: Rel. 4.8.0, 6 March 2012, [PJ_VERSION: 480]
## Path to PROJ.4 shared files: /Library/Frameworks/R.framework/Versions/3.1/Resources/library/rgdal/proj
install.packages("R.matlab")
## Error in contrib.url(repos, "source"): trying to use CRAN without setting a mirror
library(R.matlab)
## R.matlab v3.2.0 (2015-02-24) successfully loaded. See ?R.matlab for help.
## 
## Attaching package: 'R.matlab'
## 
## The following objects are masked from 'package:base':
## 
##     getOption, isOpen
library(lattice)

Data import: I download the dataset, which was originally in matlab format, to my pc and import it to R

SM2RF<-readMat("/~/P_SM2RAIN_ECVSM_BluNile_DISTR.mat")
proj <- CRS('+proj=longlat +ellps=WGS84')
SM2RF2 <- cbind(SM2RF$lonubn, SM2RF$latubn, t(SM2RF$Psim)) 
SM2RF3 <- rasterFromXYZ(SM2RF2)

The dataset is raster layer for 8766 days. To indix each layer by the time-steps, we can use time index as follows:

Timeseries <- seq(as.Date('1990-01-01'), as.Date('2013-12-31'), 'day')
SM2RF4 <- setZ(SM2RF3, Timeseries)

To extract point data based on location of ground measurments, we can create spatial points and extract the data from the layer using these spatial points
To create the spatial points of measurment stations:

ADET<-cbind(37.47,11.27); ADET<-SpatialPoints(ADET) 
BAHRDAR<-cbind(37.42,11.6); BAHRDAR<-SpatialPoints(BAHRDAR) 
CHAGNI<-cbind(36.00,10.95); CHAGNI<-SpatialPoints(CHAGNI) 
DANGILA<-cbind(36.42,11.12); DANGILA<-SpatialPoints(DANGILA) 
DEBREMARKOS<-cbind(37.67,10.33);DEBREMARKOS<-SpatialPoints(DEBREMARKOS) 
MOTTA<-cbind(37.87,11.08); MOTTA<-SpatialPoints(MOTTA) 
PAWE<-cbind(36.05,11.15);  PAWE<-SpatialPoints(PAWE) 
AYKEL<-cbind(37.05,12.53); AYKEL<-SpatialPoints(AYKEL) 
GONDER<-cbind(37.42,12.55); GONDER<-SpatialPoints(GONDER)
DEBREBRIHAN<-cbind(39.58,9.63);DEBREBRIHAN<-SpatialPoints(DEBREBRIHAN)

To extract the time series rainfall (1990-2013 daily rainfall) and create the dataframe

SM2RF_station<-data.frame(
time=seq(as.Date('1990-01-01'), as.Date('2013-12-31'), 'day'),
ADET=as.vector(extract(SM2RF3, ADET)),
BAHRDAR=as.vector(extract(SM2RF3, BAHRDAR)),
CHAGNI=as.vector(extract(SM2RF3, CHAGNI)),
DANGILA=as.vector(extract(SM2RF3, DANGILA)),
DEBREMARKOS=as.vector(extract(SM2RF3,DEBREMARKOS)),
MOTTA=as.vector(extract(SM2RF3, MOTTA)),
PAWE=as.vector(extract(SM2RF3, PAWE)),
AYKEL=as.vector(extract(SM2RF3, AYKEL)),
GONDER=as.vector(extract(SM2RF3, GONDER)),
DEBREBRIHAN=as.vector(extract(SM2RF3, DEBREBRIHAN)))

Visulisation: to visualize sample of time series and stations,subset of the datset is used. Here is the first 11 years, 1990-2000, from top to down, each plot for one year.

SM2RF_station2<-subset(SM2RF_station[0:4018, ])
xyplot(ADET+PAWE+MOTTA+GONDER+CHAGNI+DEBREMARKOS~time|equal.count(as.numeric(time), 11, overlap = 0.1),
SM2RF_station2,  
type = c("a","g"),   , strip = FALSE,
xlab="time", ylab="SM2RAIN rainfall",
as.table = TRUE, layout = c(1, 11),
auto.key = list(points = FALSE, lines = TRUE, columns = 3),
scales = list(x = list(relation = "sliced", axs = "i"),
              y = list(alternating = FALSE)))