The innovative use of satellites to predict famine

This is really interesting achievement of Gabriel Senay. Besides to his breakthrough researches and studies, he is one of Ethiopian hydrologist tries to apply his research for practical problems. I do't have much exposure, except few, to his researches and studies,  but I would like to know more.   Anyway I have seen this long description by SMITHSONIAN MAGAZINE  about him and his researches in Ethiopia at this magazine. The innovative use of satellites to predict famine

Gabriel Senay in his office (source: SMITHSONIAN MAGAZINE )

Very interesting paper on basin water yields

The very idea of budyko curve and other long term water balance approach is to characterise the basin water yeild capacity, or, the capacity of the basin to generate the discharge at the outlet, for a given precipitation. In other words, how the basin is parttioning the incoming inputs, rainfall, into different outflow (i.e ET and Q). In most studies, the formulations, the main driver that is asumed and explicitly accounted, is the basin energy-budget, in the form of PET. For me, though I didn't review literatures very well, it seems that the mere description and formulation of precipitaion, P, and potentail evapotranspiration, PET, would not explian the dischrage, Q, specially in the so-called a single hydrological year and so. I find that this (Zhou et al)\( ^{1} \) paper published in NATURE COMMUNICATIONS, and, is very promising with easy analytical formulation.
From the simplified water balance equation \[ P=E+R+ \Delta S \] where P is rainfall, E is evapotranspiration, and R runoff and \( \Delta S \) change in storage differences for a given hydrological year. Assuming the \( \Delta S = 0 \) for a given hydrological year, or even better for longer hydrological years, after re-arranging the equation will be:
\[ \frac{R}{P} = 1-\frac{E}{P} \]
From this point, there are some studies on how to estimate the \( E \) from the \( P \) and \( PET \) (e.g. Zeng and Cai, 2015\( ^{3} \) and papers cited therein), defines \( E \) as function of aridity index (\( PET/P \)), \[ E= P(1+ \frac{PET}{P} - (1+(\frac{PET}{P})w)^{\frac{1}{w}})) \]
This \( E \) formulation is substituted, and the reciprocal of the aridity index (\( P/PET \)), wetnes index, is used for analytical simplicity , i.e:
\[ \frac{R}{P}=(1+ (\frac{P}{PET})^{-m})^{\frac{1}{m}}-(\frac{P}{PET})^{-1} \]
which relates annual water yield ( R ) to a wetness index (precipitation/ potential evapotranspiration; P/PET) and watershed characteristics (m). m could be connected to many watershed characterstics, such as soil, vegetation, basin area, geomorphometry.

There is also an intersting efforts by Voepel et'al\( ^{2} \) on the controls of hydrologic partitioning at the catchment scale. One important thing in thier approach is the way they concptualize basin wetting and aridity in relation to the Normalized Difference Vegetation Index (NDVI). This would make the job easy, because it is easy to use NDVI from satellite.
Anyways, what interest me is \( m \) can be assessed with wide ranges of basin characterstics, and, further refine the equation with the dominante physical paramater(s).
References

1. \( ^{1} \) Guoyi Zhou, Xiaohua Wei, Xiuzhi Chen1, Ping Zhou, Xiaodong Liu, Yin Xiao, Ge Sun, David F. Scott, Shuyidan Zhou, Liusheng Han & Yongxian Su: Global pattern for the effect of climate and land cover on water yield.“ Nature Communications 6, Article number: 5918 DOI: 10.1038/ncomms6918
2. \( ^{2} \)Voepel, H., B. Ruddell, R. Schumer, P. A. Troch, P. D. Brooks, A. Neal, M. Durcik, and M. Sivapalan (2011), Quantifying the role of climate and landscape characteristics on hydrologic partitioning and vegetation response, Water Resour. Res., 47, W00J09, doi:10.1029/2010WR009944.
3. \( ^{3} \)Ruijie Zeng, Ximing Cai. (2015) Assessing the temporal variance of evapotranspiration considering climate and catchment storage factors. Advances in Water Resources 79, 51-60.
4. Gerrits, A. M. J., H. H. G. Savenije, E. J. M. Veling, and L. Pfister (2009), Analytical derivation of the Budyko curve based on rainfall characteristics and a simple evaporation model, Water Resour. Res., 45, W04403, doi:10.1029/2008WR007308

consultant and business idea

Apart from the research and academia world, if I would like to work (establish) consultant and business  out of my professional life, at least for the sake of supporting my research career, I would go for something like  eWaterCycle project. Since I am working on the water balance at smallest subbasin scale, my Professor brought this project to my attention. From their website,  eWaterCycle   " is a project that will provide detailed hydrological information for water management challenges around the globe. We will calculate how much water is available in each part of the world at present and in the near future". 

The ultimate objective of hydrology as a subject and hydrologist as a profession is to understand the hydrological processes to estimate the available water at different scales.  Hence, it feels for me that organising ones work for the sake of this objective will always have practical and professional career.  

Remote sensing rainfall products: part 4:TAMSAT

This is the fourth rainfall products i am looking at for hydrological research. TAMSAT stands for Tropical Applications of Meteorology using SATellite data and ground-based observations. TAMSAT is rainfall products designed for africa, calibrated with ground measurments. Detal description about it can be found here. While the dekadal and monthly estimation is avaliable for the whole of africa with 1km resolution, since 1983, the daily estimation is only avalaible 2013 onwards. Here I would like to use this products, and compare with other estimates.As usual, i downloaded the data in my harddisk, and process using R.

loading packages

library(raster)
## Loading required package: sp
library(ncdf4)
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
library(lattice)
library(rasterVis)
## Loading required package: latticeExtra
## Loading required package: RColorBrewer
## Loading required package: hexbin
library(ncdf)

Next, let’s load the raster into R.

setwd("/Users/administrator/Documents/PHDResearch/UBN_rainfall/TAMSAT/Project/")
tamsatlist<-list.files(pattern ='.nc', full.names = TRUE)

Next, let’s create a raster stack all two years of daily raster file.

#create raster stack
tamsat.stack <- stack(tamsatlist)
tamsat.stack
## class       : RasterStack 
## dimensions  : 1974, 1894, 3738756, 730  (nrow, ncol, ncell, nlayers)
## resolution  : 0.0375, 0.0375  (x, y)
## extent      : -19.03125, 51.99375, -35.98125, 38.04375  (xmin, xmax, ymin, ymax)
## coord. ref. : +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0 
## names       : Rain.Fall.Estimate.1, Rain.Fall.Estimate.2, Rain.Fall.Estimate.3, Rain.Fall.Estimate.4, Rain.Fall.Estimate.5, Rain.Fall.Estimate.6, Rain.Fall.Estimate.7, Rain.Fall.Estimate.8, Rain.Fall.Estimate.9, Rain.Fall.Estimate.10, Rain.Fall.Estimate.11, Rain.Fall.Estimate.12, Rain.Fall.Estimate.13, Rain.Fall.Estimate.14, Rain.Fall.Estimate.15, ...

These are time series raster data for the whole africa region, however, i would like to extract at some stations in my study area. let me do just for a single station. First I need to create spatialpoint object, then extract the time series layer at this location.

ADET<-cbind(37.47,11.27); ADET<-SpatialPoints(ADET) 

TAMSAT_station<-data.frame(
time=seq(as.Date('2013-01-01'), as.Date('2014-12-31'), 'day'),
ADET=as.vector(extract(tamsat.stack, ADET))
)

the time series rainfall at ADET

xyplot(ADET~time|equal.count(as.numeric(time), 1, overlap = 0.1),
TAMSAT_station,  
type = c("l","g"),  lw=2, strip = FALSE,
xlab="time", ylab="TAMSAT rainfall[mm]",
scales = list(x = list(relation = "sliced", axs = "i"),
              y = list(alternating = FALSE)))