What Four Decades of Earth Observation Tell us about Land Degradation in the Sahel

From: Mbow, C.; Brandt, M.; Ouedraogo, I.; de Leeuw, J.; Marshall, M. What Four Decades of Earth Observation Tell Us about Land Degradation in the Sahel? Remote Sens. 2015, 7, 4048-4067.

Land degradation mechanisms are related to two main categories, one related to climate change and one associated with local human impact, mostly land use change such as expansion of cultivation, agricultural intensification, overgrazing and overuse of woody vegetation. Land degradation characteristics, triggers and human influence are manifold and interrelated. Some of the indicators can be monitored using Earth Observation techniques (underlined in red):

1

During the last four decades, the Sahel was affected by below-normal precipitation with two severe drought periods in 1972–73 and in 1983–84. Because of this negative climate trend, many studies prioritized the Sahel “crisis” in terms of productivity loss and land degradation. These negative perceptions have been opposed with recent findings of improved greenness mostly in relation to recent improvement in rainfall.

cover1

The assessment of land degradation and quantifying its effects on land productivity have been both a scientific and political challenge. After four decades of Earth Observation applications, little agreement has been gained on the magnitude and direction of land degradation in the Sahel. The number of Earth Observation datasets and methods, biophysical and social drivers and the complexity of interactions make it difficult to apply aggregated Earth Observation indices for these non-linear processes. Hence, while many studies stress that the Sahel is greening, others indicate no trend or browning. The different generations of satellite sensors, the granularity of studies, the study period, the applied indices and the assumptions and/or computational methods impact these trends.

4

tab

While there is a clearly positive trend in biomass production at Sahel scale, a loss in biodiversity and locally encroaching barren land are observed at the same time. Multi-scale Earth Observation analyses show that neither the desertification nor the greening paradigms can be generalized, as both attempt to simplify a very complex reality. Heterogeneity is an issue of scale, and very coarse-scaled vegetation trend analyses reveal a greening Sahel. However, locally-scaled studies are not uniform, observing greening and degradation at the same time.

We suggest several improvements: (1) harmonize time-series data, (2) promote knowledge networks, (3) improve data-access, (4) fill data gaps, (5) agree on scales and assumptions, (6) set up a denser network of long-term fields-surveys and (7) consider local perceptions and social dynamics, as local people’s perception of land degradation/improvements often disagree with Earth Observation analyses.

Thus, to allow multiple perspectives and avoid erroneous interpretations caused by data quality/scale issues/generalizations, we recommend combining multiple data sources at multiple scales. Furthermore, we underline the relevance of field data and experience, and results achieved by remote sensing techniques should not be interpreted without contextual knowledge.

Download the full article here: Paper at MDPI

see also:

Knauer, K., Gessner, U., Dech, S., Kuenzer, C., 2014. Remote sensing of vegetation dynamics in West Africa. International Journal of Remote Sensing 35, 6357–6396. doi:10.1080/01431161.2014.954062

Renaming, converting, clipping: script based raster time series processing

When working with remotely sensed time series data (e.g. MODIS, AVHRR, GIMMS, etc.), bulk processing can save a lot of time. Using a terminal in a Linux environment, simple scripts can process thousends of files in short time. Here are some basic hints on how to start, gdal has to be installed. For Windows users or users who prefer an interface, QGIS provides a graphical interface for most of the gdal modules. Also renaming can be done more user friendly.

Renaming:

Original data-names can be very long annoying. So first we rename all files to numbers from 1 to XX (like 1, 2, 3…). To do so, navigate to the folder with the files, and these lines will rename all files originally starting with MOD (like MOD13Q1203837383922837284.hdf):

for fname in MOD* 
do
mv $fname `printf $x`
x=$(($x+1))
done

Then the following lines will create a script that renames all files numbered from 1 to 60. the final files will include year and date. So we then have data named MODIS_2001_01 until MODIS_2005_12. This can be adopted to every name and numbering. The script is then executed.

for ((i=1;$i<=60;i++)) do echo "mv $i MODIS_$(( ($i-1) / 12 + 2001 ))_$(( ($i-1) % 12 + 1 )).tif" ; done >rename.sh
sed -i 's/_1/_01/g' rename.sh
sed -i 's/_2/_02/g' rename.sh
sed -i 's/_3/_03/g' rename.sh
sed -i 's/_4/_04/g' rename.sh
sed -i 's/_5/_05/g' rename.sh
sed -i 's/_6/_06/g' rename.sh
sed -i 's/_7/_07/g' rename.sh
sed -i 's/_8/_08/g' rename.sh
sed -i 's/_9/_09/g' rename.sh
chmod 0755 rename.sh
./rename.sh

Convert:

This works fine to convert all kind of formats, also very useful to change HDF files (like MODIS, VGT…) to GeoTiff. This example converts GeoTiffs to ENVI format.

# define in and out folder and output format

in="/home/martin/SPOT_VGT/NDV/"
out="/home/martin/SPOT_VGT/NDV/converted/"
format="ENVI"

# convert all Tiffs to ENVI format, use float32 as output here

cd $in
for i in *.tif; do gdal_translate -of $format -ot Float32 $i $out`basename $i`; done

Clip:

# define in and out folder

in="/home/martin/SPOT_VGT/NDV/"
out="/home/martin/SPOT_VGT/NDV/clipped/"

# define the window which should be clipped

range="-17.6 16.25 -2.72 13.16"

# do the clipping for all Tiffs

cd $in 
for i in *.tif; do gdal_translate -projwin $range $i $out`basename $i`; done

For more information, see http://www.gdal.org/

Pixel-wise time series trend anaylsis with NDVI (GIMMS) and R

The GIMMS dataset is currently offline and the new GIMMS3g will soon be released, but it does not really matter which type of data is used for this purpose. It can also be SPOT VGT, MODIS or anything else as long as the temporal resolution is high and the time frame is long enough to detect significant trends. The purpose is to do a pixelwise trend analyis and extract only significant trends over a certain period of time fora selected region. Everything is done using open and free R software. Input data are continuous NDVI images, in this case it’s GIMMS with bi-monthly images, so 24 per year for the period 1982-2006.

so let’s load all Geotiffs lying in the GIMMS directory (should be  600 in this case):

library(raster)
setwd("~/GIMMS")
sg = stack(list.files(pattern='*.tif'))
gimms = brick(sg)
rm(sg)

first, let’s create annual sums, as autocorrelation might be present with monthly values. To keep NDVI as the unit, the results are devided by 24:

fun <- function(x) { 
 gimms.ts = ts(x, start=c(1982,1), end=c(2006,24), frequency=24)
 x <- aggregate(gimms.ts) 
 }
gimms.sum <- calc(gimms, fun)
gimms.sum=gimms.sum/24
plot(gimms.sum)

then the slope is calculated to get the direction and magnitude of trends, multiplied by the number of years to get the change in NDVI units:

time <- 1:nlayers(gimms.sum) 
fun=function(x) { if (is.na(x[1])){ NA } else { m = lm(x ~ time); summary(m)$coefficients[2] }}
gimms.slope=calc(gimms.sum, fun)
gimms.slope=gimms.slope*25
plot(gimms.slope)

now we need to see which trends are significant. Thus we first extract the p-value:

fun=function(x) { if (is.na(x[1])){ NA } else { m = lm(x ~ time); summary(m)$coefficients[8] }}
p <- calc(gimms.sum, fun=fun)
plot(p, main="p-Value")

then mask all values >0.05 to get a confidence level of 95%:

m = c(0, 0.05, 1, 0.05, 1, 0)
rclmat = matrix(m, ncol=3, byrow=TRUE)
p.mask = reclassify(p, rclmat)
fun=function(x) { x[x<1] <- NA; return(x)}
p.mask.NA = calc(p.mask, fun)

and finaly mask all insignificant values in the trend map, so we only get NDVI change significant at the 95% level:

trend.sig = mask(gimms.slope, p.mask.NA)
plot(trend.sig, main="significant NDVI change")

The result could look like that:

significant NDVI change (1982-2006) using integrated GIMMS

significant NDVI change (1982-2006) using integrated GIMMS

Simple time series analysis with GIMMS NDVI and R

GIMMS NDVI time series for a selected pixel

GIMMS NDVI time series for a selected pixel

Time series analysis with satellite derived greenness indexes (e.g. NDVI) is a powerfull tool to assess environmental processes. AVHRR, MODIS and SPOT VGT provide global and daily imagery. Creating some plots is a simple task, and here is a rough start how it is done with GIMMS NDVI. All we need is the free software R.

I assume we already have the NDVI images in a folder, so the first 3 steps can be skipped, if not, they can be downloaded with a simple command:

wget -r ftp://ftp.glcf.umd.edu/glcf/GIMMS/Albers/Africa/

now unzip it:

find . -name "*.gz" -exec unp {} \;

As it comes in Albers projection, I reproject the Tiffs to a Lat Long WGS84 projection:

for in in *tif; do gdalwarp -s_srs EPSG:9001 -t_srs EPSG:4326 -r near -of GTiff $i /home/martin/Dissertation/GIMMS/GIMMS_lat/`basename $i`; done

I delete all 1981 images, as the year is not complete.

next, we start R, load the raster package and set the directory with the GIMMS Geotiffs:

library(raster)
library(rgdal)
setwd("~/GIMMS")

now load the Tiffs as a raster brick:

sg = stack(list.files(pattern='*.tif'))
gimmsb = brick(sg)
rm(sg)

now let’s chose a pixel and create a time series object. It can either be done by Row and Col:

i=20 # row
j=23 # col
gimms = gimmsb[i,j]
gimms.ts = ts(gimms, start=1982, end=c(2006,24), frequency=24)

or by coordinates

xy <- cbind(-3.6249,14.3844) 
sp <- SpatialPoints(xy) 
data <- extract(gimmsb, sp)
data=as.vector(data)
gimms.ts = ts(data, start=1982, end=c(2006,24), frequency=24)

now the time series can be used for further analysis or plotted:

plot(gimms.ts)

A regression line and a LOESS smoothing can be added:

gimms.time = time(gimms.ts) 
plot(gimms.ts)
abline(reg=lm(gimms.ts ~ gimms.time))
lines(lowess(gimms.ts, f=0.3, iter=5), col="blue")

some playing with boxplots and annual values:

layout(1:2)
plot(aggregate(gimms.ts)); 
boxplot(gimms.ts ~ cycle(gimms.ts))
GIMMS time series analysis

GIMMS time series analysis

averaged over the sudy area:

g=cellStats(gimmsb, stat='mean')
g.ts = ts(g, start=1982, end=c(2006,24), frequency=24)
plot(g.ts)

There are great possibilities for time series analysis in R, e.g. the series can be smoothed, decomposed…etc. The zoo package provides further functions and can handle irregular series. See also the lattice package for better visual presentation. How pixelwise analyis is conducted for time series of rasters, I’ll write in another post.

Detecting environmental change using time series, high resolution imagery and field work – a case study in the Sahel of Mali

Climatic changes and population pressure have caused major environmental change in the Sahel during the last fifty years. Many studies use coarse resolution NDVI time series such as GIMMS to detect environmental trends; however explanations for these trends remain largely unknown.

map

We suggest a five-step methodology for the validation of trends with a case study on the Dogon Plateau, Mali. The first step is to monitor long-term trends with coarse scale time series. Instead of GIMMS, we use a combination of LTDR (derived from AVHRR) and SPOT VGT NDVI data, covering the period from 1982-2010 with a temporal resolution of 10 days and a spatial resolution of 5 km.

Areas with significant trends are further analysed in a second step. Here we use a decomposed MODIS time series with a spatial resolution of 250 m to discover details of the blue spot i9n Figure 1. Due to the large scaled MODIS dataset, trends can be identified at a local scale / village level, see Figure 2.

Fig. 1: LTDR-SPOT showing spatial trends of NDVI. Spatial variations can be observed at a scale of 5.6 km (here the resolution is interpolated to 1 km). South of Fiko the large blue area stands out. This seems to be an area which does not show greening trends after the droughts in the beginning of the 80s.

Fig. 1: LTDR-SPOT showing spatial trends of NDVI. Spatial variations can be observed at a scale of 5.6 km (here the resolution is interpolated to 1 km). South of Fiko the large blue area stands out. This seems to be an area which does not show greening trends after the droughts in the beginning of the 80s.

Fig. 2: The map corresponds to the area of the rectangle in Fig. 1 and shows significant trends of MODIS time series since 2000. At a resolution of 250 m, the spatial patterns are far more diverse and variations within small areas can be detected. This demonstrates that the LTDR-SPOT trends merge many processes into single pixels. Thus further steps must be taken to explain local variations.

Fig. 2: The map corresponds to the area of the rectangle in Fig. 1 and shows significant trends of MODIS time series since 2000. At a resolution of 250 m, the spatial patterns are far more diverse and variations within small areas can be detected. This demonstrates that the LTDR-SPOT trends merge many processes into single pixels. Thus further steps must be taken to explain local variations.

Using very high resolution imagery (e.g. SPOT, Quickbird) areas of interest can be compared with pre-drought Corona-imagery from 1967. This offers a detailed overview of the environmental change at tree-level. A comparison of high resolution imagery with the Corona images show major land use changes over the past fifty years. What used to be dense bush cover has partially been converted to farmer managed agro-forestry and a significant proportion is now degraded land. Furthermore, an increase of tree cover on the fields can be detected. These different trends can also be observed in figures 3 and 4.

Fig. 3:  Left: Quickbird 2007 (Google Earth) Right: Corona 1967

Fig. 3: Left: Quickbird 2007 (Google Earth) Right: Corona 1967

Fig 4:  Left: Quickbird 2007 (Google Earth) Right: Corona 1967

Fig 4: Left: Quickbird 2007 (Google Earth) Right: Corona 1967

Yet many explanations for the changes identified remain unclear.

On-site field work provides information on the land use systems, vegetation composition and the current environmental condition. An initial field trip validated the suspected soil erosion and ongoing loss of trees and shrubs outside the fields used for farming purposes. On the fields surrounding the village many useful trees of all ages were identified. Still many explanations for change can only be speculated and hypothesized.

Fig. 5: There is a huge difference of farming areas (second row) and grazing areas (first row) which are adjacent (images taken in Nov. 2011).

Fig. 5: There is a huge difference of farming areas (second row) and grazing areas (first row) which are adjacent (images taken in Nov. 2011).

Fig. 6: On the one hand trees are protected on farmer's fields, on the other hand the „brousse / forêt“ is exploited for firewood (images taken in Nov. 2011).

Fig. 6: On the one hand trees are protected on farmer’s fields, on the other hand the „brousse / forêt“ is exploited for firewood (images taken in Nov. 2011).

Still many explanations for change can only be speculated and hypothesized. For this reason, interviews with the local population are vital for providing missing details.

Interviews with local people showed that good farmer-management using traditional methods, without outside-influence of projects, led to an increase of tree cover on the fields and healthy environmental conditions.

The land outside of the current farming area is highly degraded, which locals explain by the following points:

  • the extreme droughts in the 1970s and 1980s,
  • lack of rain in the past 30 years,
  • lack of protection by farmers,
  • legal and illegal felling by inhabitants of provincial towns in the region,
  • increased livestock numbers put pressure on the soil and vegetation.

Due to the declining vegetation cover and supported by the unfavourable morphology, the susceptibility to soil erosion increases. Many useful trees and shrubs have become rare or disappeared in these areas (e.g. Butyrospermum parkii, Crataeva adansonii, Combretum micranthum, Piliostigma reticulatum, Pterocarpus lucens, Sclerocarya birrea, etc).

Fig. 6: On the one hand trees are protected on farmer's fields, on the other hand the „brousse / forêt“ is exploited for firewood (images taken in Nov. 2011).

Fig. 6: On the one hand trees are protected on farmer’s fields, on the other hand the „brousse / forêt“ is exploited for firewood (images taken in Nov. 2011).

Fig. 7: Interviewing a farmer from Djamnati (image taken in Nov. 2011).

Fig. 7: Interviewing a farmer from Djamnati (image taken in Nov. 2011).

This example demonstrates the importance of land use and how an integrative and qualitative approach as well as input of local inhabitants expands knowledge and understanding of environmental change in the Sahel. Greening and degradation have many reasons which need to be varified by field work. Our example demonstrates, that climatic factors are important drivers of environmental changes. But land use concepts lead to oppositional results in vegetation development and therefore heterogenous landscape patterns.

see the poster:

egu_poster_small

see: Brandt, M., Samimi, C., Romankiewicz, C. & R. Spiekermann (2012): Detecting environmental change using time series, high resolution imagery and field work – a case study in the Sahel of Mali. Geophysical Research Abstracts, Vol. 14, EGU2012-10583, EGU General Assembly 2012.