CHIRPS 2.0 prelim (sub 1) Tanzania data and trends Comparison to CHIRPS v1.9 1/27/2015 Laura Harrison
CHIRPS 2.0 prelim (sub 1)
Tanzania data and trends
Comparison to CHIRPS v1.91/27/2015
Laura Harrison
Overview 1: NDVI and rainfall trends
Main differences• Parts of central TZ are
wetter now on average• Acceptable• But the clear 30-year drying
trend that matched with the 30-year NDVI declines is now gone
Time series comparisons• Show that CHIRPS 2.0 prelim
seems fine• In some areas more strongly
correlated with NDVI than before
• Questionable areas– Shinyanga-Tabora central area.
2 stations here. Are they ok?– Area of greening trend (see
Overview 2: Seasonal Trends)
Rainfall isohyets (Oct-May total)Base layer: NDVI-3g (Nov-Oct mean)
CHIRPS v1.9 CHIRPS v2.0p Wetter climate
Topography
Slope of linear trend 1982-2010Oct-May rainfall total
CHIRPS v1.9 CHIRPS v2.0p
Weaker drying trend:• Central TZ• North-central TZ
Using z-scores
Slope of linear trend 1982-2010Oct-May rainfall total and Nov-Oct NDVI mean
CHIRPS v1.9 NDVI-3gPreviously in CHIRPS
(x 2.5 fix)
Box 1
Box 2
Box 3
Box 4
Using z-scores
Slope of linear trend 1982-2010Oct-May rainfall total and Nov-Oct NDVI mean
CHIRPS v2.0p NDVI-3gCurrently in CHIRPS
Box 1
Box 2
Box 3
Box 4
Using z-scores
Rainfall and NDVI time seriesBox 1: Shinyanga-Tabora west
CHIRPS v1.9 CHIRPS v2.0
NDVI improvesRainfall does not
NDVI improvesRainfall improves
Stronger correlation
No rainfall trend (1982-2010)Rainfall trend is significant and matches NDVI trend (1982-2010)
Reality check: OK
Using z-scores
Rainfall and NDVI time seriesBox 2: Shinyanga-Tabora central
CHIRPS v1.9 CHIRPS v2.0
Weaker correlation
No rainfall trend (1982-2010)Rainfall trend is significant and matches NDVI trend (1982-2010)
Extremes are mostly wetter, but not always.• Wetter wet years: 1997-98, 2006-07• Wetter dry years: 2007-08, 2010-11• Drier dry years: 1983-84, 1999-00
Wet yearsWet years
Dry years Dry years
Reality check: ?
There are 2 stationsin this area
Using z-scores
Rainfall and NDVI time seriesBox 3: Singida
CHIRPS v1.9 CHIRPS v2.0
Same correlation
Rainfall trend not significantRainfall trend is significant (1982-2010)
Minimal difference
Reality check: OK
Using z-scores
Rainfall and NDVI time seriesBox 4: Arusha
CHIRPS v1.9 CHIRPS v2.0
Stronger correlation
Reality check: OK
Using z-scores
Overview 2: Seasonal trends
Main differences• Oct-Feb trends
– Weaker drying trends– Significant drying trends gone in
Shinyanga (TZ), Congo, Rwanda– Significant drying trend still in southern TZ
• Mar-May trends– Weaker drying trends– Arusha area shows only a weak Long Rains
drying trend– New wetting trend (significant) in
northern TZ– Significant drying trends still in Tanga &
Morogoro
• Oct-May total trends (revisited)– Significant drying trends gone
in central and north-central TZ Congo, Rwanda
– Significant drying trends still in southern TZ
– New significant drying trends in east TZ
– Stronger, larger drying trends in Kenya
Slope of Sen trend 1982-2010Oct-May rainfall total
CHIRPS v1.9 CHIRPS v2.0pPercent change per decade Stippling shows significance at p < 0.05
Overall: Weaker drying trends in central and north-central TZ
Stronger Kenya trends
Previously an area with significant drying trend
Slope of Sen trend 1982-2010Oct-Feb rainfall total
CHIRPS v1.9 CHIRPS v2.0pPercent change per decade Stippling shows significance at p < 0.05
Previously an area with significant drying trendOverall: Weaker Oct-Feb trends; only
southern TZ is significant in CHIRPS v2.0p
Slope of Sen trend 1982-2010Mar-May rainfall total
CHIRPS v1.9 CHIRPS v2.0pPercent change per decade Stippling shows significance at p < 0.05
Previously an area with significant drying trendLoss of Long Rains drying trend
New wetting trends