Tropical Applications of Meteorology using SATellite data
Analysing Climate Trends in African Rainfall using
Satellite Data
Ross Maidment
TAMSAT Research Group, University of Reading
Supervisors: Dr David Grimes and Dr Richard Allan
Tropical Applications of Meteorology using SATellite data
Overview
1. Introduction2. Aims of my research3. The TAMSAT methodology to estimate rainfall4. Validation of rainfall estimates in Uganda5. Meteosat satellite data6. Initial results – Ethiopia7. Future work and Conclusions
Tropical Applications of Meteorology using SATellite data
1. Introduction Most of the African continent is economically dependent on rain
fed agriculture, therefore changes in expected rainfall patterns can have serious consequences, both economically and from a humanitarian point of view
However, future predictions must be based on a secure knowledge of the present rainfall climate
The current African rainfall climatology is poorly understood, mainly due to inadequate ground based observations
GPCC: Gauge locations – 2007, March - May
Tropical Applications of Meteorology using SATellite data
Introduction Cont...
Satellite based rainfall data sets exist but many, such as the TRMM based algorithms only cover short time periods
Those covering longer time periods, such as the widely used GPCP product tend to be a blend of different satellite sensors and gauge data, the proportions of which vary from year to year
This makes these data sets unsuitable for climate studies as it is impossible to extract meaningful trends and statistics when biases vary interanually
Tropical Applications of Meteorology using SATellite data
Overview
1. Introduction2. Aims of my research3. The TAMSAT methodology to estimate rainfall4. Validation of rainfall estimates in Uganda5. Meteosat satellite data6. Initial results – Ethiopia7. Future work and Conclusions
Tropical Applications of Meteorology using SATellite data
2. Aims of my research
1. To create TARCAT (TAMSAT African Rainfall Climatology And Time-series), a temporally homogeneous 30 year rainfall time-series and climatology for all of sub-Saharan Africa using the TAMSAT approach to rainfall estimation
2. Analyse this data set to better understand temporal trends and statistics that describe the rainfall climate
3. Evaluate model data to increase confidence in future predictions
Tropical Applications of Meteorology using SATellite data
Overview
1. Introduction2. Aims of my research3. The TAMSAT methodology to estimate rainfall4. Validation of rainfall estimates in Uganda5. Meteosat satellite data6. Initial results – Ethiopia7. Future work and Conclusions
Tropical Applications of Meteorology using SATellite data
3. TAMSAT Methodology The TAMSAT approach to rainfall estimation is based entirely on
Meteosat Thermal Infra-Red imagery to identify precipitating cumulonimbus clouds (deep convection)
Tropical Applications of Meteorology using SATellite data
TAMSAT Methodology Cont...
• Calculate Cold Cloud Duration (CCD) for each pixel (length of time cloud top is colder than Tt )• Estimate rainfall total as rain = a0 + a1 CCD a0, a1, Tt are calibrated using local gauges from historic data• Calibration parameters vary in space and time (i.e. local calibration)• Resolution: Temporal – 10 days (1 dekad), Spatial – 0.05° (sat. pixel)
• Using Meteosat TIR imagery, identify optimum cloud top temperature threshold Tt distinguishing between rain and no rain
Tropical Applications of Meteorology using SATellite data
TAMSAT Operational Product
Tropical Applications of Meteorology using SATellite data
Overview
1. Introduction2. Aims of my research3. The TAMSAT methodology to estimate rainfall4. Validation of rainfall estimates in Uganda5. Meteosat satellite data6. Initial results – Ethiopia7. Future work and Conclusions
Tropical Applications of Meteorology using SATellite data
4. Rainfall Validation in Uganda Validating rainfall products
against spatially interpolated gauge data (res: 0.5° x 0.5°)
Period of study: 2001 to 2005 for first rainy
season (Feb – June)
Figure: spatially averaged dekadal rainfall
ERA-40 ERA-Interim
TAMSAT RFE 2.0
GPCP
Product Bias/mm RMSD/mm R2
ERA-40 -9.86 18.09 0.41 ERA-Interim TAMSAT
14.06 -1.01
26.55 10.41
0.39 0.72
RFE 2.0 GPCP
-1.73 0.16
11.00 11.56
0.74 0.72
Tropical Applications of Meteorology using SATellite data
Rainfall Validation in Uganda Cont... Temporal variability using 2003 and 2004 data
Tropical Applications of Meteorology using SATellite data
Overview
1. Introduction2. Aims of my research3. The TAMSAT methodology to estimate rainfall4. Validation of rainfall estimates in Uganda5. Meteosat satellite data6. Initial results – Ethiopia7. Future work and Conclusions
Tropical Applications of Meteorology using SATellite data
5. Meteosat TIR data Meteosat archive spans 30 years, 1980 to present, giving
complete African coverage every 30 minutes (First Generation) and every 15 minutes post June 2006 (Second Generation)
Problems with the TIR archive:
1. Archive isn’t complete, with some gaps from one slot
up to several days (these gaps will be interpolated)
Tropical Applications of Meteorology using SATellite data
Meteosat TIR data - Problems
OK < 6 hours < 36 hours > 36 hours
Tropical Applications of Meteorology using SATellite data
Meteosat TIR data - Problems 2. Change in satellite sensor introducing artificial temporal discontinuities in time-series
- Timeline of warmest/brightest pixel over a ‘cloud free’ region of the Equatorial Atlantic to detect sudden changes:
Tropical Applications of Meteorology using SATellite data
Meteosat TIR data - Problems
Satellite (Meteosat 2 – 9)
Daily warmest pixel – warm ocean scene
Tropical Applications of Meteorology using SATellite data
Meteosat TIR data - Problems
MET-7 MET-8/9MET-6 MET-7
Satellite (Meteosat 2 – 9)
Daily warmest pixel – warm ocean scene
Tropical Applications of Meteorology using SATellite data
Meteosat TIR data - Problems Other problems:
3. Changes in data format from MFG (OpenMTP) to MSG (Native)
4. Corrupt images (missing pixels, lines and whole images)
5. Failure in reading the header of some OpenMTP files
6. Change in radiance definition from ‘spectral’ to ‘effective’ radiance in May 2008
7. Using data obtained from the EUMETSAT archive that is from the non-prime satellite
Tropical Applications of Meteorology using SATellite data
Overview
1. Introduction2. Aims of my research3. The TAMSAT methodology to estimate rainfall4. Validation of rainfall estimates in Uganda5. Meteosat satellite data6. Initial results – Ethiopia7. Future work and Conclusions
Tropical Applications of Meteorology using SATellite data
6. Initial results - Ethiopia Why Ethiopia?
- TAMSAT involved in joint project (with IRI, Columbia University and NMA, Ethiopia) based on Ethiopia and funded by Google (Dinku et al 2011)
- Complex and interesting rainfall climate
Have compared (so far):TAMSAT GPCP ERA-Interim
Tropical Applications of Meteorology using SATellite data
Initial Results – Ethiopia Cont...
Tropical Applications of Meteorology using SATellite data
Initial Results – Ethiopia Cont...Mean dekadal rainfall
Tropical Applications of Meteorology using SATellite data
Initial Results – Ethiopia Cont...
Period: 1989 – 2009
Tropical Applications of Meteorology using SATellite data
Initial Results – Ethiopia Cont...
TAMSAT (1983-2009)
GPCP (1979-2008)
ERA-Interim (1989-2009)
N (no. of months) 324 360 252 Mean (mm) 51.24 57.90 61.42
Standard deviation (mm) 40.11 37.87 36.62 Variance (mm2) 1609.83 1434.81 1341.69
Trend (mm/month) 0.024 -0.007 0.071 Min (mm) 0.07 1.3 1.42 Max (mm) 159.16 204.20 171.3
6 Month & Annual Running Mean – Mean Monthly Rainfall
Tropical Applications of Meteorology using SATellite data
Initial Results – Ethiopia Cont...TARCAT – Trend during 1983-2009 period
Tropical Applications of Meteorology using SATellite data
Overview
1. Introduction2. Aims of my research3. The TAMSAT methodology to estimate rainfall4. Validation of rainfall estimates in Uganda5. Meteosat satellite data6. Initial results – Ethiopia7. Future work and Conclusions
Tropical Applications of Meteorology using SATellite data
7. Future work and Conclusions• Complete TARCAT and extend study to all of Africa• Main interests:• trends over the last 30 years• changes in length and start of wet season• changes in wet/dry regimes• changes in frequency of extreme events
• Evaluate model data
The TAMSAT methodology and the Meteosat archive provides the opportunity to create a unique and consistent data set that can enhance our current understanding of the African rainfall climate and hopefully increase our confidence in future predictions
Tropical Applications of Meteorology using SATellite data
Thank you for your time
Photo courtesy NASA Earth Observatory