Comparison of the Cameroon Weather Synoptic Stations Rainfall Data with TRMM Datasets

Transcription

1 Comparison of the Cameroon Weather Synoptic Stations Rainfall Data with TRMM Datasets Igri Moudi Pascal, Vondou Derbetini, Tanessong Roméo, Komkoua Mbienda Laboratory of Environmental Modeling and Atmospheric Physics (LEMAP), University of Yaounde I, Cameroon Ecole Africaine de la Météorologie et de l Aviation Civile (EAMAC), Niamey, Niger 1 est MTGV, Sept 2013, Toulouse, France 27 septembre septembre / 46

2 Acknowledgements Acknowledgements With acknowledgements to 1 1 st MTGV organizing committee for the Financial Support 2 To my colleagues and supervisor at the LEMAP, Yaounde 27 septembre / 46

3 Introduction The context of the study Heavy precipitations and droughts greatly impact on the modification of socio-economic planning in the world, especially in developing countries (Africa for example) where populations are highly vulnerable. 27 septembre / 46

4 Introduction The context of the study Heavy precipitations and droughts greatly impact on the modification of socio-economic planning in the world, especially in developing countries (Africa for example) where populations are highly vulnerable. Context With shifting seasons Increasing water scarcity More frequent and intense extreme events Climate change is bringing a series of disaster impacts to the poorest and most vulnerable countries and communities, and is placing development assistance at risk. 27 septembre / 46

5 Continue... Context Progressively more attention has been given to converging Disaster Risk Reduction (DRR) and Climate Change Adaptation (CCA) agendas conceptually and in practice at sub-national, national and international levels. 27 septembre / 46

6 Continue... Context Progressively more attention has been given to converging Disaster Risk Reduction (DRR) and Climate Change Adaptation (CCA) agendas conceptually and in practice at sub-national, national and international levels. Context Weather Forecasting and Climate projections could help reducing such Disasters. But Models results are generally validated based on direct comparison with gauge data or reanalysis data over the region of interest, but the spatial repartition of synoptic weather stations is sparse and heterogeneous in these regions as the meteorological observation network is seriously deteriorating. 27 septembre / 46

7 Continue... Context In fact, stations in stations at the beginning of the year 1991 The reduction of the number of synoptic stations is particularly affecting Cameroon and Nigeria 27 septembre / 46

8 Aim of the study Objective The aim of this study is to evaluate how accuracy are the 3B43 TRMM data by a direct comparison with the Cameroon National Meteorological Department (NMD) data and to show whether synoptic stations data could be replaced by these datasets (TRMM) in Cameroon when validating models. The study is done for the period septembre / 46

9 Study domain 27 septembre / 46

10 Study domain Domain 27 septembre / 46

11 Data used Data Rain gauge data Tropical Rainfall Measuring Mission (TRMM). 27 septembre / 46

12 Data used Data Rain gauge data Tropical Rainfall Measuring Mission (TRMM). Data Version 6 data 3B43 combined is used. It provides mean monthly estimations of rainfall on a grid of As the stations are assumed to be a gridded points, the corresponding TRMM data have been retrieved using NCAR Command Language (NCL) tool. 27 septembre / 46

13 Methodology Main climatic regions in Cameroon Cameroon is divided into five agroecological zones, representing the different weather and climate of Cameroon : The soudano-sahelian zone (approximatively 400 to 1200 mm/year, the hottest, the driest and the sunniest) The High savannah zone (on average 1500 mm/year) The Highlands zone (on average 1500 to 2500 mm/year) The forest bi-modal zone (about 2000 mm/year) The forest mono-modal zone (2500 mm/year) 27 septembre / 46

14 Agro-ecological areas Climatic zones FIG.: Study domain with different agro-ecological areas 27 septembre / 46

15 Methodology Processing methods Results are presented and discussed according to the five agroecological zones. TRMM data are compared with gauge data (GD) at 0.25 x 0.25 grid resolution over the five agroecological zones. The accuracy of the TRMM data is assessed in terms of annual correlation coefficient and annual bias. The bias is calculated as the mean departure between TRMM and GD data. Mathematical formulation of bias and correlation coefficient (r) is given respectively by 27 septembre / 46

16 Methodology Processing methods Bias = 1 N N (x i y i ) = x ȳ (1) i=1 r = N i=1 (x i x)(y i ȳ) N i=1 (x i x) (2) 2 N i=1 (y i ȳ) 2 where N is the number of year, x i is the value of the TRMM rainfall to the i th station, y i is the value of the GD rainfall to the i th station. x is the time average of TRMM data ; ȳ is the time average of GD data (station). 27 septembre / 46

17 Results Mean monthly accumulated rainfall : The Sudano-Sahelian zone 27 septembre / 46

18 Results Mean monthly accumulated rainfall : The Sudano-Sahelian zone 27 septembre / 46

19 Discussion Discussion The sudano-sahelian zone of Cameroon is the driest one It receives approximatively on average 3 months of rainfall during the African monsoon period The rainy season effectively starts in June and ends in September. Maximum rainfall occurs in August with amount exceeded 200 mm in Garoua and Kaele. This zone, which empasses the Maroua, Garoua main cities, is the most vulnerable zone of the country to extreme events such as droughts and floods. 27 septembre / 46

20 Results and Discussion Mean annual correlation 27 septembre / 46

21 Discussion Discussion Good correlation coefficient between GD and TRMM datasets which reaches 0.8 on average for different year. Moreover, the annual correlation is generally higher than 70% in all the period of study. 27 septembre / 46

22 Results and Discussion Mean annual biases 27 septembre / 46

23 Discussion Discussion Among the years studied, we have in general negative biases in Kaele and Garoua, that lead to a strong tendency of overestimation by GD, although there is a remarkably good agreement between the TRMM and GD in MAM and ASON. Bias appears to be random with either overestimation or underestimation and ranging from -18 mm to 14 mm. This analysis tends to confirm the trend seen above. 27 septembre / 46

24 Results Mean monthly accumulated rainfall : The High Savannah zone 27 septembre / 46

25 Results Mean monthly accumulated rainfall : The High Savannah zone 27 septembre / 46

26 Discussion Discussion Both TRMM algorithm and GD in these stations have similar seasonal distributions during the monsoon period, especially in August when rainfall occurrence reaches it peak. Discrepancy between these datasets is particularly more highlighted in Banyo where TRMM retrieved data overestimate the GD. 27 septembre / 46

28 Discussion Discussion The global tendency is overestimation by TRMM algorithm. Both datasets statistically agree by displaying high annual correlation. For example, annual correlation in Ngaoundere station is at least 0.85 during the eight years of available data (from 1998 to 2005). 27 septembre / 46

29 Results Mean monthly accumulated rainfall : The Highland zone 27 septembre / 46

30 Results Mean monthly accumulated rainfall : The Highland zone 27 septembre / 46

31 Discussion Discussion The same distribution compared to the High Savannah zone (Mountains areas) 27 septembre / 46

32 Results Mean annual correlation 27 septembre / 46

33 Results Mean annual biases 27 septembre / 46

34 Discussion Discussion Not particular results different from the High Savannah zone (Mountains areas) 27 septembre / 46

35 Results Mean monthly accumulated rainfall : The Bi-modal forest zone 27 septembre / 46

36 Results Mean monthly accumulated rainfall : The Bi-modal forest zone 27 septembre / 46

37 Results Mean monthly accumulated rainfall : The Mono-modal forest zone 27 septembre / 46

40 Discussion Discussion Departure between TRMM 3B43 and GD data is remarkable in the forest zones The main features is overestimation. Monthly mean zonal analysis for the study period reveals that such features are persistent during the whole year 27 septembre / 46

41 Discussion Discussion Departure between TRMM 3B43 and GD data is remarkable in the forest zones The main features is overestimation. Monthly mean zonal analysis for the study period reveals that such features are persistent during the whole year Discussion But, an exception behaviour has been shown in the Douala station where TRMM data are underestimated all over the study period, and this behaviour can be partially attributed to sea-breeze. 27 septembre / 46

45 Results Mean annual biases 27 septembre / 46

46 Discussion Discussion The highest negative biases are recorded in the Douala station with values reaching -110 mm/yr and the highest positive biases are recorded in the Mpundu station with a value higher than 200 mm/yr Forest zone has the poorest agreement for all years considered. This can be caused by orography or other synoptic perturbations. In both forest zones (bi and mono modal), the poorest agreements with rain GD are recorded in the monsoon period corresponding to the peak of the wet season (JJA). This zone has the worst correlation 27 septembre / 46

47 Conclusion Concluding remarks Comparisons of the rain GD in the Cameroon weather synoptic stations with TRMM 3B43 algorithm has been investigated over the five agroecological zones. 1 TRMM 3B43 is slightly underestimated in the Sahel, Savannah and Highlands zones 2 TRMM 3B43 is overestimated in the mono-modal and bi-modal forest zones 3 Bias is generally most pronounced in the bi-modal forest zone 27 septembre / 46

48 Conclusion Concluding remarks For instance, the best agreements with rain gauge data are obtained for the Cameroon northern zone rather than for its southern counterpart leading to two majors climatic regions in Cameroon : the North region of Cameroon with the rainy season in JJA and the South region with the rainy season ranging from May to November. 27 septembre / 46

49 Conclusion Perspectives We intend to use the Megha Tropiques data to validate Rain Gauge data in Cameroon and then validate WRF 3D-Var model using these data sets 27 septembre / 46

50 The END THANKS FOR YOUR KIND ATTENTION 27 septembre / 46