REPORT ON SOWING PERIOD AND FLOODING RISKS IN DOUALA, CAMEROON. Cameroon Meteorological Department

Transcription

1 REPORT ON SOWING PERIOD AND FLOODING RISKS IN DOUALA, CAMEROON BY GERVAIS DIDIER YONTCHANG Cameroon Meteorological Department PROJECT WORK FOR PART FULFILMENT OF THE e-siac COURSE (STATISTICS IN APPLIED CLIMATOLOGY APRIL AUGUST 2010) MAINTAINED BY THE MET. OFFICE COLLEGE AND THE SSC WITH SUPPORT FROM THE UK MET. OFFICE AND WMO. August

2 TABLE OF CONTENTS REPORT ON SOWING PERIOD AND FLOODING RISKS IN DOUALA, CAMEROON... 1 TABLE OF CONTENTS... 2 A. BACKGROUND: ). Extreme climate events increasingly important for Cameroon ) Location and climate of Douala, Cameroon in Africa... 3 B. OBJECTIVES:... 4 i. Assessment of the annual distribution of sowing dates... 4 ii. Assessment of the flooding risks during the rainfall peak period... 4 iii. To quantify the risk of having floods with crop failure... 4 iv. Towards development of a possible cautious strategy... 4 C. METHODOLOGY: Find the dates when sowing are possible using the suitable definitions Assessment of the risk of flooding with crop failure Boxplots for visual comparison, descriptive statistics as dates and Comment on the summaries Some conclusions for the client (Ministry of agriculture) based on these summaries and risks calculations D. DATA AND DEFINITIONS USED: Daily rainfall data for the period of 1950 to 2006 collected at the Douala international Airport DATA reformatted using SSC-Stat tool before it could be imported from EXCEL to INSTAT... 6 E. RESULTS:... 1) - Find the dates when sowing and flooding is possible using suitable definitions: Assessment of the risk of flooding with crop failure: RISK OF FLOODING WITH CROP FAILURE Text boxes showing the chance (proportion) of experiencing floods and crop failure during the month of August: Boxplots for visual comparison, descriptive statistics as dates and Comment on the summaries: F. IMPLICATIONS: The sowing strategy definition that I would recommend to a farmer The advantages The disadvantages G.APPENDICES: Appendix A: References and sources documentation Appendix: Daily data for Douala, Cameroon Day numbers in the year

3 BACKGROUND: 1). Extreme climate events increasingly important for Cameroon Climate extremes such as droughts and floods have increased in frequency and severity in Africa over the past 30 years (Nicholson, 1987). The probability of extreme climate events is becoming increasingly important for Cameroon because of the inherent vulnerability of its agricultural system and even the associated risk of the potential spread of plant, animal as well as human diseases (malaria, typhoid, cholera, etc.) that are highly linked to climate variability and climate change. 2) Location and climate of Douala, Cameroon in Africa Cameroon lies within latitudes 1 45 N to 13 N and Longitude 8 25 E to E. It is situated on the west coast of Africa, running north to south from Sahara desert to the Atlantic Ocean. The country is bounded on the north by Lake Chad; on the east by Chad and the Central African Republic; on the south by the Republic of the Congo, Gabon, and Equatorial Guinea; and on the west by the Bight of Biafra (an arm of the Atlantic Ocean) and Nigeria (Figure 1). The country is shaped like an elongated triangle, and forms a bridge between West Africa and Central Africa. Cameroon has a total area of 475,442 km 2. Yaoundé is the capital, and Douala is the largest city. The major industries (petroleum refineries, fishing industries, international air transports and others) are located in this region. Douala city is situated at the junction of the southern coastal plain region with dense equatorial rain forests and the west area of high forested mountains of volcanic origin, where lies Mount Cameroon Mountain (4095m), the highest peak in western Africa and an active volcano. One of the country s most fertile soils is found in this region. July and August are seen to be the peak of the rainy season, with a lowest total of 207.4mm in July 2003 and one year (August1966) with over 1240mm. The rains are usually from January to December, though December, January and February are occasionally dry. The mean maximum relative humidity occurs here due to its proximity to Atlantic Ocean (exposure to the warm Guinean Gulf s currents). Here, the type and intensity of the weather is determined by the meridional variations in the depth of the monsoon layer (moisture). The southwest (SW) monsoon flow at lower levels, source of humidity to a large extent determined by the large scale atmospheric circulation and 3

4 sea-air interaction over much of the tropical Atlantic sector.(lamb,1983). Douala Cameroon (about 4ºN) is well usually marked by a zone with deep active and convective clouds, thunderstorms and squall lines (SW winds) and heavy precipitations. The African Waves are often associated with Squall lines that form, develop and dissipate within the waves. Figure 1: Location of Douala, Cameroon in Africa B. OBJECTIVES: The main objective of the study is to investigate the sowing period variability and flooding risks in the city of Douala. These will be achieved by addressing the following specific objectives: i. Assessment of the annual distribution of sowing dates; ii. iii. iv. Assessment of the flooding risks during the rainfall peak period; To quantify the risk of having floods with crop failure for each week in the month of August, where 100mm of rain or more in a day has caused flooding and crop destruction. Again we compute the number of years in the time series when at exactly one day in the week in question has registered 100mm or more of rain. If the event was observed at least once, then we define that a week experiences a crop failure. Week 1 is from day number 212 to day number 218; Week 2 is from day number 219 to day number 225; Week 3 is from day number 226 to day number 232; Week 4 is from day number 233 to day number 239. Towards development of a possible cautious strategy. 4

5 C. METHODOLOGY: The methods used to achieve the objectives mentioned above are the following: All these results describe the rainfall data for Douala, Cameroon and were computed (plots, graphs, tables, figures, risks or variability calculations) using Instat+ v3.36 and Excel packages. To define the events of interest (start of the season, dry spell) the climatic>>events menu options from INSTAT were used. Then the sowing dates were extracted and analyzed based on the below definitions. 1. Find the dates when sowing are possible using the suitable definitions: a)- First date from 1 st March getting more than 40mm in 1,2,3 or 4 days; b)- First date from 1 st April getting more than 40mm in 1,2 or 3 days; c) First date from 1 st May getting more than 40mm in 1 or 2 days; d) First date from 1 st June getting more than 40mm in 1 day 2. Assessment of the risk of flooding with crop failure e) Fist occasion after 1 st August with more than 100mm totalled over 2 consecutive days and no dry spell of 3 days or more in the next 30 days. If the event was observed at least once we define that a year experiences a flood in low-land areas. f) First occasion after 1 st August with more than 100mm totalled exactly over one day. If the event was observed at least once we define that a year experiences a general flood with crop failure. g) The average rainfall total for August will be compared with the average totals for the other months to get a general idea as to how wet the month is. Then the month will then be broken down in 4 weeks as specified above, where the average total for each week will be looked at and the highest rainfall totals will be highlighted. After this, a count of the number of days with at least 100mm in exactly 1 day will be looked at for each week, then find the average number of times per week each of these occurred over the past 57 years. This will give the probability of the event of flooding, and therefore the level of risk for encountering crop failure in either week 1, 2, 3 or 4 will be estimated. 3. Boxplots for visual comparison, descriptive statistics as dates and Comment on the summaries. 4. Some conclusions for the client (Ministry of agriculture) based on these summaries and risks calculations. 5

6 D. DATA AND DEFINITIONS USED: 1. Daily rainfall data for the period of 1950 to 2006 collected at the Douala international Airport Daily rainfall data for the period of 1950 to 2006 collected at the Douala international Airport which was provided by the Cameroon meteorological service. The accumulated number of years of daily data is 57, because almost all the data was available. A rainy day in this instance is considered to be a day with at least 0.85mm of rainfall. The data described are the daily rainfall values from 1950 to The mean annual total rainfall over this period was 3895mm with an average of about 220 rainy days. The highest annual total was mm in 1956; the lowest total recorded was mm in The daily rainfall data was originally provided in EXCEL format by the Meteorological office. It was then examined to ensure its quality. A thorough inspection was done in an effort to clean the data from any erroneous values whenever necessary. There where very few missing data and no negative values. The measurement units millimeters were consistently used throughout the dataset. The data had to be reformatted. 2. DATA reformatted using SSC-Stat tool before it could be imported from EXCEL to INSTAT DATA was reformatted using SSC-Stat tool before it could be imported from EXCEL to INSTAT where supplementary checks were conducted to ensure adequate coding for nonleap years and proper calendar year settings. The coding used fro non leap years was 9988 represented as ***. 6

7 E. RESULTS: These results describe the rainfall data for Douala, Cameroon. This section gives all the results (plots, graphs, tables, figures, risks or variability calculations, using Instat+ v3.36). It is also the space for analysis of the events. To define the events of interest (start of the season, dry spell) the climatic>>events menu options from INSTAT were used. Then the sowing dates were extracted and analyzed based on the given definitions. 1) - Find the dates when sowing and flooding is possible using suitable definitions: Table 1: Sowing & flooding dates YEAR 1st Mar 1st Apr 1st May 1st Jun 1st Aug floods 1st Mar 1st Apr 1st May 1st Jun 1st Aug flood s Mar 3-Apr 24-May 3-Jun 22-Aug 10-Sep Apr 23-Apr 4-May 8-Jun 1-Aug 1-Sep Mar 14-Apr 4-May 1-Jun 14-Aug 14-Aug Mar 29-Apr 1-May 7-Jun 21-Aug 18-Sep Mar 13-Apr 12-May 7-Jun 4-Sep 12-Sep Mar 8-Apr 2-May 14-Jun 9-Aug 28-Aug Mar 8-Apr 20-May 2-Jun 11-Aug 11-Aug Mar 1-Apr 18-May 9-Jun 1-Aug 5-Aug Mar 12-Apr 11-May 17-Jun 7-Aug 14-Sep Mar 4-Apr 2-May 7-Jun 2-Aug 2-Aug Mar 17-Apr 9-May 19-Jun 5-Aug 5-Aug Mar 1-Apr 29-May 29-Jun 11-Aug 13-Aug Mar 14-Apr 4-May 7-Jun 1-Aug 11-Aug Apr 5-Apr 7-May 4-Jun 14-Aug 14-Aug Mar 7-Apr 3-May 4-Jun 14-Aug 14-Aug Mar 8-Apr 6-May 9-Jun 2-Aug 15-Aug Mar 1-Apr 5-Jun 5-Jun 4-Aug 23-Aug Apr 12-Apr 11-May 16-Jun 16-Aug 24-Aug Mar 19-Apr 5-May 9-Jun 30-Aug 21-Sep Mar 3-Apr 9-May 22-Jun 11-Aug 2-Sep Mar 12-Apr 16-May 25-Jun 1-Aug 3-Sep Mar 4-Apr 20-May 13-Jun 2-Aug 2-Sep Mar 17-Apr 2-Jun 2-Jun 10-Aug 10-Aug Mar 4-Apr 30-May 1-Jun 6-Aug 1-Sep Mar 23-Apr 8-May 29-Jun 4-Aug 14-Aug Mar 18-Apr 8-May 7-Jun 17-Aug 29-Sep Apr 2-Apr 7-May 5-Jun 1-Aug Mar 5-Apr 28-May 12-Jun 12-Aug 12-Aug Apr 5-Apr 11-May 1-Jun 3-Aug 19-Aug Mar 19-Apr 1-May 1-Jun 2-Aug 27-Sep Mar ** 11-May 2-Jun 5-Sep 6-Sep 77 *** Mar 1-Apr 3-May 8-Jun 1-Aug 22-Aug Mar 9-Apr 2-May 28-Jun 21-Aug 12-Sep Mar 1-Apr 9-May 22-Jun 2-Aug Mar 4-Apr 17-May 21-Jun 8-Aug 8-Aug Mar 1-Apr 2-May 13-Jun 5-Aug 30-Aug Mar 2-Apr 3-May 4-Jun 11-Aug 3-Nov Mar 4-Apr 28-May 26-Jun 17-Aug 17-Aug Mar 21-Apr 20-May 17-Jun 24-Aug 24-Aug Mar 16-Apr 4-May 12-Jun 12-Aug 12-Aug Apr 24-Apr 3-May 2-Jun 9-Aug 25-Aug Apr 8-Apr 1-May 21-Jun 5-Aug

8 Rainfall ( mm ) Mar 14-Apr 6-May 13-Jun 5-Oct 5-Oct Mar 13-Apr 14-May 3-Jun 4-Aug 30-Aug Mar 3-Apr 2-May 22-Jun 9-Aug 9-Aug Mar 15-Apr 8-May 5-Jun 3-Aug 3-Aug Mar 11-Apr 27-May 1-Jun 18-Aug Mar 6-Apr 1-May 16-Jun 1-Aug Apr 6-Apr 2-May 15-Jun 13-Aug 13-Aug Mar 3-Apr 7-May 18-Jun 3-Aug 3-Sep Mar 3-Apr 5-Jun 5-Jun 3-Aug 16-Aug Mar 2-Apr ** 3-Jun 20-Sep 23-Sep *** Mar 1-Apr 1-May 15-Jun 15-Aug 15-Aug Mar 9-Apr 24-May 7-Jun 5-Aug 5-Aug Mar 4-Apr 3-May 9-Jun 0-Jan Mar 21-Apr 4-May 5-Jun 6-Aug 14-Sep Mar 1-Apr 4-May 14-Jun 1-Aug 17-Aug Earliest 1-Mar 1-Apr 1-May 1-Jun 1-Aug 1 st -Aug Latest 24-Apr 29-Apr 5-Jun 29-Jun 5-Oct 3-Nov Assessment of the risk of flooding with crop failure: Graph of the Average monthly rainfall for years at Douala International Airport Average minimum monthly rainfall Average monthly rainfall Average maximum monthly rainfall Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Graph (i): Average monthly rainfall for years (Douala) According to graph (i) (see also table1, page 25 of appendix), August is the wettest month of the year with an average of mm over 57 year period. 8

9 Graph (ii): Total rainfall in August for years (Douala Airport) Graph (ii) shows the yearly totals of rainfall for the month of August throughout the years. The year with the highest totals was 1966 with 1240 mm and the year with the lowest total was 1954 with a mere mm of rainfall. 3. RISK OF FLOODING WITH CROP FAILURE Table 3: showing the yearly totals of rainfall within the month of August Year Week 1 Week 2 Week 3 Week ,4 81,1 135,7 231, ,8 132,1 82,9 117, ,7 94,4 292,4 188, , ,9 40,6 31,5 77, ,1 271,8 130,9 125, , ,4 37, ,8 206,8 185,7 271, ,7 243,7 127,7 134, ,6 262,4 195,7 220, ,5 508, , ,4 270,3 334,9 197, ,9 217,5 168,4 152, ,2 117, , ,3 32, , ,5 155,1 317,6 210, ,5 229,2 290,5 444,4 9

10 ,3 175,8 198,8 369, ,4 126,9 65,5 58, ,5 144,8 189,8 193, ,7 170,4 198, ,2 211,7 138,7 239, ,9 296,5 167,9 66, , ,4 137, ,2 67,7 497,9 140, ,3 82,7 134,6 211, , , ,5 257,8 26, ,2 143,6 268,2 189, ,6 97,6 126,7 92, ,4 91,7 82,3 106, ,9 132,3 219,2 285, ,8 73,7 96,8 267, ,8 84,4 60,4 104, ,8 177,6 94,9 50, ,8 125, , ,1 42, ,4 90,3 310,9 323, ,9 259, ,7 213,9 352,4 200, ,9 226,2 127,3 211, ,4 131,6 206,6 199, ,1 49,9 59,2 72, ,5 149,9 133,7 174, ,3 248,1 117, ,3 105,3 171,6 126, ,4 58,6 272,4 64, ,6 124, , , ,8 170, ,4 42,3 61, ,7 72, , ,5 83,6 55, ,8 189,5 252,2 159, ,1 161,8 281, ,5 55,1 57,1 43, ,5 134,7 125,8 200, ,9 21,2 288,1 161,6 Total , ,9 9742,2 10

11 Graph (iii): Yearly totals rainfall in August for week 1, 2, 3 & 4 Table (3) and graph (iii) show that week 2 and 4 are those with the least amount of rainfall while week 3 and 1 are the wettest. Assessment of the weekly risk of flooding with crop failure: To quantify the risk of experiencing a flood for each week in the month of August, when 100mm or more rain in exactly one day has occurred. We compute the number of years in the time series when at least one day in the week in question has registered 100mm or more of rain. If the event was observed at least once, then we define that a week experienced a flood. TABLE 4: Showing the frequency of having 100 mm or more rainfall in a day (possible flooding with crop failure) for weeks 1, 2, 3 & 4 for the month of August over the years. Year WEEK-1 WEEK-2 WEEK-3 WEEK

12 Total nonzero The frequency of possible crop failure shown in table (iv) indicate that over the past 57 years, there were 12 years when 100 mm or more fell in week 1, compared with 9 years in week 2, 16 years in week 3 and 11 years in week 4. 12

13 4. Text boxes showing the chance (proportion) of experiencing floods and crop failure during the month of August: Simple Models - Normal Distribution, One Sample TINt 'Flooding' Normal model, one sample Column Flooding Sample size 57 Minimum 0 Maximum 308 Range 308 Mean 214 Std. deviation Standard error of mean = with 56 d.f. 95% confidence interval for mean to From our sample of 57 years, the true mean date of the start of threatening flooding ( from 1 st August, 100mm or more of rain/ 1day) is estimated to be day 214(1.August) and is highly likely to be between day 193 (11.July) and day 234 (21.August). I am 95% confident that the margin error is ± 20 days (about 3 weeks). WEEK 1 CHANCE OF CROP FAILURE One proportion - binomial model BINomial;stats 57 12;SIMple;EXAct Binomial model, single sample Sample size 57 Successes 12 Proportion Approx s.e. of proportion = Exact results: 95% confidence interval for prop to Simple normal approximation: 95% confidence interval for prop to The 95% confidence interval for the true risk of flooding with crop failure for the first week in August is from 11.4% to 33.9%. Estimated value (proportion) = (21.1%) s.e. = (5.4%) The risk of crop failure during week 1 of August is (12/57): 21.05% or about 21% with a return period of (1/ = 4.75) about 5 years. I am 95% confident that the estimate of 21.1% is within ± 11.25% of the true risk of flooding with crop failure for the first week.. 13

14 WEEK 2 CHANCE OF CROP FAILURE One proportion - binomial model BINomial;stats 57 9;SIMple;EXAct Binomial model, single sample Sample size 57 Successes 9 Proportion Approx s.e. of proportion = Exact results: 95% confidence interval for prop to Simple normal approximation: 95% confidence interval for prop to The 95% confidence interval for the true risk of flooding with crop failure for the second week in August is from 7.5% to 27.9%. Estimated value (proportion) = (15.8%) s.e. = (4.8%) The risk of crop failure during week 2 of August is (9/57 = ): 15.79% or about 16% with a return period of (1/ = 6.333) about 6.5 years. I am 95% confident that the estimate of 15.8% is within ± 10.2% of the true risk of flooding with crop failure for the second week. WEEK 3 CHANCE OF CROP FAILURE One proportion - binomial model BINomial;stats 57 16;SIMple;EXAct Binomial model, single sample Sample size 57 Successes 16 Proportion Approx s.e. of proportion = Exact results: 95% confidence interval for prop to Simple normal approximation: 95% confidence interval for prop to The 95% confidence interval for the true risk of flooding with crop failure for the third week in August is from 17% to 41.5%. Estimated value (proportion) = (28.1%) s.e. = 0.06 (6%) 14

15 The risk of crop failure during week 3 of August is (16/57 = ): 28.07% or about 28% with a return period of (1/ = 3.56) about 3.5 years. I am 95% confident that the estimate of 28.1% is within ± 12.25% of the true risk of flooding with crop failure for the third week. WEEK 4 CHANCE OF CROP FAILURE One proportion - binomial model BINomial;stats 57 11;SIMple;EXAct Binomial model, single sample Sample size 57 Successes 11 Proportion Approx s.e. of proportion = Exact results: 95% confidence interval for prop to Simple normal approximation: 95% confidence interval for prop to The 95% confidence interval for the true risk of flooding with crop failure for the fourth week in August is from 10% to 31.9%. Estimated value (proportion) = (19.3%) s.e. = (5.2%) The risk of crop failure during week 4 of August is (11/57 = 0.193): 19.3% or about 19% with a return period of (1/0.193 = 5.2) about 5 years. I am 95% confident that the estimate of 19.3% is within ±10.95% of the true risk of flooding with crop failure for the fourth week..5. Boxplots for visual comparison, descriptive statistics as dates and Comment on the summaries: 15

16 Boxplots for visual comparison: Graph (iv): Boxplots showing the distribution of sowing and flooding dates ( day numbers) for the given definitions: From the above boxplots, it is obvious that the bold (1 st Mar) distribution presents a median date of sowing almost on day 76 (16.March). As a matter of fact ( boxplot stdr1lo, based on definition e ) the median occurrence of flooding in low lands is on day 221 (8.August). Boxplot of flooding with crop failure based on definition f shows that from 1 st August,100mm or more rainfall in exactly one day is highly likely to be between day 214 (1.August) and day 280 (6-October). From 6 years out of 57 years, the general risk of having this event is 10.53%, which is a 9.5 years return period. 16

17 Graph (v): Boxplots showing the distribution of rainfall in August for week 1, 2, 3 & 4 Graph (v) shows that week 3 and 1 have more variations in the rainfall totals for the week. Comparing week 3 and 1, it shows that week 3 has more variation than week 1 with almost the same amount of outliers. Descriptive statistics as dates: Conditions Table 5: Statistics summaries of sowing and flooding day numbers missin g count Mean SDE Min 5% 25% 40% Media n 60% 75% 95% Maximum 1 st Mar st Apr st May st Jun st Aug flooding

18 Table 6: Statistics summaries of sowing and flooding dates Conditions Minimum 25% Median 75% Maximum 1stMar(bold) 1.Mar 8.Mar 16.Mar 26.Mar 24.Apr 1 st Apr 1.Apr 3.Apr 7.Apr 14.Apr 29.Apr 1 st May 1.May 3.May 7.May 18.May 5.Jun 1 st Jun(latest) 1.Jun 4.Jun 9.Jun 16.Jun 29.Jun 1 st Aug(lowlands) 1.Aug 2.Aug 8.Aug 14.Aug 5.Oct 1 st Aug(flooding) 1.Aug 9.Aug 17.Aug 3.Sep 3.Nov Table 7: summaries of flooding risk with crop failure for each week in August August Week 1 Week 2 Week 3 Week 4 RETURN PERIODS ( of flooding with crop failure) 5 years 6.5 years 3.5 years 5 years Comment on the summaries: Graph (v) shows that week 3 and 1 have more variations in the rainfall totals for the week. Comparing week 3 and 1, it shows that week 3 has more variation than week 1 with almost the same amount of outliers. What is the distribution of sowing dates? In Douala, farmers always sowed before the start of the peak of rainfall when the soil is saturated, especially in August. So, the maximum or latest sowing date is the same for both farmers: day 214 (1.August). Earliest sowing date for bold farmer is day 61 (1 March) opposed to day 153 (1.June) for latest farmer. How wider is the season of the bold farmer compared to that of a latest farmer? 18

19 19

20 Difference in the length of growing season as experienced by bold and latest farmers Bold farmers Latest farmers Year Day_number Date Day_number Date Difference Mar Jun Apr Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Apr Jun Mar Jun Mar Jun Mar Jun Apr Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Apr Jun Mar Jun Apr Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Apr Jun Apr Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Apr Jun Mar Jun Mar Jun 75 20

21 Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun Mar Jun 91 Minimum Mean Range Std.deviation Maximum 57 years 39 days days 72 days days 111 days 5.5weeks 12weeks = 3months 10weeks 2weeks 16weeks = 4 months How often will the bold farmer have a longer growing season than the latest farmer? This question is answered by quantifying how often the season was longer for a bold farmer than for a latest farmer (conditional on differences being greater than zero). It will be found that the bold farmer who sows earlier in the season will always have the benefit of a longer season, which is a full 100% of the time. Talking about the length of the growing season, when we subtract starting date (as day number) of the latest farmer from that of the bold farmer we get 57 non-zero values, indicating that the former experienced a longer growing season every single year. On average his season was 85 days longer, at last 39 days longer and at most 111days longer, with a standard deviation of about 15.5 days. How far is the sowing season of the bold farmer compared to the occurrence of the flooding period? The maximum or latest sowing date for bold farmer is day 115 (24.April). Earliest flooding occurrence date is day 214 (1.August) according to our given definition. So, the bold strategy is at least 99 days (almost 3 months) before the first flooding strike which is almost 2.5 weeks (19 days) more variable, and completely off after 3 rd November (day 308). 21

22 F. IMPLICATIONS: Some conclusions for the client (Ministry of agriculture) based on these summaries and risks calculations. At last, we could say the highest true risk of experiencing flood with crop failure is observed once every 3.5 years during the third week of August. There is little difference between weeks 1 and 4 with a chance of flooding with crop disruption every 5 years. The smallest risk is observed in week 2 with a 6.5 years return period. 1. The sowing strategy definition that I would recommend to a farmer The sowing strategy definition that I would recommend to a farmer is the bold method. 2. The advantages The advantages could be: The greatest (range =54 days) and variable (s.d. =13 days) spread of sowing dates. The possibility to harvest more than one type of crops within this long sowing period. The possibility of choosing the accurate crop to match the expected season length, with higher yield. A crop growing during a shorter season (3 months) can be sow, with (57/57) 100 % risk or zero risk not to meet a long dry spell. As the threshold rainfall for sowing is high (40 mm), chance of crop failure is low even if there is long dry spell later. Hence, by sowing earlier, the risk of early crop failure due to flood could be greatly reduced even in low lands because after three months, the growing stage enables more plants maturity to face flooding events and the farmer can gain in confidence in the whole system of recommendations. 3. The disadvantages The disadvantages could be: Even with a better seasonal forecast, there still be at least once every three (3) years a chance of crop destruction in Douala, during the peak period (July, August, and September) of rainfall when the soil is saturated, especially in August. The flooding event in low land areas has a range of 65 days (about 2 months). There is highest interannual variability i.e. spread in flooding dates with a variability of 32 days (about 4.5 weeks mainly within the month of August). So maximum awareness must be taken around day 221(8.August) and also the farmer needs to keep himself in state of preparedness for this period so as to take action of the 1 st occasion to protect or to secure the crop). That is why I strongly recommend to the farmers dwelling in Douala and surrounding: Please! Sow earlier!. 22

23 G.APPENDICES: Appendix A: References and sources documentation Instat Climatic Guide Chapters 4-7 Roger Stern, Derk Rijks, Ian Dale, Joan Knock ed. January 2006 Confidence and significance: Key Concepts of Inferential Statistics, the University of Reading Statistical Services Centre, February 2006 Climatic e-books ( ) CAST for SADC ( Appendix: Daily data for Douala, Cameroon Fig. 1.a Time series plot of annual rainfall totals and number of rainy days at Douala, Cameroon, from 1950 to The dotted horizontal lines show one and two standard deviations above and below the mean. 23

24 PLOT OF ANNUAL NUMBER OF RAINY DAYS IN DOUALA;CAMEROON( ) Year The individual daily values or rainfall for every year, together with monthly summaries, are given in the Appendix. Fig. 2. Daily rainfall values for 1950 and

25 Figure 2 shows the pattern of rainfall through the year for two years. Notice the false start to the season in each year, where a few early days of rainfall are followed by a long spell of dry days. Fig. 3. Summary values of monthly rainfall PLOT OF THE SUMMARY STATISTICS AGAINST THE MONTHS IN DOUALA; CAMEROON gfedcb 1200 gfedcb gfedcb gfedcb 1000 gfedcb gfedcb X76 X77 X78 X79 X80 X Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec MONTHS 25

26 Table1: Summary values of monthly rainfall Column Min. 25% Mean Median 75% Max. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 3, Table 1 and 2 show some summary values for the monthly totals. July and August are seen to be the peak of the rainy season, with a lowest total of 207.4mm in July 2003 and one year (August1966) with over 1240mm. The rains are usually from January to December, though December, January and February are occasionally dry. Table 2: minimum, mean and maximum monthly rainfall summaries MONTH MINIMUM MEAN MAXIMUM January February March April May June July August September October November December

27 Table 2: Monthly rainfall totals in each year ( ). Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

28 Appendix: Daily data for Douala, Cameroon Daily data for: Yr1950 Mon Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Day

29 Daily data for: Yr1951 Mon Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Day

30 Day numbers in the year Month: Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Day First Last