Maximum and minimum temperatures time series of Congo-Brazzaville are analyzed for trend and discontinuities over the period 1932 to 2010. Temperatures series show an irregular increase. A total of 8 synoptic stations show positive trends in their annual mean maximum temperature series, and 7 of them are significant, with higher trends for urban stations. Annual mean minimum temperature showed 6 stations having positive trends. This increase is in relation with observations at regional scale. However, the differences are observed between large towns (Brazzaville and Pointe-Noire), and small or rural towns (Dolisie, Sibiti, Impfondo, Djambala). Trends in diurnal temperature range (DTR) are large positive trends in maximum temperature that are mainly observed in cities. The curve of DTR shows a decreasing trend which indicates the increasing of minimum temperatures. The effects of urbanization on temperature trends are investigated. Most stations regarded as urban stations are still useful for trend analysis; being situated on the suburban of the studied cities, they are therefore, not substantially influenced by the urban heat island.
This study assesses trends of minimum and maximum temperature in the Congo-Brazzaville from 1932 to 2010, a period for which reliable temperatures database are available. A large number of studies documented climate variations and changes across the 20th century in West Africa, e.g., [
The climate variables (temperature, rainfall, insolation, wind, etc.) are important components on agriculture development process. The rural economy in Congo relies essentially on agriculture, which in turn depends on weather and climate. In the same way climate components (e.g., air temperatures) play an important role in the water balance for the crops. However, successful agriculture yield requires suitable combinations of various factors like soil, crop, water, regime of temperature, etc. Therefore climate plays a crucial role in the growth and production of crops. It is then important to know the influence of climate parameters on plant growth and agriculture yield. The aim of analysing climate components (e.g., air temperatures) is to understand the effect of climate factors on agriculture. This analysis can provide global trend on temperature to detect the period of drought, during which air temperatures play a crucial role. The air temperature is the most important weather parameters because of physical and chemical for evolution within plants. Each crop has its own lethal maximum, minimum limits and the optimum temperature for its normal growth and reproduction.
The aim of this paper is to identify fluctuations and possible changes (trends or abrupt shifts) in the air temperature series across the country on annual and seasonal scales. In particular, one will examine whether the warming trend found in some parts of the region since the 1970s or 1980s in previous studies [
The Section 1 is the introduction; data have been presented and average temperature in Section 2. In Section 3, we describe the methods used. In Section 4, we describe the temperatures trends. The Section 5 is reserved for the discussion of the different resultants. In conclusion, we examine some environments impacts of the area study.
The data used in this study come from the archives of the “Direction Météorologie Nationale” of Congo-Braz- zaville. These dataset are extracted from the database of the Centre de Recherche et d’Etude sur l’Environne- ment (CR2E) of the Marien Ngouabi University. The database concerns the temperature of the air on a monthly time scale, from 1932 to 2010 of thirteen synoptic stations (
We examine the monthly temperature of 12 synoptic stations. Descriptive statistics for the annual and seasonal maximum and minimum temperature for the 12 stations are given in
. Stations used their altitude and statistical parameters (mean temperatures minimum and maximum) over the period 1932-2010
Altitude (m) | Period and years of site changes | Minimum | Maximum | |
---|---|---|---|---|
Impfondo | 326 | 1932-2004 | 20.7 | 30.6 |
Ouesso | 351 | 1932-2010 | 20.5 | 31.0 |
Souanké | 549 | 1951-2001 | 18.5 | 29.0 |
Makoua | 378 | 1956-1956/1956-1963/1963-2008 | 20.4 | 30.4 |
Gamboma | 376 | 1949-2010 | 21.0 | 31.1 |
Djambala | 789 | 1950-1957/1957-2010 | 18.4 | 27.5 |
Mpouya | 311 | 1941-2010 | 20.6 | 31.0 |
Brazzaville | 314 | 1932-1966/1966-2010 | 20.7 | 30.2 |
Mouyondzi | 511 | 1949-2010 | 19.4 | 28.1 |
Sibiti | 530 | 1950-2010 | 19.3 | 27.1 |
Dolisie | 330 | 1947-1957/1957-2010 | 20.9 | 29.2 |
Pointe-Noire | 16 | 1932-2010 | 22.1 | 28.3 |
Makabana | 161 | 1963-1998 | 19.5 | 32.8 |
Locations of the 13 synoptic stations in Congo-Brazzaville used in the study
Temperatures are transformed in absolute temperatures before their analysis. The annual and seasonal mean absolute temperatures at each station are standardized to obtain the annual and seasonal mean temperatures anomalies according to the relation:
where xij is the monthly or annual mean absolute temperatures for the station i in the year j, xi, the mean monthly or annual mean absolute temperatures for station i, and σi the standard deviation of annual totals.
Standardized time-series climate were used by many authors such as [
The 12 synoptic stations were subjected to two non-parametric Mann-Kendall and Pettitt tests to detecting trends and abrupt changes in time-series over the period 1932-2010. The principles of this test have been largely described by Sneyers [
The distribution of t, under the null hypothesis, is practically a normal distribution with the average and the variance given by the following Expressions (3) and (4):
(3) and (4)
The reduced statistics of the test, given by
Average monthly temperatures in the Congo-Brazzaville expressed in normalized anomalies (over 1932-2010)
The null hypothesis can, therefore, be rejected for high values of
The Mann-Kendall test consists in calculating two series of statistical values, one from the beginning of the series, another from the end; these series are shown in the form of two curves called respectively the direct curve (ui) and the retrograde curve (
Another
We consider xi, i = 1, 2, …, n, the terms of a continuous chronological series. For every t, t = 1, 2, …, n, we calculate the statistics:
Utn = 2Wt – t(n+1) with
Under the no hypothesis of the unpredictable character, the Wt are distributed according to a normal law with average and variance as:
(6) and (7)
The probability UE associated with the statistics is given by:
when we calculate the various statistics utn for t from 1 to n we can, in this series, look for the extreme values of utn. The critical value at the level of α = 0.05 is given by:
The value of t such as
Different sub-periods from 1932 to 2010 were compared to examine if there was any change in temperature trends throughout the period of investigation. A change was detected and dated. A comparison between the two sub-periods was carried out using the t-test (8). Trends are indicated to be significant in the discussions of results. It is only when the “t”-test is significant that the shift in the average climate is validated. In the t-test, if we suppose that two sub-samples are different the observed difference between the averages of two sub-samples is significant.
Here is the analytical expression of “t”-test:
where
Next section relating to the results our analyses are about on the determination of the trends.
Our analyses of monthly trends of all stations point out large temporal differences in trends between months for individual stations. The monthly minimum and maximum temperatures average have been standardized and plotted for every station (
The southern Congo-Brazzaville shows consecutive year periods with positives anomalies during the last year, whereas the northern have shown non-persistent mean annual minimum and maximum temperatures since the late 1970s. The major warming covered the whole Congo-Brazzaville. During the two last decades 1990s and 2000s, the mean annual minimum and maximum temperatures in all regions were above average.
The season DJF is characterized by the warm months in major stations of Congo-Brazzaville. The time series of the DJF mean seasonal minimum and maximum temperatures (
The season MAM is the period with higher warming months in the annual cycle at all stations over the country. It’s the first warming peak over annual. The time series of the MAM mean seasonal minimum temperatures (
The time series of the JJA mean seasonal minimum temperatures (
The SON season constituted the second peak of warming months. The minimum and maximum temperatures evolutions show a predominance of the positive anomalies during the last two decades (
Standardized time series of mean annual minimum and maximum temperatures (bars) and 5-year moving averages (lines) with coefficient of variation over the period
Standardized time series of mean seasonal DJF minimum and maximum temperatures (bars) and 5-year moving averages (lines) with coefficient of variation over the period
Standardized time series of mean seasonal MAM minimum and maximum temperatures (bars) and 5-year moving averages (lines) with coefficient of variation over the period
Standardized time series of mean seasonal JJA minimum and maximum temperatures (bars) and 5-year moving averages (lines) with coefficient of variation over the period
Standardized time series of mean seasonal SON minimum and maximum temperatures (bars) and 5-year moving averages (lines) with coefficient of variation over the period
With regard to what precedes, the evolution of the mean minimum and maximum temperatures is recorded in different ways over the country. This evolution is marked by predominance of the positive anomalies over the two last decades in the second half of the 20th century and the start decade for 21st century.
The non-parametric tests Mann-Kendall and Pettitt applied on the annual maximum and minimum temperatures series data showed an increase over major stations in Congo-Brazzaville (
It also is interesting to note that 5 of the stations showed significant positive trends. It is already established that most stations showed a significant increase in maximum and minimum annual temperature for the period 1932 to 2010, as shown in
Globally, the last decade of the 20th century was shown to be substantially warmer than previous decades. There is much difference between the trends for the two periods and the average trend for the latter last period which is slightly higher than for the former.
Trends in annual mean minimum and maximum temperature for the period 1932-2010 (significant trend at the 95% level)
We can conclude that there is gradual increase in annual mean temperature trends from 1960 to 2010. These results are consistent with those of Mpounza et al. [
To investigate if there was an abrupt increase in annual mean maximum and minimum temperatures on a shorter time scale. The time series was subject to a Student’s t-test for the differences in mean of annual mean temperatures between the period from 1932 to a specific year and the period after that year to 2010. The higher the absolute value of the test statistic, the larger the difference is deemed to be in mean temperature between the period before a specific year and the period after that year, for the number of years utilized for each period. The change in the t-test statistic is shown in
Student’s test applied on two sub-samples allowed us to quantify the average difference in trends of mean annual temperatures, between the periods 1930 to 1970/1980 and 1980 to 2010 (
. Test of Student on sub-periods defined by Mann-Kendall and Pettitt test; corresponding unlike Temperatures minimum between two sub-periods (over the period 1932-2010)
Stations | DJF | MAM | JJA | SON | Years | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | |
Impfondo | 1932-1966 1967-2004 | 20.7 20.2 | −0.5 | 1932-1940 1941-1966 1967-2004 | 22.2 21.0 21.2 | −1.2 +0.2 | 1932-1946 1947-2004 | 20.9 20.5 | −0.4 | 1932-1948 1949-2004 | 20.8 20.4 | −0.3 | 1932-1948 1949-2004 | 21.0 20.6 | −0.4 |
Ouesso | 1932-1940 1941-2010 | 20.3 19.9 | −0.4 | 1932-1980 1981-2010 | 20.8 21.3 | 0.5 | 1932-1972 1973-2010 | 20.5 20.4 | −0.1 | 1932-1942 1943-1950 1950-2010 | 20.4 21.4 20.3 | +1.0 −1.1 | |||
Souanké | 1951-1981 1982-2001 | 18.0 17.8 | −0.2 | 1951-1959 1960-1983 1984-2001 | 18.7 19.1 18.9 | +0.4 −0.2 | 1951-1959 1960-2001 | 18.4 18.7 | +0.3 | 1951-1959 1960-2001 | 18.6 18.7 | +0.1 | 1951-1958 1959-1978 1979-2001 | 18.3 18.7 18.5 | +0.4 −0.2 |
Gamboma | 1949-1977 1978-2010 | 21.4 21.8 | +0.4 | 1949-1979 1980-2010 | 20.8 21.2 | +0.4 | |||||||||
Djambala | 1950-1962 1963-2010 | 19.1 18.6 | −0.5 | 1950-1965 1966-2010 | 19.3 18.9 | −0.4 | 1950-1954 1955-2010 | 18.6 18.5 | -0.1 | 1950-1964 1965-2010 | 18.6 18.4 | −0.2 | |||
Mpouya | 1941-1987 1988-2010 | 20.9 21.4 | +0.5 | 1941-1981 1982-2010 | 21.2 21.8 | +0.6 | 1941-1985 1986-2010 | 19.5 20.4 | +0.9 | 1941-1981 1982-2010 | 20.7 21.3 | +0.6 | 1941-1984 1985-2010 | 20.6 21.2 | +0.6 |
Brazzaville | 1932-1980 1981-2010 | 21.1 22.3 | +1.3 | 1932-1976 1977-2010 | 21.2 22.4 | +1.2 | 1932-1980 1981-2010 | 17.7 19.9 | +2.1 | 1932-1972 1973-2010 | 20.4 21.7 | +1.3 | 1932-1988 1989-2010 | 20.3 21.7 | +1.4 |
Mouyondzi | 1949-1974 1975-2010 | 20.1 20.3 | +0.2 | 1949-2005 2006-2010 | 20.3 20.7 | +0.4 | 1949-1990 1991-2010 | 17.3 18.0 | +0.7 | 1949-2007 2008-2010 | 19.7 20.1 | +0.4 | 1949-1998 1999-2010 | 19.3 19.7 | +0.3 |
Sibiti | 1950-2007 2008-2010 | 20.3 20.0 | −0.2 | 1950-1966 1967-2010 | 19.3 19.4 | +0.1 | 1950-1970 1971-2010 | 19.1 19.3 | +0.2 | ||||||
Dolisie | 1947-1993 1994-2010 | 21.8 22.0 | +0.2 | 1947-1963 1964-2010 | 20.8 21.2 | +0.4 | 1950-1994 1995-2010 | 20.8 21.3 | +0.5 | ||||||
Pointe-Noire | 1932-2005 2006-2010 | 23.5 24.3 | +1.0 | 1932-2002 2003-2010 | 23.3 24.3 | +1.0 | 1932-1992 1993-2010 | 18.9 20.5 | +1.6 | 1932-2000 2001-2010 | 22.4 23.4 | +1.0 | 1932-2000 2001-2010 | 22.0 23.2 | +1.1 |
Makoua | 1956-1974 1975-1994 1995-2010 | 20.4 20.5 20.4 | +0.1 −0.1 | 1956-1996 1997-2010 | 21.0 20.9 | −0.1 | 1956-1973 1974-2010 | 20.3 20.2 | −0.1 |
. Test of Student on sub-periods defined by Mann-Kendall and Pettitt test; corresponding unlike Temperatures maximum between two sub-periods (over the period 1932-2010)
Stations | DJF | MAM | JJA | SON | Years | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | P | M(˚C) | D(˚C) | |
Impfondo | 1932-1942 1943-2004 | 30.3 31.3 | +0.9 | 1932-1940 1941-2004 | 30.4 32.0 | +1.6 | 1932-1946 1947-2004 | 29.4 29.6 | +0.2 | 1932-1944 1945-2004 | 29.5 30.0 | +0.5 | 1932-1942 1943-2004 | 30.0 30.7 | +0.7 |
Ouesso | 1932-1971 1972-2010 | 32.4 31.1 | −1.3 | 1932-1963 1964-2010 | 32.5 31.7 | −0.8 | 1932-1968 1969-2010 | 30.9 30.2 | −0.7 | 1932-1944 1945-2010 | 32.2 30.7 | −1.5 | |||
Souanké | 1951-1976 1977-2001 | 28.9 29.7 | +0.8 | 1951-1977 1978-2001 | 30.0 30.7 | +0.7 | 1951-1971 1972-2001 | 27.2 28.0 | +0.8 | 1951-1976 1977-2001 | 28.4 29.0 | +0.6 | 1951-1974 1975-2001 | 28.6 29.3 | +0.7 |
Gamboma | 1949-1955 1956-2010 | 31.5 31.1 | −0.4 | ||||||||||||
Djambala | 1950-1980 1981-2010 | 27.4 28.0 | +0.6 | 1950-1958 1959-2010 | 26.7 27.0 | +0.3 | 1950-1982 1983-2010 | 27.3 27.8 | +0.5 | ||||||
Mpouya | 1941-1987 1988-2010 | 30.8 32.1 | +1.3 | 1941-2006 2007-2010 | 31.9 33.0 | +1.2 | 1941-2005 2006-2010 | 30.7 32.1 | +1.4 | 1941-2005 2006-2010 | 31.1 32.3 | +1.2 | |||
Brazzaville | 1932-1991 1992-2010 | 30.7 31.0 | +0.3 | 1932-1991 1992-2010 | 31.4 32.1 | +0.7 | 1932-1994 1995-2010 | 27.9 28.6 | +0.8 | 1932-1986 1987-2010 | 30.3 31.0 | +0.7 | 1932-1976 1977-2010 | 30.0 30.6 | +0.6 |
Mouyondzi | 1949-1956 1957-2010 | 30.2 29.4 | −0.8 | 1949-2000 2001-2010 | 28.3 28.7 | +0.4 | |||||||||
Sibiti | 1950-1982 1983-2010 | 23.9 24.4 | +0.5 | 1950-2003 2004-2010 | 27.1 28.1 | +1.0 | |||||||||
Dolisie | 1947-2000 2001-2010 | 30.2 31.0 | +0.8 | 1947-2002 2003-2010 | 30.8 31.9 | +1.1 | 1947-1999 2000-2010 | 26.5 27.9 | +1.4 | 1947-1996 1997-2010 | 29.0 30.1 | +1.1 | 1947-1999 2000-2010 | 29.1 30.2 | +1.1 |
Pointe-Noire | 1932-1944 1945-2010 | 29.5 29.8 | +0.3 | 1932-2000 2001-2010 | 25.7 26.3 | +0.6 | 1932-2006 2007-2010 | 28.3 29.4 | +1.1 | ||||||
Makoua | 1956-1976 1977-2010 | Ns | 1956-1964 1965-2010 | 30.6 30.8 | +0.2 | 1956-2000 2001-2010 | 31.5 31.9 | +0.4 | 1956-2000 2001-2010 | 29.2 29.6 | +0.4 | 1956-1961 1962-1994 1995-2010 | 30.5 30.4 30.8 | −0.1 +0.4 |
Annual mean minimum temperatures know the first increase after period 1970 and after 1970 (between +0.2˚C and +1.4˚C). However, trends of annual mean minimum temperature showed clear and significant changes, for comparison with the results of this study. Maximum temperatures increased between +0.5˚C and +1.5˚C.
Temperatures increase is more important in urban (big Town) stations than non-urban (small Town). For urban stations the minimum temperatures have increase trend more +1.0˚C on average and for non-urban stations a trend of under +1.0˚C.
The results of the above seasons are shown in
The non-parametric Mann-Kendall trend test results, shows a significant trend in the seasonal minimum and maximum temperatures average in the Congo-Brazzaville over the period 1932-2010. The
The first and second sub-periods correspond respectively to the temperature decrease and increase period. The seasonal temperatures decrease between two sub-periods reaching up to −0.2˚C at −1.2˚C. The seasonal temperatures from March to May in the southern stations point out significant increase (Pointe-Noire, Dolisie, Brazzaville); whereas the northern stations (Impfondo, Ouesso) show a decrease (
Trends in seasonal DJF mean minimum and maximum temperature for the period 1932-2010 (significant trend at the 95% level)
Trends in seasonal MAM mean minimum and maximum temperature for the period 1932-2010 (significant trend at the 95% level)
Trends in seasonal JJA mean minimum and maximum temperature for the period 1932-2010 (significant trend at the 95% level)
Trends in seasonal SON mean minimum and maximum temperature for the period 1932-2010 (significant trend at the 95% level)
From the t-test student the decline was compared between two sub-periods revealed by Mann-Kendall (
The global trend is for the DTR trend to be negative over the last half century due to the fact that, trends of minimum temperatures are higher than trends in maximum temperature (Figures 13-14). Four stations showed positive DTR trends, with five of them significant, and six stations showed negative trends. The stations in the southern showed high positive trends due in part to the high trends in annual mean maximum temperature, also because of the small negative trends in annual mean minimum temperatures for some of the stations. Other stations with the same tendency are situated in the northern, but their trends in DTR were much smaller, although significant for Mpouya station. Most some of these stations with negative or positive trends showed a substantial significant increase in minimum temperatures, whatever the trends in maximum temperatures. The results obtained are similar with those of the other studies on the in country Africa [
In this section we compare the temperature trends of the urban stations with the rest of stations in small cities. The stations that knew substantial increase of urbanization in their vicinities during the study period are Brazzaville, Pointe-Noire and Dolisie. All these stations of big town knew an increase of the trends maximum and minimum temperatures more than +1.0˚C. Although, the stations in small towns in the forest zone (Sibiti, Impfondo, Makoua), show a increasing of trends trends maximum and minimum temperatures. By contrast, the trends maximum and minimum temperatures of Brazzaville, Pointe-Noire and Dolisie, differ strongly to other stations, which are located in the small towns. This is due certainly by the effect of urbanization. These stations show a significant increase in mean annual minimum temperature more than +1.0˚C compared with +0.2˚C to +0.7˚C, in the small city.
In the average annual minimum temperature trends, the urban stations (Brazzaville, Pointe-Noire) show, on average, a significant positive trend of +1.1˚C and +1.4˚C compared with a significant +0.4˚C or +0.6˚C for non-urban stations (Impfondo, Sibiti, Mouyondzi); for average annual maximum temperature trends is a significant +0.6˚C, at +1.1˚C (urban) compared with a significant more than +1.1˚C (non-urban).
By splitting urban stations and non-urban, in general the average minimum annual temperatures trends of urban stations showed more positive trend than non-urban stations. This means that there might be a strong increase in minimum temperatures compared with maximum temperatures for urban stations than the non-urban stations.
The reason for the difference in trends seems to be the increase of the mean minimum temperature for urban stations than the non-urban stations, although the differences in mean maximum temperature between urban and non-urban station also seem to play a role to a certain point.
This study documents minimum and maximum temperature in Congo-Brazzaville, over the west part of Atlantic Equatorial Africa. The results obtained are similar with those of the other studies on the in country Africa [
Diurnal temperature range (DTR) for the period 1932-2010 (significant trend at the 95% level)
Trends diurnal temperature range (DTR) for the period 1932-2010 (significant trend at the 95% level)
those in the same non-urban stations. The maximum and minimum temperature evolution of Congo corresponds to that of the intertropical convergence zone (ITCZ) as published by the WMO. The years 80’s were less warm than the years 90’s. Nevertheless two periods clearly individualize the recent temperature evolution of Congo: before 1970 and after. The first period shows a temperature decrease or low temperatures. The second period is very uniform on the totality of the country and is marked by warmer temperatures. Trends in temperature range also show that results on a regional basis coincide with the general global trends, which have been positive over the last century [
In this evolution of temperatures in Congo, there are urban zones which have known the most important warming, without any doubt linked to the phenomenon of urbanization. But the even generalized rising in weak urbanization zones seems to be a sign of sensitivity space of Congo to the global climate change.
However, there does not seem to be much spatial coherence in the results of temperature trends for other seasons. Temperature trends were also found not to be consistent between seasons, with MAM being the season of highest temperature trends on average and DJF being the season of lowest trends. DTR trends mean annual minimum and maximum temperature trends; many stations showed positive trends.
The conclusions are as follows:
• It is possible to establish a general increase of temperatures for all stations, and the annual temperature is characterised by a rise of temperatures quasi generalized to all stations; all series show a relative precipitation amount in the course of decades 1970 and 1990 compared to other decades. However, temperatures seem to begin a rise from the 1990s.
• It was found that there was rather a relatively increase minimum temperatures in the early 1970s, this being the main cause of the general increase average temperatures over the whole period from 1932 to 2010.
The authors wish to thank the Direction de la Météorologie nationale du Congo-Brazzaville for providing the data used in this study.