Analysis of Precipitation and Temperature Variability over Central Africa (1901-2015)

Africa is already experiencing the impact of climate change. Some of the manifestations of climate change in Africa are, changing weather patterns resulting in, flooding and drought. Temperature change has impacted health, livelihoods, productivity of food, availability of water, and state of security. This study examines the long-term climate variations in Central African Countries (Gabon, Cameroon, Republic of Congo, Central Africa Republic, Chad and Democratic Republic of Congo) for the period 1901 to 2015, and then investigates the possible influence of increases in greenhouse gas con-centrations. To investigate climate patterns and trends in the Central African Countries, precipitation and temperature were analyzed on annual time scales using data collected from the World Bank Group Climate Change Knowledge Portal. Data was further aggregated using annual average blocks of 10 years. Linear and polynomial regression was performed. Also, linear time series slopes were analyzed to investigate the spatial and temporal trends of climate variability in Central African countries. Results of the analyses indicated that the mean annual temperature and precipitation records in some of the Central African Countries had both warming and cooling trends over the study period from 1901 to 2015. For perature rose according to a cubic model from about 24.5˚C to just below 27˚C during the period 1901-1940. This was followed by a brief drop between 1940 and 1960. From 1960 to 2015 it rose according to the model to almost 28˚C. By 2040 the temperature is expected to reach about 29.5˚C if this trend continues. Gabon was found to be the wettest country in Central Africa. Between 1901 and 1960, its average rainfall rose from about 144 mm to a maximum of approximately 160 mm. It had a general average rainfall/10 year’s increase from 1901 to 2015. The paper concludes by outlining policy recommendations in the form of improving national and regional environmental policies and regulations in the region, community involvement in decision making processes both at local and the national levels so as to contribute their input in the daily management of the forest resources, poverty alleviation in the region as well as building regional information system (RIS) incorporat-ing Geographic Information System (GIS), remote sensing and other environmental and socio-economic data to help reduce anthropogenic emissions of greenhouse gases.


Introduction
Climate is among important drivers of ecosystems' health, their compositions and other earth systems. Rainfall, atmospheric pressure, temperature and humidity are among the variables that control weather and climate respectively. Climate describes statistical variations of components of weather over time spanning from months to many centuries [1]. In the last two decades, there have been global concerns about the degradation of the world's ecosystems because of its attendant impact on greenhouse gases. The loss of biodiversity fueled by anthropogenic influences has been attributed in part to increased levels of greenhouse gases [2] [3] [4] [5] [6]. These changes in greenhouse gases and aerosols, taken together, are projected to lead to regional and global changes in temperature, precipitation, and other climate variables-resulting in global changes in soil moisture, an increase in global mean sea level, and prospects for more severe extreme high-temperature events, floods, and droughts in some places [5] [7]- [12]. Climate variations in temperatures and precipitations have been examined in different regions of the world over recent decades e.g. [13] [14] [15] [16] [17]. A study by Rosenbluth et al. [18] in Chile and Argentina reported variations in temperature due to influence of El Ninõ, Southern Oscillation on surface temperature along the Pacific South American coast 18˚S to 53˚S. Whitfield et al. [19] analyzed monthly temperature and precipitation time series for several hundred weather stations across Canada. Results showed variations in temperature and precipitation at different temporal and spatial scales. They further con-cluded that, differences observed in temperature and precipitation could have been the result of variations in atmospheric circulations.
Global changes especially warming could impact regional climates, especially rainfall patterns in the long-term. Local activities such as modification of land cover and disruption of local and regional hydrology could also impact climates [20]. Among the physical parameters of climate, temperature and rainfall play roles of paramount importance [21] since conditions of the environment in different places of regions impacting agricultural productivity depend on them [22]). The warming earth alters temperature, rainfall and other patterns of weather and climate. This poses threats to human health, plant and animal lives respectively. This paper examines rainfall and temperature data in central African countries in order to determine existing patterns in variation of the data.

The Study Area
Central Africa is a region of the African continent comprising of Angola, Burundi, Cameroon, the Central African Republic, Chad, the Democratic Republic of the Congo, the Republic of the Congo, Equatorial Guinea, Gabon, Rwanda, and São Tomé and Príncipe. This study focused on data from Democratic Republic of Congo (DRC), Republic of Congo, Cameroun, Chad and Gabon ( Figure  1). The area is home to the world's natural rainforests. As the region embarks upon an unprecedented phase of economic expansion, it faces several environmental challenges fueled partly by the pressures caused by human activities through economic activities, such as housing development, and road construction for transportation, economic development and demographic changes. In the last few decades, the region has been experiencing increasing resources degradation, fueling climate change [23]. Human activities inside these countries, for example, deforestation and degradation of forest resources, large scale and illegal mining, urbanization, slash and burn agriculture, resulting from shifting cultivation and uncontrolled exploitation to satisfy the needs of both the local and nonlocal populations are some of the major issues confronting the region. Rapidly growing population and poverty are stretching the resources of the basin beyond its carrying capacity.

Data Acquisition and Processing
For this study, temporal analysis of rainfall data from, the Democratic Republic of Congo, Cameroun, Gabon, Republic of Congo, Central African Republic and Chad were used. Rainfall and temperature data sets of 1950-2015 for the stated countries were downloaded from World Bank Group Climate Change Knowledge Portal for Development Practitioners and Policy Makers [24]. Before modeling, rainfall and temperature trends for each country, tables for 10-year blocks were made. Table 1 illustrates the data used to compute the average monthly rainfall for the years 1911 to 1920, which is shown as 124.44 mm in the table. All the rainfall data collected from January 1911 to December 1920 was summed up and divided by the total number of months, yielding rainfall per month per 10 years. Microsoft Excel spread sheet was used to compute the averages. This procedure was used to compute all entries in Table 2, for Democratic Republic of Congo. The same procedure was also carried out for the other central African countries to yield data illustrated in Table 3 and Table 4 for Cameroun and Gabon respectively.  To develop models that identify trends in rainfall and temperature data sets, aggregation of data using batches of 10 years was adopted for all the countries. In the modelling of the variations between variables in atmospheric data, regression was used.
The models for average rainfall and temperature based on data sets of 10 consecutive years were modeled based on the following respective formulations.

( )
Rainfall, R f t ∝ For cubic model, For quadratic model, For linear model, where, R is the average rainfall based on 10 consecutive years in mm/year, n is the polynomial index and t is time in years. A 1 , A 2 , A 3 and B are constants.
( ) For cubic model, For quadratic model, where, T is the average temperature based on 10 consecutive years in ˚C/year, n is the polynomial index and t is time in years. A 1 , A 2 and A 3 are constants.
Statistical software (SPSS, Microsoft Excel data analysis toolkit, etc.) was used to fit data and build models. The strength of model developed was based on the magnitude of the coefficients of correlations between variables (R square), which ranged from 0 to 1.0 (0% -100%).
A model was accepted if the corresponding coefficient of correlation was close to 100%. Linear and curvilinear modellings were applied depending on the correlations between data pairs, which attempts to develop single models for data sets spanning from 1901 to 2015 yielded models of low coefficients of correlations for most data sets. To develop models that were more representative of the data sets, illustrating the rises and falls in environmental data, data aggregates from time spans shorter than the entire data collection range of time  were used in modeling. Modeling for rainfall trends was carried out for, the Democratic Republic of Congo (DRC), Gabon and Cameroun since the difference between their maxima and minima rainfall data was at least 10%. Only Chad had a percentage change of temperature higher than 15%. Hence temperature change was only analyzed for Chad. The modeling processes and corresponding results for each of the countries are discussed as follows.

Results and Discussion
Democratic Republic of Congo (DRC)

Rainfall
The data used in modeling rainfall for the Democratic Republic of Congo (DRC) is presented in Table 2.
The difference between the maximum and the minimum rainfall data was 13 mm. The percentage rise from the minimum to maximum temperature was ((132.4 − 119.2)/119.2) × 100 = 11.04%. Regression analysis was used to model the Democratic Republic of Congo's average monthly rainfall data per consecutive 10 years versus years. The data from 1901 to 1940 was fitted into a third order polynomial equation. The general average rainfall trend from 1901 to 1940 is represented by a gradual rise from about 123 mm in 1901 to a maximum quantity of 126 mm in 1910 ( Figure 2). It then dropped gradually to a minimum value of 124 mm in 1930. It then gradually rose to 126.5 mm in 1940. The model is represented by the equation illustrated in Figure 2. This equation explains about 91% of the variation in average rainfall data for the specified span of time.
A polynomial function of order 3 was used to model rainfall data from 1940 to 1980 by curve fitting. The equation and the curve are given and illustrated in Figure 3.
The average rainfall/10 years dropped from about 127 mm in 1940 to a local minimum of 125 mm in 1945. It then rose to a local maximum of about132 mm in 1970, and then gradually dropped to a value between 125 mm and 126 mm in   1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 Average monthly rainfall mm/10 years  Figure 5 and Figure 6 respectively.

Cameroun
The difference between the maximum and the minimum rainfall data for Cameroun's data is 13 mm. Hence, the percentage rise from the minimum to maximum    Table 3.
The mean rainfall data per10 consecutive years was fitted to a cubic equation.
Where, y is the average rainfall based on every 10 consecutive years in mm/year and x is time in years.
This model suggested a decreasing trend for rainfall from 138 mm to 130 mm.
R squared for the cubic model was 0.4038; hence, it represents or explains about 40% of the data.
Better models for this data were developed by fitting the data for 1901-1960 to a quadratic polynomial using Microsoft Excel curve fitting tool. The model had R squared equal to 0.8465. Hence, the model explains about 85% of the data ( Figure 8).
The data for 1960-2015 was fitted to a third order polynomial model with R squared equal to 0.732 ( Figure 9). Hence the model; explains about 73% of the data.
Stronger models for the data for 1980-2015 were developed by modeling the data for the time periods of 1980-2000 and 2000-2015. The corresponding models are illustrated in Figure 10 and Figure 11 respectively.
The model for the period 2000-2015 ( Figure 11) had R squared equal to 1. Hence, it represented 100% of the variability in the data.
The model illustrates a rise from 127 mm to about 135mm in 2008, followed a drop to 130 mm by 2015. According to the model, Cameroun's average rainfall was expected to drop as years progressed beyond 2015.

Gabon
The difference between the maximum and the minimum rainfall data is 15 units. The data used in modeling rainfall Gabon is presented in Table 4.        1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Average monthly rainfall mm/10 years   Where, y is the average rainfall based on every 10 consecutive years in mm/year and x is time in years.

Chad Temperature
The difference between the maximum and the minimum average temperatures per ten-year data aggregates was 3.5˚C (5.3 F), which corresponded to a percentage rise equal to (4.5/24.5) = 18.36%.
The average monthly temperature for Chad per 10 years is presented in Table  5.   A correlation analysis for the average temperatures per 10 years data and was run, using SPSS 20 software. The correlation between the average temperatures per 10 consecutive years was found to be statistically significant (p = 0.006), with coefficient of correlation being equal to 71% (Table 6).
Microsoft Excel software was used to fit a cubic model the variation between the data sets using regression.     and . The corresponding cubic and quadratic models had R squared (coefficients of determination) equal to, 0.907 and 0.901 respectively. These models are illustrated in Figure 16 and Figure 17 respectively Between 1995 and 2015 the average temperature of chad rose sharply. According to the models illustrated in Figure 13 and Figure 15, the average temperature of Chad will likely rise significantly over future decades and possibly impact the environment negatively.
If the temperature continues to rise according to the model, Microsoft Excel statistical tool kit can be used to model its value by 1940 by entering 15 years in the forecasting tab. The temperature for the period 2015-2040 is illustrated in the following graph ( Figure 18). The temperature by 2040 is expected to be 24.5˚C.     The average rainfall of the Democratic Republic of Congo had a first general trend of rising, followed by a decrease to a minimum of about 123 mm. It then rose gradually to about 127 mm. The average rainfall for Cameroun dropped from about 138 mm in 1905 to 133 mm in 1960. It then rose to about 140 mm in 1980, followed by a drop to about 139mm. In general, Cameroun's rainfall data tended to decrease from 1905 to 2015 ( Figure 6). The average rainfall of Gabon increased to a maximum of approximately 160 mm according to the model illustrated in Figure 12 and then dropped to a minimum of about 150mm according to the model illustrated in Figure 13. However, in general, the rainfall increased ( Figure 11). The temperature of Chad was observed to rise from about 24.5˚C to about 28˚C from 1900 to 2015 according to a cubic model ( Figure 14). Variability in climate and anthropogenic activities have impacted Lake Chad, with the former accounting for about 34% to 45% during the time span 1972 to 2013 [25].

Policy Recommendations and Conclusion
Anthropogenic activities such as clearing of land for agriculture, industry, slash-and-burn agriculture and other human activities in the region have an impact on the concentration of greenhouse gases which ultimately leads to global warming. Perhaps, introduction of a policy which focuses on improving national and regional environmental policies and regulations in the region would go a long way to minimize some of the anthropogenic activities especially disforestation which are very rampant in the region [26]. Similarly, since most of these forest resources are spread out in the villages and rural areas, communities should be involved in decision making process both at local, regional and the national level, so as to contribute their input in the daily management of the forest and other natural resources. Any policy directed towards managing these forest resources from the top hierarchy level without community involvement is likely to fail. Local communities have played prominent roles in addressing pressing environmental issues so as to improve ecosystems. Hence, community based environmental protection should be included in environmental protection policies [27]. Also, in most African countries, environmental degradation is driven by poverty. Policies that deal with addressing poverty and raising the income levels as well as increasing the welfare of village communities in an eco-friendly manner will help alleviate environmental degradation in the region [28]. To reduce greenhouse gas emissions and forest management, there is the need to design and build a comprehensive Regional Information Systems (RIS) in the form of periodic inventorying, monitoring and evaluation with full support of the governments in the study area. RIS would entail combining remote sensing and geographic information system (GIS) data, climate data, field survey data, land use, forest cover and forest cover change into one system. Developing such a system will offer the decision makers access to the appropriate temporal-spatial data for monitoring the pressures mounted on the areas' ecosystems and socio-economic needs. Such a tool could act as an effective decision Y. A. Twumasi et al. support system in order to keep development in harmony with environmental sustainability, efficient land use planning and management of protected areas [29] [30].