Analysis of temperature, rainfall and their projection in Rwanda's highland. Case study: Gicumbi

Abstract

Climate variability and change are some of the major challenges the world is facing today, particularly in African countries where the ability to cope with this challenge is also another challenge. As scientists we need to keep track of climate variability and change in order to inform Policymakers. The main objective of this study was to determine the variability, trend, and future changes in temperature and rainfall in the Gicumbi district. Observation data used in this study was collected from Rwanda Meteorological Agency (Meteo-Rwanda) and was used with simulated data of Regional Climate Model(RCM) REMO2015 driven by NCC-NorESM1-M from Coordinated Regional Downscaling Experiment(CORDEX). In this study, graphical and statistical methods have been employed to achieve our goal where temperature and rainfall were plotted at different time scales, seasonal and annual. Mann Kendall a non-parametric trend test was employed to determine the trend of the above climate variabilities and in the attempt to determine variability patterns, the coefficientofvariabilityandprecipitationconcentrationindex(PCI)wascalculatedfortheperiodstartingfrom 1983to2021. Lastly,theprojectedchangewasfoundbycomparingthemeanofabaseperiodof1976to2005 and the future projection 2024-2053. The result from the Mann Kendal trend testhasrevealed thatthere isa trend but not significant inrainfall atall stations and all months, seasons, and annual except January at Murindi station which is -1.130 mm/year. The minimum, maximum, and mean temperatures have shown a significant trend in some stations and at a given timescale. Highvariabilityhasalsobeenrevealedintherainfall. Thechangeinaveragerainfallwascomputed and revealed that the JJA season has the highest change which is about 100%, The results also from Mann Kendal test of minimum, mean, and maximum temperature show that there is a mixed positive and negative trend, but the overall shows that it temperature of this are was increased at all stations. The study’s results indicate that there is a notable degree of monthly and seasonal variability in rainfall, falling within the category of moderate to high variability. However, when compared to the monthly and seasonal variations, the annual variability is relatively small, falling within the range of moderate to low variability. These findings highlight that the coefficient of variation (CV) for rainfall is typically at its highest during the drymonths,specificallyfromJunetoAugust(JJA)andtheircorrespondingseason. Incontrast,itisatitslowest during the wet months, spanning from September to May, including their corresponding seasons (MAM and SOND). This difference is attributed to the more even distribution of rainfall during the wet months, while the drymonthsexperienceconcentratedheavyrainfallevents. Moreover,thedatasuggeststhattheCVofrainfallis generallyhigherattheKaramboandRweserostationscomparedtotheByumba,Murindi,andKabezastations. Based on the visual representation, it became evident that there existed year-to-year monthly, seasonal, and annual fluctuations in rainfall. The highest recorded rainfall in this region reached 2072 mm/year, observed at Karambo station in 2012, which was attributed to an unusual surge in rainfall during September, October, and November. In contrast, the lowest recorded rainfall was 350 mm/year in 2017. The data also suggested v that in 2017, there was a noticeable deficiency in rainfall, possibly influenced by climate change. Among the five stations, three exhibited a slight declining pattern, with rates of approximately −1.89 mm/year, −7.03 mm/year, and −0.82 mm/year at Karambo, Murindi, and Kabeza, respectively. Conversely, Rwesero and Byumba stations displayed an increasing trend, with magnitudes of around 2.82 mm/year and 2.98 mm/year, respectively. Theoverallannualtrendwascalculatedtobeapproximately−0.79 mm/year,whichisdeemedas not statistically significant. ItwasdeterminedthatKabezahadthelowestaverageannualminimumtemperature,whichmeasured12.74◦C, withastandarddeviationof 0.55◦C,resultinginacoefficientofvariabilityof 4.32%. Incontrast,Rweserowas foundtohavethehighestaverageminimumtemperature,at 15.20◦C,withastandarddeviationof 0.48◦C,correspondingtoacoefficientofvariabilityof 3.13%. Theanalysisalsoindicatedthatthecoefficientofvariability for minimum temperature remained consistently low, ranging from 3% to 7% throughout all months. Furthermore,itwasobservedthatinJune,Byumbahadthelowestaveragemeantemperature,recording16.76◦C,with a standard deviation of 0.66◦C, resulting in a coefficient of variability of 3.94%. On the other hand, Rwesero displayedthehighestaveragemeantemperature, whichreached 16.76◦C, withastandarddeviationof 0.78◦C, leading to a coefficient of variability of 3.60%. The coefficient of variability for mean temperature remained relatively small, ranging from 2.4% to 3.96% in all months, and at the seasonal scale, the annual coefficient of variability also remained low, spanning from 1.5% to 2.83%. It was also observed that Byumba had the lowest average maximum temperature in May, registering at 20.78◦C, with a standard deviation of 0.88◦C, resulting in a coefficient of variability of 4.21%. Conversely, Rwesero recorded the highest average maximum temperature, at 27.56◦C, with a standard deviation of 1◦C, and a coefficient of variability of 3.64%. The coefficient of variability for maximum temperature exhibited a relatively small range, spanning from 2.36% to 4.21% across all months. At the seasonal scale, the annual coefficient of variability also remained low, ranging from 2.59% to 4.24%. These findings collectively indicate that the minimum, mean, and maximum temperatures in this region exhibit relatively low variability. Itcanbededucedthattheaverageminimum,mean,andmaximumtemperatureinthisareadisplayedapositive significant trend in some stations with an estimated average rate of around 0.28◦ C/decade, 0.2◦ C/decade, 0.1◦C/decade respectively. From visual representation, It was also shown that the inter-annual variability of minimum,mean,andmaximumtemperaturewasnothighinthisareaandtwostations(RweseroandKarambo ) registered a higher temperature than the remaining stations(Byumba, Murindi, and Kabeza), This is because those two stations are located in the region near eastern and central plateau which are among hotter regions in the country. In Gicumbi district, future rainfall projections are derived from bias-corrected data using a linear scaling approach. During the JF season, rainfall is expected to increase at all stations, ranging from 2.2% (4.63 mm) to 8.3% (10.94 mm) under RCP 2.6, while a decrease is projected, ranging from 0.89% (-1.2 mm) to 7.9% (-16.48 mm) under RCP 8.5. For the MAM season, all stations are anticipated to decrease, with declines from 3.7% (-16.7 mm) to 12% (-55.5 mm) under RCP 2.6 and 2.82% (-10.2 mm) to 10% (-54 mm) under RCP 8.5. JJA brings a substantial increase, from 27.54% (23.4 mm) to 95.36% (107.4 mm) under RCP 2.6, along with a moderate decrease from 8% (-6.88 mm) to 12% (-6.87 mm) at most stations. In SOND, three out of five stations are expected to increase, while two are projected to decrease under RCP 8.5. In the annual context, stations show increases under RCP 2.6 and decreases under RCP 8.5 for the period of 2024-2053. Minimum temperature projections for Gicumbi district utilize bias-corrected data. During the JF season, an increaseinminimumtemperatureisexpectedatallstations,rangingfrom6.97%to19.42%underRCP2.6and 8.91% to 10.71% under RCP 8.5. In the MAM season, increases range from 7.78% to 23.73% under RCP 2.6 vi and9.18%to11.25%underRCP8.5. ForJJA,stationsexperienceincreasesfrom2.29%to21.76%underRCP 2.6, with a moderate rise from 11.26% to 11.68% under RCP 8.5. During SOND, most stations increase, with Byumba,Mulindi,andKabezastationsshowingsubstantialincreases,whileRweseroandKarambostationsare expected to decrease under RCP 8.5. In the annual context, all stations increase under RCP 2.6 and decrease under RCP 8.5. Future projections for mean temperature in Gicumbi district employ bias-corrected data with linear scaling. During the JF season, an increase in mean temperature is expected at all stations, with increases ranging from 6.97% to 19.42% under RCP 2.6 and 8.91% to 10.71% under RCP 8.5. In the MAM season, stations are projected to experience increases from 7.78% to 23.73% under RCP 2.6 and 9.18% to 11.25% under RCP 8.5. JJA shows increases ranging from 2.29% to 21.76% under RCP 2.6, with a moderate rise from 11.26% to 11.68% under RCP 8.5. During SOND, most stations indicate increases, with Byumba, Mulindi, and Kabeza stations showing substantial increases, while Rwesero and Karambo stations are expected to decrease. In the annual context, all stations show increases under RCP 2.6 and decreases under RCP 8.5 for the period of 2024-2053. Maximum temperature projections in Gicumbi district are based on bias-corrected data using linear scaling. During the JF season, an increase in maximum temperature is anticipated at all stations, ranging from 5.19% to 6.54% under RCP 2.6 and 6.28% to 8% under RCP 8.5. In the MAM season, all stations are expected to experience increases, ranging from 5.82% to 7.32% under RCP 2.6 and 7.95% to 9.91% under RCP 8.5. JJA showsincreasesrangingfrom4.4%to5.10%underRCP2.6,withamoderaterisefrom6.03%to6.99%under RCP 8.5. In SOND and annually, stations are projected to show increases under both RCP 2.6 and RCP 8.5, with varying percentages.