Tuesday, August 15, 2017

Climate Change Impact - Part 12 - Kagera Basin (Rwanda, Burundi, Uganda and Tanzania)

Climate Change Impact

Part 12: Example – Kagera Basin (Rwanda, Burundi, Uganda and Tanzania)


The Kagera basin flows into Lake Victoria and as such it forms part of the Nile Basin. An extensive data base of climate and flows was available and was used to calibrate the HYSIM hydrological model to 22 sub-basins. Climate projections show that rainfall is projected to increase but temperature (and hence evapotranspiration) is also expected to increase. Whilst the two changes to some extent balance out, the increase in temperature still has important implications for the future of agriculture.


The Kagera River Basin and its tributaries flow within four countries (Rwanda, Burundi, Uganda and Tanzania). The Kagera River flows into Lake Victoria which in turn forms part of the Nile River Basin. The aim of the study was to assess the water resources potential of the basin and also to estimate the potential effect of climate change.

Figure 1 is a map of the river basin with the sub-basins used for hydrological modelling delineated.

Figure 1 Kagera River basin showing sub-basins for hydrological modelling

Current situation

The project team was provided with a climate and hydrometric data base developed in the Lake Victoria Environmental Management Program, Phase I. This included river flow, precipitation, temperature and other variables such as wind speed, sunshine and relative humidity needed to calculate potential evapotranspiration. Data were available up to the year 2000.

The methodology was based on use of a hydrological model, HYSIM (in its monthly variant). A hydrological model requires continuous input data so gaps in the data had to be infilled by reference to nearby stations with data.

Temporal infilling is standard option from the HYSIM program.  The program does this as follows:

  •         Reads the data into a monthly array of data from all stations.  If there are more than a set number of days with data, then the total for those days is adjusted pro-rata upwards to give the month's total (in the case of precipitation) or monthly mean, in the case of other variables.
  •         Calculates a matrix of totals/means for concurrent periods for all pairs of stations.
  •         Uses the above totals to calculate the ratio of the totals/mean for all pairs of stations.
  •         Uses these ratios to estimate the errors in the relationship between all possible pairs of stations.
  •         Infills the monthly data using whichever station with data gives the lowest error and which has not itself been infilled.

Even when all stations with data had been infilled there were still large parts of the basin without measurements. To complement the observed values, data on a 10’ geographical grid from the Climatic Research Units of the University of East Anglia were also used.

The data were used to calculate potential evapotranspiration and precipitation on a 10’ grid. The following figure shows contours of rainfall minus potential evapotranspiration (PET). As can be seen the western parts of the basin are in surplus but those to the east are in deficit.

Figure 2 Precipitation minus potential evapotranspiration

A similar process was used for other climatic parameters.

The data base of rainfall and precipitation were used with the HYSIM hydrological model to simulate river flows in the basin. Flows were simulated for each of the 22 sub-catchments that are shown on figure 1.

The following figure shows the simulated and observed flows for the Rivubu at Gitega.

Figure 3 Simulated and observed flows - River Rivubu at Gitega

For most of the time the simulation is good however the accuracy drops off markedly after the early 1990s. Whilst the data base is good up to the end of the 1980s the number of climate stations becomes much fewer for the later years and this can be considered to explain the change in accuracy.

Climate change

The A1B scenario has a balanced emphasis on all energy sources. It is generally considered to be the projection of temperature that might occur in the absence of any international agreed and binding protocol to reduce carbon emissions. This scenario was therefore used.

The choice of models was based on those listed on Table 6 of the IPCC “General Guidelines on the Use of Scenario Data for Climate Impact and Adaptation Assessment”, Version 2, June 2007.  Seven models were listed. The projections were based on the average of those models.

The following figure shows the current average annual temperature in the basin and the projected temperature for two different time horizons 2020 to 2049 and 2070 to 2099.

Figure 4 - Observed and projected basin temperature fro two time horizons

For the period 1970 to 1999 the average basin precipitation was 1570 mm/year. For the period 2020 to 2049 it is projected to be 1644 mm/year and the period 2070 to 2099 is projected to be 1740 mm/year.

The overall conclusion was that the increase in precipitation would counter-balance the increased potential evapotranspiration at a basin level but for individual sub-catchments the picture was more complex. Even though the overall water balance might be the same the higher temperature had implication for the types of crops which could be grown and the amount of irrigation required.

Monday, August 14, 2017

Climate Change Impact - Part 11 - Turkey

Climate Change Impact

Part 11: Example –Turkey


The Yesilirmak Basin in northern Turkey drains into the Black Sea.  The basin is currently highly developed for hydropower and irrigation. The impact of climate change will be to reduce annual flow. Currently snow melt in late spring provides water at the start of the irrigation season; critically, the biggest reduction in flow will be during this period. To maintain current levels of irrigation will require additional storage. 


The Ye┼čilirmak basin in northern Turkey has a drainage area of around 36,000 km2 and flows into the Black Sea. The basin is mountainous with parts of the basin reaching elevations in excess of 2,000m. The aim of the study was to develop an understanding of the water balance of the basin and then to examine the potential effects of climate change on the basin. The basin is heavily developed, mainly for irrigation and hydropower.

Figure 1 Yesilirmak River Basin

Current situation

Meteorological data were available from two sources. The first source was supplied locally and covered the period 1961 to 2007. The second was from an internationally supported internet site which provided data from 1931 to 2006. There was considerable overlap in data from the two sources and we were able to create a consistent data set from 1931 to 2007. Annual precipitation reaches 1000mm/annum near the coast and is less than 500 mm/annum inland. Average annual temperature is around 14 °C to the north of basin and less than 10 °C in higher areas inland.

From 1931 to 2007 the data showed little trend in annual precipitation or temperature.

Figure 2 Monthly precipitation - Yesilirmak basin
 For both precipitation and temperature there were differences in seasonal values. Winter precipitation and spring temperature both showed increases.

Figure 3 Seasonal temperature - Yesilirmak basin

Reservoirs in the basin play an important part in flow regulation. The total storage is over 5,000 million m3. This is close to the annual average flow of 5,500 million m3/annum. The calculation of the effect of reservoirs was complicated by three factors:

  •         Little information was available on their operating rules.
  •         There are transfers of water between basins
  •         When a reservoir is newly constructed water is used to fill the reservoirs.

Annual irrigation use is 500 million m3/annum and is, of course, concentrated in the summer months when flows are at their lowest.

Climate change

The basin was modelled in two ways. First a water resources model was developed which simulated as far as possible the progressive changes in water use and reservoir storage over the period with data. This model was used to estimate the natural flow in the basin. Secondly the monthly rainfall/runoff model HYSIMM was used to simulate the flows in the basin.

The climate projections were based on the IPCC task force on Data and Scenario Support 2007 report “General Guidelines on the Use of Scenario Data for Climate Impact and Adaptation Assessment”. The SresA1B scenario, which assumes rapid growth and a balanced use of energy sources, was used

As an initial appraisal of the potential impacts of climate change, the simulated flow for the period 1980 to 1999 was compared with simulated flow for projected values for the period 2080 to 2099, i.e. 100 years in the future relative to observed.

The following chart shows three lines. The blue line is the observed average monthly flow – or, more accurately, the estimated natural flows after accounting for storage and abstractions. The red line shows the flow simulated by the hydrological model. The third line, in green shows the projected monthly flow after the impact of climate change. The impact of climate change is represented by the difference between the red and green lines.

Figure 4 Monthly flows - with and without climate change

This shows that two importance changes are projected:

  •         Overall flows will be lower,
  •         Currently the peak of flow is in late spring which coincides with the start of the growing season. In effect, snow in the mountains is acting as a reservoir. In the future, the peak flow will be in winter.

The conclusion is that to maintain current levels of agriculture additional reservoir storage will be needed.

Climate Change Impact - Part 10 - Zambia

Climate Change Impact

Part 10: Example –  Zambia


Zambia has a climate typical of Southern Africa with cool dry winters (June to August) with hot wet summers. The potential impact of climate change was studied as part of a project enhancing the country's skills in Integrated Water Resources Management (IWRM). Temperature are expected to rise throughout the year and over the whole country with slightly higher increases in the south-east. The changes in precipitation are less consistent with some months projected to have an increase and others to have a reduction.


The Government of Zambia was fully aware of the principles of Integrated Water Resources Management.  The need for climate change to integrated in water resources planning was recognised in the National Water Plan of 1994. The Water Resources Management Act of 2011 took this a stage further. This act sought to create a National Water Authority. Section 8 of the Act on ‘Functions of the Authority’ said its functions should include:

  •         minimising the effects of climate change
  •         support proactive climate change planning and management
  •         in consultation with the institution responsible for national statistics, establish and maintain an information system, which will be accessible by both gender, in accordance with regulations issued by the Minister providing for the content of the system, which shall include relevant hydrological, hydrogeological, meteorological, climatological, water quality, water storage and supply and use data, and relevant information on potentials for the use of water
  •         publish forecasts, projections and information on water resources

The aim of this project was to assist the Government in the integration of climate change into Integrated Water Resources Management.

The first, and at the time of the project only, National Communication on Climate Change prepared under the auspices of the United Nations Framework Convention on Climate Change was produced by the Ministry of Tourism and Natural Resources (MoTNR) in 2002. The report considered mitigation options. Under the heading ‘Vulnerability and Adaptation Assessment” it was estimated that maize production might fall but sorghum could increase and groundnuts remain steady. There was no clear indication of the effect on livestock. In terms of water resources, it was suggested that southern parts of the country might be particularly vulnerable.

A report on the National Adaptation Programme of Action on Climate Change was produced by the MoTNR. Zambia had experienced a number of climate related hazards over several decades. Using multi-criteria analysis, it had identified most urgent needs to prioritize ten immediate adaptation interventions. Zambia was divided into 3 ecological-climatic regions based on rainfall. The wettest regions were toward the north of the country. According to the report, the projections suggested that the wettest region would have an increase in rainfall but the drier regions would have less rainfall. The driest region is projected to produce less agricultural produce and livestock. Wildlife could be heavily stressed due to reduced rainfall and increased migration. Malaria is likely to increase in areas with increasing rainfall.

Current Climate

The average annual temperatures for four stations are shown on figure 1. All four stations show a similar trend: a maximum around 1930, a general fall until about 1975 and then an increase to a new maximum around 2005. Temperatures during the 5 years 1927 to 1931 were about 0.5 °C higher than temperatures from 2001 to 2004.

Figure 1 Average annual temperature in Zambia - four representative stations

Temperatures are lowest in June and July. In terms of geographical distribution of temperatures, they are highest in the south-east and the north though the variation is not great – most of the is in the range 22°C to 24°C.

Figure 2 shows the seasonal distribution of precipitation. It shows that rainfall is highly seasonal with very little rain in the period June to August.

Figure 2 Average monthly precipitation - three representative stations

There is considerable variation in rainfall from year to year. The wettest station showed a slight increasing trend and the driest station showed a slight decreasing trend. The geographical distribution of rainfall showed it as being higher to the north of the country.

Climate change projections

Climate change projections were based on the A1B scenario. This is considered to be the ‘business-as-usual’ scenario. The projection used was the average of 23 climate models used to inform the IPCC Assessment Report.
Temperatures are expected to increase by from 3.2°C to 3.9°C by the end of the century. Figure 3 shows the geographical distribution of the temperature changes.

Figure 3 Geographical distribution of climate change

In the case of temperature, the increases are fairly uniform throughout the year. In the case of precipitation there is a marked difference in the changes at different times of the year.

Figure 4 Projected change in precipitation

This shows that rainfall will decrease in the currently driest periods of the year and will increase most in the wettest periods.


Temperatures are projected to rise throughout the year and over the whole country with slightly higher increases in the south-east. In the case of precipitation there are seasonal variations in the changes. In January, there are increases in precipitation over the whole country but larger increases in the north. In November, precipitation is expected to fall over the whole country with larger falls in the south.  December could be considered a ‘pivot’ month with increases in the North and reductions in the South.

Climate Change Impact - Part 9 - Kyrgyzstan

Climate Change Impact

Part 9: Example –Kyrgyzstan


Kyrgyzstan has a continental climate with cold winters and hot summers. Most of the rain falls in the summer months and temperatures are below freezing for most of the winter months. It is projected that storm rainfall will increase by up to 20% and that the duration of lying snow will decrease.


Kyrgyzstan is in central Asia and has severe winters with temperature below zero for many months, particularly in mountainous areas.

Figure 1 Map of Kyrgyzstan showing project road

Climate change can affect roads in many ways. The most obvious is storm rainfall; an increase in storm rainfall could require modification to current design for culverts and longitudinal drains. Other factors include daily temperature range, which could affect expansion joints, and maximum temperature, which could affect the choice of binding agent.

Once current values of these parameters have been determined then the extent to which they will change in the future can be assessed.

The observed climate data were downloaded from internet sites which process data facilitated by international organisations such as the World Meteorological Organisation.

Other documents related to climate change for Kyrgyzstan were downloaded. These included:

  •         UNFCCC Country Brief 2014: Kyrgyzstan
  •         Climate Profile of the Kyrgyz Republic
  •         The Kyrgyz Republic: Intended Nationally Determined Contribution
  •         The Kyrgyz Republic’s Second National Communication to the United Nations Framework Convention on Climate Change

These documents describe the potential change to the climate in general terms but were not specific enough for road design.

Current Climate

Kyrgyzstan has a continental climate with warm summer and cold winters.

 The following chart shows the location of the climate stations which were used to determine current climate parameters. Only stations in the area of the project road are shown. The data were at a daily time step and included: precipitation, depth of snow, average daily temperature, daily maximum temperature and daily minimum temperature. Data were downloaded for the period 1950 to the present.

Figure 2 - Location of sites with climate data

Figure 2 shows the location of sites with climate data. Sites with a solid diamond have precipitation, temperature and snow data. Sites with an open diamond only have precipitation data.

The chart also shows the area covered by the nearest climate model cell as an orange rectangle. It is convenient that the section of road of interest corresponds to one of the climate cells.

Rainfall is highest in the summer months. In winter, average monthly temperatures are often below zero. The road itself passes between two areas with higher elevations and maximum daily rainfall is lower than areas with higher land – around 20 mm per day.

Climate projections

To examine the performance of climate models, their simulation relative to past observed climate for the period 1970 to 1990 was examined. The four models chosen on this basis were:

  •         bcc-csm1-1: Beijing Climate Center Climate System Model (China)
  •         IPSL-CM5A-MR: The Institut Pierre Simon Laplace (France)
  •         CCSM4: The Community Climate System Model Version 4 (USA)
  •         NorESM1-M: Norwegian Earth System Model (Norway)

The conclusions relating to climate change and daily maximum rainfall were:

  •   The average percentage increase for the final 50 years of the present century is 7.3% for the RCP 8.5 projections and 5.5% for the RCP 6.0 projections.
  • The projections of 5-day rainfall, more relevant for rivers which are crossed by the road showed a similar increase which could be translated in into increased flooding.
  • Winter and summer temperatures have been rising in recent years. The rate of increase has been 3.3 °C per century, slightly lower than the projected 5.5 °C per century. This significance of these increases relate to icing in winter and the heat-resistance of the road surface in summer. There is no indication that the diurnal temperature range (important for expansion joints) will increase.
  • On average, the depth of snow reaches 500 mm or even more in an average year.
·         When a range of models was ranked on the projected increase between the present and the year 2100, the difference in projection between the upper and lower quartile was quite modest; of the order of 15%. This applied for both RCP 6.0 and RCP 8.5.

·         When four selected models were compared their difference in projected values for the year 2100 was larger.

·         In most cases the largest percentage increase in daily rainfall for any projection occurred not in the final year of this century but in an earlier year. At any time in the current century the increase in precipitation rarely exceeded 20%, apart from the few models with the highest rate of increase in precipitation. A further consideration following from this is that the maximum increase in rainfall could occur during the projected life of the road.

There are no specific projections related to snow depth. Figure 3 shows two alternative metrics. The first is ‘icing days’; these are days when the daily maximum temperature is below zero. The second is ‘frost days’; these are days when the daily minimum temperature is below zero. The decline in these two variables indicates two things. Firstly, that snow fall will be less frequent and secondly that it will lie for a shorter time.

Figure 3 Days with temperature below freezing for all or part of a day


The main conclusions are that there will be an increase in storm rainfall of up to 20% during the life of the road. The period with lying snow will reduce.

Climate Change Impact - Part 8 - Samoa

Climate Change Impact

Part 8: Example – Samoa


Some of the roads on Samoa had been damaged in recent storms and the objective of the project was to prepare the rehabilitation taking account of climate change. Climate data, including rainfall at a 10-inute time step for two stations were obtained. The data showed that there was a significant increase in rainfall with elevation (which might explain why the most severe damage to the roads was at highest elevations). A methodology was developed to estimate the storm intensity for a range of durations and return period taking account of climate change.


Samoa consists of two main islands shown on the following map. Both islands have a road network. On Upolo there are roads around and across the island. On Savai’I the roads run around the island.
There is a third island, to the east of and smaller than these two, which is a US territory.

Figure 1 Map of Samoa

The main aim of the project was to upgrade some of the roads on the two Islands taking account of climate change. In particular one of the cross-island roads on Upolu had been damaged during a storm and it considered that its reconstruction should not suffer from the same problem.

As the roads around the islands are often close to sea level, the possibility of sea level rise also had to be considered.

Current climate

There are three types of data available in digital format:

  •         10-minute data from 2010 to 2015 for two stations, Nafanua and Afiamalu.
  •         Daily data from 1984 to 2014 for two stations, Faleolo and Apia.
  •         Monthly data from the early 1980s and in some cases earlier for four stations on Savai’i and one on Upolu.

These data were measured by the Samoa Meteorology Division (SAMET). Other data were abstracted from reports. Additional data on climate and flow were also obtained for US Samoa.

The weather of Samoa is influenced by four main factors:

  •         The sub-tropical high-pressure zone in the Eastern Pacific is a large semi-permanent anticyclone.
  •         Trade winds which blow from between east and south-east which contributes to a rain shadow effect to the north and west of the islands.
  •         The South Pacific Convergence Zone whose position helps to determine the seasonal pattern of the rain in which rain from November to March is above the monthly average.
  •         The Southern Oscillation which when in positive mode leads to increased rainfall.

The rain shadow effect is illustrated by Figure 2 which shows isohyets (contours) of the mean annual rainfall. Areas to the north and west of both of the main islands have less rain than areas to the south and east. The map also shows the effect of elevation on rainfall, with higher rainfall being associated with higher elevations.

Figure 2 Mean annual rainfall Samoa

Climate change projections

  1. The only daily record available for Samoa, of good quality and a long duration, is for Apia. This record was used to estimate the daily rainfall of a given frequency of occurrence.
  2. Two records of rainfall measured at 10-minute intervals are available for a period of up to 6-years, to the south of Apia. The rainfall stations are at different elevations and the one at the higher elevation records more rain than the other. However, when the rainfall at a short duration (from 10 minutes to a few hours) is expressed as a proportion of the daily value, the results are almost identical for both stations. Combining the two records, enables a single curve relating rainfall at a short duration to be calculated, as a proportion of the daily rainfall.
  3. The 10-minute rainfall records are at different elevations (796 m and 128 m) and have different daily storm rainfall (331 mm/day and 206 mm day). This implies that storm rainfall is higher at higher elevations. This is potentially an important conclusion but the 10-minute rainfall records are of short duration. These two records were combined with daily data from Apia, and charts and tables from earlier reports covering both islands, to arrive at a justifiable value for this effect.
  4. The data presented in some earlier reports implied that aspect is an important factor in storm rainfall, with storms on a south-facing slopes having twice the rainfall of slopes on the north or east. It was concluded that the limited data available do not allow an accurate value to be ascribed to this effect.
  5. The relationship between monthly and hourly rainfall was examined. The correlation for the two stations was weak at one station and non-existent at the other.
  6. Two methods are used to calculated the flow resulting from the storm rainfall: The Rational Method for small catchments and the Generalised Tropical Flood Model for larger catchments. These were complemented by the use of a hydrological model of American Samoa.
  7. Climate projections were based on 4 climate models: CSIRO, GFDL, HadGem and MIROC. These had been found to perform well in the region.
  8. Projections were provided for 3 time-horizons: 2030, 2055 and 2090.

 The 2055 projection this represents the highest intensities in this century. And was used for drainage design.
For daily values, this represents an increase of 17% on the current daily rainfall figure and for the standard deviation an increase of 7%. Both the daily values and the standard deviation are used to calculate the rainfall intensities for different frequencies of occurrence.
It was mentioned above that flow and climate data were used for a stream on US Samoa. This is very small catchment, 1.52 km2. (It is interesting to note that in another posting in this series that same model was able to successfully simulate flows in the Mekong River at a point where its drainage area was 660,000 km2.) The HYSIM rainfall/runoff model was run at an hourly time step, though the raingauge was outside the catchment. Figure 3 shows the simulated and observed daily flow.

Figure 3 Simulated and observed flow - Pago Stream - US Samoa

The rainfall and flow data were analysed to estimate an appropriate runoff coefficient. It was found that the coefficient increased for storms of higher return periods and was higher for 2-hour storms that 1-hour storms.
Sea level data were also analysed and it was found that in recent decades sea levels had been increasing by 5 mm a year. This is comparable with the projected values.

Climate Change Impact - Part 7 - Vanuatu

Climate Change Impact

Part 7: Example – Vanuatu


The project examined projected changes to road flooding for four islands of the Vanuatu archipelago. The conclusion was that rainfall intensities would increase for all islands, particularly at the lower durations and return periods critical for road drainage.


This example looks at the estimation of road flooding in Vanuatu. The islands of Vanuatu stretch from 13°S to 20°S and 116.e°E to 170.25°E. They lie about 2000 km from the coast of Australia.

The following chart shows the layout of the islands of Vanuatu. The four islands highlighted in green, Ambae, Pentecost, Malekula and Tanna, were included in the study. The red crosses mark the location of climate measurement sites. The roads on the islands are being upgraded and for this it was necessary to ensure that road drainage would be effective for the whole life of the road taking into account projected climate change.

Figure 1 Vanuatu showing the site of climate stations

Flooding on the islands can occur very suddenly and result in flash flooding with rapid increase in flow depth. The following photograph was taken the day following a storm.  The stream itself can been to the left of photograph. The highlighted area, near the tree, shows floating material trapped in the branches showing the depth of flooding. This indicates the degree of flood problems to tackled.

Figure 2 Material trapped in branches during a flood.

Current Climate

Daily observed climate data were obtained from 3 sources:
  •  The Vanuatu Meteorological and Geohazards Department (MGHD)
  •  The National Climate Data Center (NCDC) which is part of the National Oceanographic and Atmospheric Administration in the USA
  • TuTiempo web site

Data from the Meteorological Office was for Bauerfield, Efate and three stations on Tanna. All stations had daily precipitation and Bauerfield included temperature and wind speed. Data from NCDC was for Pekoa, Spiritu Santo, for daily precipitation and temperature. The data from TuTiempo were available for 6 sites and included precipitation, temperature, wind speed and relative humidity.

Where data were available from different sources for overlapping periods, their values were compared and were found to be compatible.

Data were also obtained on storm rainfall profiles. For road drainage, the critical time of a storm is often of the order of a few minutes so daily data on its own is not sufficient.

Climate projections

Projections were provided from PACCSAP (Pacific-Australia Climate Change Science and Adaptation Planning) and included precipitation projections in NetCDF (Network Common Data Format) for the whole world at the grid spacing of the original models. These are:

  •         ACCESS1-3, 1.875° longitude, 1.25° latitude (220km x 147km)
  •         CNRM-CM5, 1.4° longitude, 1.4° latitude (165km x 165 km)
  •         GISS-E2-R, 2.5° longitude, 2.0° latitude (294km x 235 km)

A further data set of projections was provided for which all the files had RX1Day in their title. They had been produced from the 50-km downscaling CCAM normal-cubic atmospheric model, a stretched-grid atmospheric model. They followed the guidelines of “Expert Team on Climate Change Detection and Indices (ETCCDI)”. These had projections of daily rainfall for three RCPs (RCP2.6, RCP4.5 and RCP8.5) and for 4 time-horizons (2030, 2050, 2070 and 2090).

Downscaling was carried out using the Delta Method.

The baseline period was 1987 to 2013 which had observed data. For consistency, the 2030 projection was based on the same number of years, 2027 to 2043. The 2055 projections used values from 2042 to 2068.

The temperature projections were consistent with different models showing similar increases. For precipitation, the projections were less consistent with higher values of RCP leading to higher rainfall for some models and lower rainfall for others.

The conclusions for the four islands were:

  • Ambae: Rainfall intensity is likely to increase for 2030 by about 15% for the 1-in-2-year storm up to about 30% for higher return periods. The additional increase for 2055 relative to 2030 is about 4%.
  • Pentecost: As with Ambae, the percentage increase is less for small return periods, 20% for 1-in-2, but up to 32% for 1-in-100. The 2055 projection is almost identical to the 2030 projection.
  • Malekula: The projections suggest storm rainfall intensity will increase from the baseline to the 2030 time horizon but after that will remain more or less constant. It is also noticeable that increases in storm intensity for low return periods are small, 18%, but increase for longer return periods, about 33%. The 2055 projection is slightly lower than the 2030 projection.
  • Tanna: The projections for this island were lower than those for the other islands.

Figure 3 Daily rainfall intensity for different  islands and return periods

The above chart shows the daily rainfall intensity for different return period on each island for 2055.

Sunday, August 13, 2017

Climate Change Impact Part 6: River Tagus - Tajo - Tejo

Climate Change Impact

Part 6: Example – Tejo-Tajo-Tagus


The Tagus (Tejo or Tajo) is one of the most important rivers in Europe. It rises in Spain and flows through Portugal to the Atlantic Ocean. The river basin is highly developed with several large dams and abstractions for irrigation and urban water supply. Data on reservoir storage and abstractions were used to estimate the natural flow of the river. A hydrological model of the river (HYSIM) was calibrated to observed flows. The calibrated model was then run with projected climate changes. The conclusions were that flows in the river would reduce but that the changes associated with human activity were of a similar order of magnitude.


The Tagus River (in English, Tajo in Spanish and Tejo in Portuguese) rises in the hearth of Spain and flows into the Atlantic close to Lisbon. The total basin area is 80,000 km2 and the length of the river is 1060 km.

The general approach to hydrology and water resources of the Tejo/Tajo basin was to calibrate hydrological model (HYSIM) which uses precipitation and potential evapotranspiration as input data and flow data for model calibration. Once the model had been calibrated the observed precipitation and climate data could be replaced with projected values related to climate change scenarios and the impact of climate change on river flows assessed.

Figure 1 shows the Tejo/Tajo basin and the principal flow stations.

 The Tejo/Tajo is a highly developed basin and this development has been ongoing for several decades. The total reservoir storage is 9 billion m3 in Spain and 11 billion m3 in Portugal. For comparison, the average natural flow at Alcantara is 6.5 billion m3/year.

Current climate

The simulation started with Spain and then moved on to Portugal. Some of the data sources had data for both countries and others were only for one country.

In Spain, comprehensive climate and flow data were available from a variety of web sites run either at a national or European level. These included precipitation and the climate parameters necessary to calculate potential evapotranspiration. Data were also available for Madrid water supply, one of the principal users. Annual data were available at province level for irrigation. 

The following shows the cumulative effect of artificial influences on flow.

Figure 2 River Tajo at Alcantara - Spain - artificial  influences on flow

The red line shows the influence of reservoirs. Basically, during winter, the reservoirs fill and during summer water is released – principally for hydropower. The Blue line shows the effect of irrigation with abstractions in the summer. The green line represent others uses of water, principally for urban populations. These data, combined with the observed discharge of the Tajo at Alcantara, were used to calculate the natural flow in the river.

The following chart shows the simulated and naturalised daily flow for the River Tajo at Alcantara.

Figure 3 Simulated and naturalised flow - River Tajo at Alcantara - Spain

Simulation of the naturalised flow was not straightforward. Normally peak flows have a major influence on the calibration but in this case the effects of storage masked some of the variation. However, the general pattern of low and high flows is well represented.

For the Tejo river basin in Portugal data on flow, climate, storage and abstractions was also available at the locations in figure 1. The station furthest downstream with flow data for the Tejo is at Almoural. To calculate the naturalised flow at this point it was necessary not only to include the influence of reservoirs and abstractions in Portugal but the cumulative effect including those in Spain as well. The following chart shows the simulated and naturalised flow.

Figure 4 Simulated and naturalised flow - River Tejo at Almoural - Portugal

Considering the many factors that went into simulating flow, the accuracy was good. The correlation coefficients were 0.91 for daily flows and also 0. 91 for monthly flows. The simulated and observed means were 353 m3/s in both cases.  The simulated and observed standard deviations of daily flows were 717 and 716 m3/s and for monthly flows were 550 and 563 m3/s.

Climate change

To examine the effect of climate change six scenarios were considered: A1B, A1B max, A1B min, A2 and B1. The A1B is considered the most probable projection.

Figure 5 Flow duration curve - observed and project flow - River Tajo at Alcantara

A flow duration shows the percentage of the time that a flow was higher than a given value. On this curve, the flows are plotted using a logarithmic axis. Some inferences which can be drawn from this graph are:
  •         Downstream of Alcantara flows are zero for more than 15% of the time.
  •         The difference between the natural and observed flow is of a similar order of magnitude to the effect of climate change.
  •         Average flows will be about 30% lower by the mid-century.

The next chart shows the average monthly flow for the River Tejo at Almourol. This is the lowest point of the Tejo with flow measurement.

Figure 5 Average monthly  observed, naturalised and projected flow - River Tejo at Almourol

This chart confirms that climate change will, not surprisingly, have a similar impact on flows at Almourol as at Alcantara. As with Alcantara, the effect of abstraction and reservoirs is of a similar magnitude to the effects of climate change. It also shows that under the A1B scenario, average flow will be reduced.

Climate Change Impact - Part 5 - Great Lakes of Africa (Lakes Victoria,Tanganyika and Malawi)

Climate Change Impact

Part 5: Example – Great Lakes of Africa


The Great Lakes of Africa are an important source of fish. The United Nations Food and Agriculture Organisation (FAO) wished to know to what extent climate change would influence the water temperature in the lakes. A study was carried out which found that lake temperature would increase by around 1°C by the middle of the 21st.century.


FAO initiated an activity to investigate the possible effects and impacts of climate change on fish and fisheries production on the African Great Lakes; Lakes Victoria, Tanganyika and Malawi.
Figure 1 Great Lakes of Africa

Figure 1 Great Lakes of Africa

Current climate

The Joint Research Centre (JRC) at Ispra, Italy, maintains databases of unpublished satellite data including water surface temperatures on the African Great Lakes on an 8-day basis. A time series of a number of decades was required to be analysed in order to present the temperature fluctuations on the Great Lakes
The satellite data from which temperatures are estimated are held in TIF (tagged image format) files, one for each year from 1985 to 2008. Each file holds data for 45 passes of the satellite.  For each pass a value is recorded for each cell on a 400 by 250 grid, provided there is no cloud cover and provided it is over water. Each cell is approximately 10 km by 10 km.

One major problem with the data is cloud cover.  For each of the lakes the approximate percentage of time for which temperature can be calculated is:

  • Lake Malawi – 70%
  • Lake Tanganyika – 60%
  • Lake Victoria – 30%
For all lakes, the problem is seasonal and is related to the rainy season.

The method adopted was as follows:
  •          Calculate the average lake temperature for each of the 45 passes and develop a temperature profile.
  •          Assume that although temperature in different parts of the lake would be higher or lower than the average, the distribution throughout the year would be the same.
  •          If data were missing, due to cloud cover, for one of the satellite passes find the value of the previous and of the following passes which had data for that cell.
  •          Base the temperature that satellite pass on the weighted average of the previous and following pass.

This enabled a complete grid of ‘observed’ surface temperature data to be prepared for all three lakes.

The following chart shows an example, for one cell of Lake Malawi.


Figure 2 Example of infilled lake temperature data

Climate change impact

As a first stage in assessing impact, an air temperature record was established for each of the lakes. The data from climate stations was of limited availability; few stations and long gaps in the data. As an alternative, the temperature data based on RSS (Remote Sensing Systems) estimates was used. This is one of two the ‘standard’ temperature records based on (Advanced) Microwave Sounding Unit data (AMSU/MSU). Comparing the limited observed data and the satellite derived data, showed similar trends but less variation. The difference in the temperature variation was due to the fact the satellite data were based on a 2.5 ° grid, not a single point, and therefore represented values over an area. In fact, this data was in that way more suitable than point data.

For each of the lakes a relationship between air temperature and water temperature was developed.

The climate change projections were based on the average of six climate models using the A1B (‘business as usual’) scenario. The models were those used in the IPCC “General Guidelines on the Use of Scenario Data for Climate Impact and Adaptation Assessment”. This gave the projected change in air temperature
The final stage was to use the relationship between lake surface temperature and air temperature to estimate the change in surface temperature of the lakes.

This shows that, for all lakes, the increase in surface water temperature would be around 1°C in the middle of the century.