# Socioeconomic Downscaled Projections

Follow Us: Twitter Follow Us on Facebook YouTube Flickr | Share: Twitter Facebook## Country-Level Population and Downscaled Projections Based on the SRES B2 Scenario, v1 (1990 – 2100 )

The IPCC SRES emissions scenarios (A1, A2, B1, B2) used population projections from both the United Nations (UN) and the International Institute for Applied Systems Analysis (IIASA).

**General Background Information for the UN Projections**

The SRES B2 population scenario was based on the UN 1998 Medium Long Range Projection for the years 1995 to 2100. The official version projects population for 8 regions of the world: Africa, Asia (minus India and China), India, China, Europe, Latin America, Northern America, Oceania. However, the UN also prepared a special 'unofficial' Long Range projection for the IPCC SRES for 11 regions of the world: North America, Western Europe, Pacific OECD, Central and Eastern Europe, Newly independent states of the former Soviet Union, Centrally planned Asia and China, South Asia, Other Pacific Asia, Middle East and North Africa, Latin America and the Caribbean, Sub-Saharan Africa.

The UN 1998 Long Range Projections are a regional extension of the 1996 UN "Revision" which projects population by country between 1995 and 2050. Projections are done for 3 variants: high, medium and low. The above link lists these three 1996 UN Revision projections for countries or areas. Countries with populations less than 150,000 are not included here but will be made available at a later date.

We have also included 1990 as the base year for our database for consistency with the base year used in the SRES report. We obtained 1990 population estimates from the UN Common Statistics Database (UNSTATS). The data were accessed at: http://unstats.un.org/ in April 2002.

**Extended Projection of the 1996 UN Medium Country-level Projection Using the 1998 Long Range UN Medium Regional Rates of Change (for SRES B2 Scenario)**

We have extended the country-level 1996 medium projection beyond 2050 and out to 2100 by applying the appropriate regional rates of change from the UN’s 1998 Long Range medium projection – reallocated by Thomas Buettner of the UN Population Division to the 11 IIASA MESSAGE model sub-regions - to each country. Therefore, in our extension, all countries falling within the same UN region experience the same rates of change for their populations.

To explain the steps, take Angola as an example. Angola falls within the tailored UN projected region Sub-Saharan Africa (SSA). Angola’s population projection from 1995 to 2050 is supplied by the UN 1996 Revision. To approximate Angola’s population in 2055 we first calculated the SSA regional *annual* population growth rate between 2050 and 2055 using the UN SSA population data for those two years. This annual growth rate was calculated using the following formula:

**r_SSA _{2050-2055} = log_{e}[P_SSA(2055)/P_SSA(2050)]/5** (1)

where P_SSA(2055) and P_SSA(2050) are the regional SSA population totals from the UN for years 2055 and 2050 respectively. The log formula accounts for the fact that the annual growth rates are applied to a continuously changing population base.

Then, starting with Angola’s population in 2050, and using the above rate, Angola’s population in 2055 was estimated as:

**P_Angola(2055) = P_Angola(2050) * Exp[r_SSA _{2050-2055}*5]** (2)

Angola’s population in 2060 is obtained using the same formula but substituting the 2055 population on the right-hand-sign and so forth. We followed this procedure for the entire country-level list in the base year.

Equations (1) and (2), applied together, constitute exactly a linear scaling of the country population with the regional population. Additionally, since the rates are applied uniformly to each country within a region, the method is linear with respect to regional totals. This means that if we begin with a base-year country population list which sums to the exact regional total, the agreement with the regional totals will remain exact for the remainder of the downscaling period. Similarly, if the base year country population sums to ±D% of the regional total, this base year difference will be exactly preserved at each time step for the remainder of the downscaling period.

However when the country lists are subsequently summed to the larger four SRES reporting regions, the linearity is not preserved because of the changing contributing weights of each region to the SRES reporting regions. This feature is seen in the accompanying difference tables for our downscalings, which show varying differences between our totals and the published four SRES reporting region totals. The variance is not large however and is usually at most a few percent.

It is useful to note that the method we use above is mathematically identical to keeping the ratio (or fraction) of a country’s population to the regional total, constant over time. In other words, in our tables, if a country starts off at x % of some regional total, it remains x % for the duration of the downscaling period. This can be understood by noting that if the ratio of a country population to a regional population remains constant over time, the country population will scale linearly with the regional population. If it scales linearly with the regional population, the country and region will have the same growth rates.

**Population Downscaling Discontinuities: **

Artifacts arise with the present downscaling procedure. The problems occur because of the post-2050 transition to the uniform growth rate method. If a country is projected by the UN 1996 Revision to have a declining (or growing) population at 2050 but falls within a larger region that has a growing (or declining) population after 2050, a discontinuity will occur. For example, Cuba and Barbados are problematic in this regard and should not be used. Other countries may have a slower or faster projected growth rate at 2050 than the regional projection. In these cases, the population slope for such countries will show a discontinuity, post-2050.

If we attempt to remove these discontinuities on a case-by-case basis, such as by using additional country-specific information, or even deleting them from the database altogether, then the regional totals will likely develop discrepancies with those in the SRES report.

Overall there are clear artifacts associated with the simplicity of the linear downscaling approach, and they can only be removed with more sophisticated treatments, or by relaxing constraints on consistency with the SRES report.

Finally, we have not included population data for 44 small countries, with populations less than 150,000, because these were not readily available from UN data sources in *electronic form*. Many small island nations vulnerable to sea level rise are in this category, so that climate impacts researchers will eventually need such data. This gap will corrected with future work.

The first accompanying spreadsheet on the previous data page presents our extension of the UN 1996 medium projection. The first 50 years are identical to the official UN 1996 medium projection to 2050 linked above.

**B2 Population Comparison**: On the second tab of the B2 spreadsheet we show the result of re-aggregating our downscaled B2 population estimates from the above website, and then comparing these sums to the aggregated totals in the SRES report. As seen, the differences are extremely small, if not zero, and apart from the base year, are on the order of less than 0.1%. The slightly larger base year differences (<0.5%) are due to the fact that 1990 is not the base year for the 1998 UN Long Range projection used in the SRES report – which is 1995. As indicated above, we accessed 1990 country-level population data from a recent UN Common Statistics database at: http://unstats.un.org/ in April 2002. The SRES report had to similarly use an independent 1990 source for population so that source evidently had small differences with our accessed data.