Estimating net migration by ecosystem and by decade: 1970–2010






Estimating net migration by ecosystem and by decade: 1970–2010




Government Office for Science UK


This project generated estimates of net migration (NM) by ecosystem over the four decades from 1970 to 2010. Because of the lack of globally consistent data on migration, indirect estimation methods were used. We relied on a combination of data on spatial population distribution for five time slices (1970, 1980, 1990, 2000 and 2010) and subnational rates of natural increase in order to derive estimates of NM on a 30 arc-second (~1 km) grid cell basis, which were then summed by ecosystem. We ran 13 geospatial NM estimation models based on outputs from the same number of imputation runs for urban and rural rates of natural increase. We took the average and standard deviation of the runs to produce the results described in the ‘Results’ section and Appendix A (see Maps A1–A4 for maps of the results by decade).
In summary, and noting that ecosystems are not mutually exclusive (the same grid cell can be counted for example in cultivated, island and coastal ecosystems), we found that:
• Most outmigration occurs over large areas, reflecting its largely rural character, whereas areas of net inmigration are typically smaller, which reflects its largely urban character.
• Coastal ecosystems (as defined circa 2000) have experienced the highest levels of net inmigration, with levels ranging from ~30 million (m) persons in the 1970s to 1980s to +82m in the 2000s.
• Inland water ecosystems have experienced the second highest levels of net inmigration, with levels ranging from +23m in the 1980s to +53m in the 2000s.
• Mountain, forest, cultivated and dryland ecosystems all show high levels of net outmigration, ranging from –12m to –43m across all decades. Mountain ecosystems have the highest net outmigration over the four decades, totalling –126m. The patterns across these ecosystems are consistent with global trends in rural-to-urban migration over the past 40 years.
• Considering their generally small populations, island ecosystems have high levels of outmigration, ranging from –3m to –4.5m.
• The largest population countries such as China and India tend to drive global results for all the ecosystems found in those countries.
• There are large standard deviations for the Asia model runs, especially in the decade from 2000 to 2010. This is due to small variations in rates of natural increase generated by the model runs, which, when multiplied by large populations, result in large standard deviations.
There are a number of uncertainties and potential sources of error in these estimates. The uncertainties include measurement errors in the spatial and tabular datasets used, potential biases in the results of the imputed time series of urban and rural rates of natural increase and issues arising from the simplifying assumptions we applied in our processing steps. These uncertainties are addressed in greater detail in the ‘Evaluation of results and next steps’ section, along with efforts to evaluate our results. However, we note here that the lack of observed population distribution data from 2010 round censuses means that the results for the 2000s are subject to greatest uncertainty.


Foresight. (2011). Estimating net migration by ecosystem and by decade: 1970–2010. London : Government Office for Science UK. URL :