Research highlights: modelling to assess climate change impacts and promote development

Abstract

We highlight four recent articles on biophysical modelling for the Ecosystem Services and Poverty Alleviation (ESPA) Deltas project in the Ganges–Brahmaputra–Meghna (GBM) delta system. These publications are part of a themed collection in Environmental Science: Processes & Impacts and contribute to a larger body of collaborative work that aims to assess the impacts of changing climate, policy, and development efforts on vulnerable populations in the GBM delta.

Introduction

Although the fate of human populations is closely intertwined with that of the environment, the environmental and social sciences remain largely separated. Structuring research collaborations to answer fundamental questions while simultaneously generating comprehensive outputs relevant to regional managers is challenging, but increasingly demanded by funding agencies. A long-term research program, Ecosystem Services for Poverty Alleviation (ESPA), is addressing this challenge. This article highlights four recent publications on an ESPA project in the Ganges–Brahmaputra–Meghna (GBM) delta system, emphasizing their balance between fundamental questions and user-oriented outputs.

Deltas are vulnerable to many changing climate factors, like sea level rise, changes in upstream catchment management, and changing storm frequency. Many deltas are highly populated and the wellbeing of these populations is often closely linked to ecosystem services. The largest delta in the world, the GBM delta, is home to 150 million people.¹ The region has a monsoonal climate, with a dry winter and wet summer. Environmental, agricultural, and social processes are closely linked with the timing, length, and magnitude of the monsoons, which provide 70 to 80 percent of the annual precipitation.² Agriculture employs a quarter of the population in coastal Bangladesh,³ where over half of the households are practically landless.⁴ Despite the delta's significance, spatial and seasonal heterogeneity coupled with poor data availability have hindered the development of regional models useful for scenario analysis. The goal of the ESPA Deltas project is to overcome this knowledge gap and create models to “assess the future of coastal Bangladesh, and the role policy and development can have in shaping that future.”⁵

Climate modelling

Due to computational limitations, global climate models typically use a coarse resolution. This resolution does not capture the effects of topographical features, like coastlines and mountains, at the scale necessary to model and manage climate variability and change within the spatially heterogeneous GBM delta system. In a recent paper, Caesar et al. used a standard approach, known as dynamical downscaling, to develop a higher resolution regional climate model (Fig. 1).²

Fig. 1 Comparison of the regional climate model (left) and global climate model (right) resolutions for an example month of simulated precipitation. Reproduced from ref. 2 with permission from The Royal Society of Chemistry.

In this process, the authors focused on developing a model capable of accurately reproducing historical precipitation and surface temperature, the climate variables most relevant to regional interests in agriculture and development. Available historical data was sparse; different data sets agree on the spatial distribution of precipitation, but disagree on its magnitude. This is reflected in the model output, where 17 regional projections—based on different global climate projections, as a measure of uncertainty in the climate model—reproduce rain events and quartiles but not absolute magnitude. This result emphasizes the importance of uncertainty in validation data sets for high resolution climate models, an uncertainty often neglected relative to natural climate variability, uncertainty in greenhouse gas emissions, and model uncertainty.

All of the future climate projections generated by Caesar et al. predict increasing temperatures and rainfall in the GBM delta by the end of the century. This increasing total rainfall comes despite a decrease in the predicted frequency of wet days during the summer monsoon, which is compensated for by an increase in heavy precipitation events. This predicted shift is likely to have significant consequences for regional agriculture, water quality, and water storage, which are being addressed by subsequent studies within the ESPA Deltas project.

Hydrological modelling

The lack of spatially distributed precipitation measurements⁶ and flow gauge data⁷ have been identified as key barriers for the widespread development of useful hydrological models in the GBM basin. Futter and coworkers overcame these barriers by using the temperature and precipitation projections generated by Caesar et al. and showing that a simple conceptual model is able to recreate flow gauge data using only these input variables.⁸

Flow gauge measurements are important to create and validate hydrological models. When these data are absent, the ability to create plausible flow predictions is the best alternative. Using the Precipitation, Evapotranspiration and Runoff Simulator for Solute Transport (PERSiST) framework,⁹ a rainfall-runoff model generally applied to small research catchments, Futter et al. implement a minimalistic and a basic conceptual hydrological model of the GBM basin. Their predictions were compared with data from four gauged sites in the Ganga and one site in the Brahmaputra. The minimalistic model and basic conceptual model generated similar qualitative trends, reproducing the timing of peak and low flow conditions at these flow gauges. The conceptual model did a better job reproducing actual flow amounts, although statistical tools which weigh model performance by the number of free parameters suggested that one model is not preferred over the other. However, varying parameters within one model influenced the model results less than whether the minimalistic or basic conceptual model was used. The authors argue that, within the PERSiST framework, structural uncertainty was more significant than parameter uncertainty.

The authors confront an ongoing literature discussion about the appropriate level of complexity for hydrological modelling by applying relatively simple models to a major river basin, the GBM basin. They show that their models are able to simulate the flows and water balance components necessary for nutrient flux modelling reasonably well. The models perform best during low flow periods; although this makes the models less useful as a flood forecasting mechanism during monsoon season, the ability to perform water quality simulation and drought analysis during the dry season, when crop irrigation is needed, is very useful. Futter et al. show that their simple model structures are capable of producing plausible flow predictions without flow gauge data requirements, paving the way for an important management tool in the numerous data poor river basins worldwide.

Framework for model to integrate social and environmental processes

Most models incorporate either environmental or social processes. In their recent publication, Lazar et al. describe a preliminary model framework suitable to link environmental and climatic processes within the GBM delta system to the welfare of local households.⁴ This framework integrates the different distance and time scales required for the multitude of model components (Fig. 2). It is the first version of a more complete simulation model which will be used to explore how different interventions, under different plausible future socioeconomic and environmental scenarios, could alleviate environmental risks and promote development.

Fig. 2 The proposed integrated model structure. The spatial and temporal scales of the model elements are shown in the grey boxes. Reproduced from ref. 4 with permission from The Royal Society of Chemistry.

The model focuses on the southwestern coastal zone of Bangladesh and incorporates recent shifts in agricultural practices and demographic changes. In the model, agricultural productivity depends on water, salinity and temperature stress, and atmospheric fertilization by carbon dioxide. In this prototype model, most variables, like salinity and migration, are assumed to vary linearly. Households are assumed to take out loans when unable to cover their monthly expenses. Finally, the model considers only income related to agriculture, whereas a recent field study indicated that only a quarter of households depend solely on agricultural income.⁴

Because of these assumptions, the model results themselves are rudimentary. The prototype model suggests that temperature limits crop growth, sometimes as much as half, and that households in coastal Bangladesh have been caught in a loan trap since 1990. Further, the model suggests that “progressive” farming is lucrative but high risk compared to traditional rice farming, because of high upfront costs for crops that are more sensitive to weather extremes. The two measures used to assess household welfare—financial wellbeing and a modeled hunger index—show different trends with time, where financial poverty decreases as hunger increases. These are intriguing preliminary results, which the subsequent model can revisit.

Non-linear salinity buildup in agricultural soils

During the dry season, salinity in coastal Bangladeshi soils can build up through irrigation and capillary rise from saline groundwater.^10,11 If the monsoon rains are insufficient or draining infrastructure is not adequate to leach away these salts, salinity accumulates (Fig. 3) and limits agricultural productivity. This accumulation, unlike climate variables, can create a “tipping point” where current agricultural activities are no longer profitable and land use changes. Salinity accumulation was identified as a major concern for agriculture and drinking water by farmers in coastal Bangladesh.

Fig. 3 Seasonal variation of soil salinity and precipitation in southern Bangladesh. Salinity data are district average values. Reproduced from ref. 3 with permission from The Royal Society of Chemistry.

In a recent paper, Clarke et al. combine crop and water management models to determine at what threshold plant irrigation requirements and irrigation water salinity cause salt accumulation greater than the monsoonal leaching capability.³ They then apply their results to identify the districts within coastal Bangladesh at the greatest risk of soil salinity accumulation in the future, assuming irrigation water is riverine and using predicted river salinities.¹² With this information in hand, efforts to implement agricultural practices that avoid salt accumulation in the root zone can focus on the communities most likely to benefit.

When extrapolating salinity trends linearly from historical data, Lazar et al. found that salinity accumulation will have only a marginal impact on agricultural productivity in the GBM basin. Here, Clarke et al. showed that salinity accumulation, due to the interplay between salinity buildup in the dry season and monsoonal leaching, is non-linear. In the absence of salinity accumulation, this model predicts that interannual variability – as suggested by the prototype model in Lazar et al. – will remain the primary determinant of dry season agricultural productivity in coastal Bangladesh, and the risk of reaching a “tipping point” is low in all but the most coastal districts of Bangladesh. By identifying the districts at greatest risk of chronic salt accumulation, Clarke et al. have made it possible to focus mitigation and adaptation efforts on the most vulnerable and most likely to benefit districts, a key complement to the large-scale modelling efforts by Lazar et al.

Concluding remarks

The articles highlighted here are part of an integrated effort to model the environmental and social processes surrounding ecosystem services and socioeconomic development within the densely populated GBM delta system. Caesar et al. developed a high resolution climate model for a region characterized by complex topography and limited data for model validation. Futter and colleagues showed that, using only this data set and a basic conceptual model, it is possible to develop a functional hydrological model for the GBM basin. Lazar et al. developed a framework to integrate models from several different disciplines and identified the additional components and datasets necessary to validate and implement the model. Clarke et al. investigated a non-linear aspect of the prototype model, salinity accumulation in soils, and identified the minimum water quality necessary for sustainable agriculture in coastal Bangladesh.

Together, the ensemble of papers in this themed collection in Environmental Science: Processes & Impacts offers an impressive perspective on an effort to integrate the natural and social sciences. The project has paved the way for computationally cheap modelling in populous areas with little data, a significant breakthrough for a future with changing climate and developing populations. Identifying major sources of uncertainty in these efforts shifts the focus from other well recognized and thoroughly studied uncertainties to these factors, ultimately supporting the creation and refinement of powerful models for numerous other data poor regions.

References

1. C. A. Wilson and S. L. Goodbred, Annu. Rev. Mar. Sci., 2015, 7, 67–88.
2. J. Caesar, T. Janes, A. Lindsay and B. Bhaskaran, Environ. Sci.: Processes Impacts, 2015, 17, 1047–1056.
3. D. Clarke, S. Williams, M. Jahiruddin, K. Parks and M. Salehin, Environ. Sci.: Processes Impacts, 2015, 17, 1127–1136.
4. A. N. Lazar, D. Clarke, H. Adams, A. R. Akanda, S. Szabo, R. J. Nicholls, Z. Matthews, D. Begum, A. F. M. Saleh, M. A. Abedin, A. Payo, P. K. Streatfield, C. Hutton, M. S. Mondal and A. Z. M. Moslehuddin, Environ. Sci.: Processes Impacts, 2015, 17, 1018–1031.
5. R. J. Nicholls, P. Whitehead, J. Wolf, M. Rahman and M. Salehin, Environ. Sci.: Processes Impacts, 2015, 17, 1016–1017.
6. J. D. Miller, W. W. Immerzeel and G. Rees, Mt. Res. Dev., 2012, 32, 461–467.
7. R. Johnston and V. Smakhtin, Water Resour. Manag., 2014, 28, 2695–2730.
8. M. N. Futter, P. G. Whitehead, S. Sarkar, H. Rodda and J. Crossman, Environ. Sci.: Processes Impacts, 2015, 17, 1070–1081.
9. M. N. Futter, M. A. Erlandsson, D. Butterfield, P. G. Whitehead, S. K. Oni and A. J. Wade, Hydrol. Earth Syst. Sci. Discuss., 2013, 10, 8635–8681.
10. H. Brammer, Clim. Risk Manag., 2014, 1, 51–62.
11. J. M. Beltran, Agr. Water Manag., 1999, 40, 183–194.
12. S. Dasgupta, F. A. Kamal, Z. H. Khan, S. Choudhury and A. Nishat, World Bank Policy Research Working Paper No. 6817, 2014.

---