Humans have needs – food, water, energy, health, security – and development is required to provide for these needs. Such development, and a rapid increase in population, have led to improper land use change, disorderly urbanization, and unstable governance, with societal problems the result. Development has also caused environmental issues on a global scale: climate change, desertification, deforestation, and loss of biodiversity. States must consider how to balance these issues with needed development.

In 2015, global attempts were made to reconcile the relationship between development and environmental issues. The Sendai Framework for Disaster Risk Reduction was developed in March that year (Aitsi-Selmi et al., 2016). The Sustainable Development Goals (SDGs) followed in September, the Paris Agreement on Climate Change (Paris Agreement) in December. As the SDGs and the Paris Agreement state, measures to mitigate climate change are the same as those that build resilience against climate-related disasters. In this regard, it is important to identify and evaluate under-recognized disaster risks that hinder sustainable development. This can be done by studying hazards and vulnerabilities that arise from societal and environmental problems and the way these risks change.

To build resilient societies, scientists and engineers must follow two paths in developing and using science and technology (Future Earth, 2013): promote interdisciplinary research between natural/applied sciences and humanities/social sciences and advocate transdisciplinary cooperation to facilitate the implementation of science and technology in social contexts. We need to advance scientific and technical knowledge, build relevant data infrastructures that allow for the anticipation of future events with greater accuracy, develop data archives, and expand our understanding of how changes and disasters unfold across different regions and sectors.

For the above reasons, we developed the Data Integration and Analysis System (DIAS), which collects and stores data from satellites, ground observation stations, and numerical weather prediction models; integrates these data with geographical and socio-economic information; and provides results for the crisis management of threats such as global environmental issues and natural disasters. These results are available in Japan and overseas. This paper gives an overview of DIAS and summarizes its application to climate change analysis.

The Coupled Model Intercomparison Project (CMIP; Meehl et al., 2000; Covey et al., 2003) is a standard experimental protocol for studying the output of coupled atmosphere–ocean general circulation models (coupled GCMs). CMIP was established in 1995 under the World Climate Research Programme’s (WCRP) Working Group on Coupled Modelling (WGCM). GCMs have been employed for decades in theoretical investigations of the mechanisms of climatic change, including the detection of anthropogenic effects in past climate records and the projection of future climate change due to the human production of greenhouse gases and aerosols. The third phase of CMIP, known as CMIP3 (Meehl et al., 2007), delivered an unprecedented, comprehensive, and coordinated set of model outputs that served the Intergovernmental Panel on Climate Change (IPCC) Working Group I. These outputs were used to prepare the IPCC Fourth Assessment Report in 2007 (IPCC, 2007). Future climate projections delivered by CMIP3 were based on a set of greenhouse gas emission scenarios for the 21st century, as formulated in the Special Report on Emission Scenarios (SRES; Nakicenovic et al., 2000). The thoughtful design of the experiments and open access to the coordinated dataset enabled a large group of climate researchers to analyze these data and carry out an extensive range of studies based on them. Nonetheless, due to the complex format of CMIP3 data in the Program for Climate Model Diagnosis and Intercomparison (PCMDI) home archive and its relatively large volume (33 TB), the data were not easily accessible, especially to researchers from other disciplines and users outside the scientific community.

To make the CMIP3 data more easily accessible to a wide range of users, a dedicated CMIP3 analysis tool was developed on DIAS (accessible at http://www.dias.nii.ac.jp/modelvis/cmip3/) and launched in 2010. This tool enables users to quickly access, browse, visualize, analyze, subset, and download CMIP3 data, which have also been replicated to the DIAS archive from the original site at PCMDI. These functions can be performed for a specified time or space through simple screen operations and without the need for users to have the data on their own computers. The DIAS CMIP3 tool has been used extensively, especially by hydrologists in Asia studying the impact of future climate change on hydrological regimes and water resources at the river-basin scale. With water resources being a key prerequisite for sustainable development, such assessment studies are in demand by those managing and developing policies for these resources. For relevant examples, see Koike et al., 2015.

Following the IPCC AR4 publication and reflecting on gaps in information that CMIP3 could fill, new coordinated climate model experiments were designed under CMIP. These experiments comprise the fifth phase of CMIP (CMIP5; Taylor et al., 2012) and are much broader in scope than those undertaken for CMIP3. The CMIP5 dataset, which includes outputs from more than 20 modeling groups worldwide, was used to prepare the IPCC’s Fifth Assessment Report (IPCC, 2013). The future climate projections provided through CMIP5 are based on new IPCC greenhouse gas emission scenarios—the so-called representative concentration pathways (RCPs; Moss et al., 2008; van Vuuren et al., 2011). Since the CMIP5 dataset is massive and multidimensional, only users in command of advanced data processing techniques can handle its original format. Building on its experience with the CMIP3 tool, the DIAS community has developed a similar analysis tool with extended capabilities, which is dedicated to CMIP5 data. The CMIP5 dataset was a new challenge for the DIAS team, mainly due to its volume (1.7PB) and complexity in terms of the number of experiments, GCMs, their versions, and ensemble runs.

Development of the CMIP5 tool was possible due to the unique DIAS environment, which was created by the integration of three components: a top-technology infrastructure with large storage capability and a high-performance analysis system, an application development platform, and an R&D community. While the advanced IT infrastructure of DIAS is vital, the development of an application such as the CMIP5 tool would not be possible without the close R&D environment DIAS affords. Intensive and iterative collaboration between the DIAS IT team and a group of CMIP5 data users from the hydrological and environmental science community led to the creation of a tool based on user needs. Its interface is simple and intuitive, and use of the tool can be explained through short training sessions or self-learned via the Help document attached to the tool.

The CMIP5 tool offers the following functions:

The DIAS CMIP5 tool has made future climate change projection data available to Japanese research institutions and public and private bodies. At the same time, international contributions to the promotion and social implementation of research in the field of climate change have expanded through the use of the CMIP5 tool as part of Official Development Assistance (ODA) in the Philippines, Indonesia, Tunisia, and other countries. The CMIP5 tool has greatly simplified the selection and processing of GCM output for hydrological and water resource applications at the basin scale.

Koike et al. (2015) described the methodology for using CMIP3 or CMIP5 data in assessing the impact of climate change on hydrological regimes and water budgets in river basins. GCM outputs suitable for regions of interest are used as forcing data for hydrological models. The CMIP5 tool function, which allows for 2-D comparison of model output with reference data, can be used to select suitable GCM outputs. Since there is substantial bias in the precipitation output of GCMs across the entire intensity spectrum (insufficient extreme events, biased mean intensity, extensive light intensity drizzle, too few dry days), this precipitation output cannot be used to force hydrological or other impact models without some form of prior bias correction. A three-step statistical bias correction method for precipitation, developed by Nyunt et al. (2013, 2016), has been incorporated into the CMIP5 tool.

To yield solid results, this method requires reliable precipitation observation data of high spatial resolution. All users of the CMIP5 tool have the option of using precipitation data from the Asian Precipitation Highly-resolved Observational Data Integration Towards Evaluation of Water Resources (APRHODITE) project (Yatagai et al., 2012). APRHODITE develops state-of-the-art daily precipitation datasets with high-resolution grids for Asia (0.25° and 0.5° resolutions). The datasets are created primarily with data obtained from a rain-gauge observation network and the resulting gridded product is freely available to all users. The accuracy of APRHODITE data depends on the density of the rain-gauge network, which is used to generate the gridded product. In addition, the spatial resolution of the freely available product may not be sufficient to capture the precipitation distribution at the basin scale, nor do APRHODITE datasets cover the whole globe. The CMIP5 tool, however, offers an option to use in situ precipitation observations to correct for bias in GCM precipitation outputs. This option is only available to users who provide their own in situ observation data, and depends on the data provider and the DIAS team having reached agreement regarding data policy. Such functions, dedicated and tailored to individual users, are the result of the strong R&D component of DIAS and its versatile application development platform. The Asian Water Cycle Initiative (AWCI) has used in situ observation data for bias correction in climate change impact assessment studies conducted in AWCI member countries. Section 4 includes more information regarding AWCI.

In 2013, discussions on and preparations for the sixth phase of CMIP were initiated (CMIP6; Meehl et al., 2014). The great interest of the wider scientific community in CMIP6 planning resulted in a wider range of scientific questions and a substantially larger number of proposed and endorsed experiments (Eyring et al., 2016). The CMIP6 simulations commenced in 2016 and will be completed in several years. The total CMIP6 output volume has been estimated to reach approximately 20 PB (Balaji, 2016), which poses a great challenge for data replication and storage. Encouraged by the success of the CMIP3 and CMIP5 tools, the DIAS team is considering the development of a similar tool for the CMIP6 output. However, the actual scope of the experiments to be embraced and the set of provided functions is subject to discussion. The functions available on the CMIP3 and CMIP5 tools, as described above, will be maintained but can also be expanded based on the further requirements of users. While the DIAS CMIP3 tool might be the first instrument to enable easy access to the CMIP data, which were tailored to the needs of the climate impact research community, other similar endeavors have emerged during the CMIP5 phase. One is the Infrastructure for the European Network for Earth System Modeling (IS-ENES) project under the European Union’s Seventh Framework Programme for research (FP7), which developed the Climate4impact tool (Plieger et al., 2015)—a platform enabling impact researchers to explore climate data and perform analyses. There are certain similarities as well as differences between the DIAS CMIP5 and Climate4impact tools. DIAS is a centralized system with its own large data archive, and the DIAS CMIP5 tool works with the data stored there. The Climate4impact tool is a data analysis platform built on the distributed Earth System Grid Federation (ESGF) network and accesses data and some services on remote servers of the ESGF data nodes. Regarding the provided services, the data search, selection, and visualization functions are similar in scope, but there are differences in the data processing and analysis functions. Both tools offer subsetting with respect to time and area, 2-D visualization of selected indices over a specific area, and model intercomparison of 2-D plots or time series. In addition, the DIAS CMIP5 tool provides analysis functions such as the comparison of model output with reference data, inter-annual variation (time series and 2-D plots), and statistical trend analysis and rainfall bias correction, as noted above. On the other hand, the Climate4impact tool provides the calculation of climate indices supported by a dedicated wizard and offers statistical downscaling. Also, the Climate4impact tool provides more explanatory text including generic use cases and other valuable advice regarding climate impact studies. This may be one of the next steps for the DIAS team to make the DIAS CMIP5 (and later CMIP6) tool more self-explanatory.

One of the main aims of DIAS is to facilitate and support problem-solving by integrating the assessment of social problems with relevant data. In this section, we focus on the water cycle and water resources, since the provision of ample and clean water is essential for reducing poverty, promoting health, ensuring food and energy security, and maintaining healthy ecosystems and biodiversity.

Accurate, timely, and long-term water cycle information is essential for integrated water resources management, particularly given the threats posed by climate change. With this in mind, GEOSS AWCI was established in 2007, bringing together 18 Asian countries. A similar collaborative framework was established in Africa in 2009. Several countries and transboundary river basin authorities and organizations are involved in the GEOSS Africa Water Cycle Coordination Initiative (AfWCCI). Recognizing the advanced capabilities of DIAS—data sharing, integration, the translation of data into information relevant to water resource practitioners and policy decision makers—both AWCI and AfWCCI collaborate closely with DIAS. Practically, this means that AWCI and AfWCCI member countries and authorities provide DIAS with meteorological and hydrological in situ observations from selected basins. In turn, DIAS shares satellite observations and GCM output data, as well as analysis and research. Such collaboration would not be possible without a strong commitment from policymakers in the AWCI and AfWCCI countries. To ensure a commitment to ongoing collaboration, the AWCI and AfWCCI research communities have demonstrated the capabilities of DIAS and the benefits of its use to members of government, agencies, and academia. A number of meetings were held in which representatives of government organizations and academia were invited to present outcomes of successful implementations of DIAS as a data integration platform, e.g., the Water–Climate–Agriculture workbench in Cambodia (Monichoth et al., 2014) and the real-time flood prediction and dam operation optimization system in Japan (Shibuo et al., 2016). In addition, selected functions of DIAS were demonstrated at these events, e.g., the support system for in situ data providers to submit data, perform data quality control, and generate metadata and metadata documents compliant with ISO standards. Moreover, training sessions were held regarding the use and applicability of CMIP tools, which were open to government organizations and their nominees. These face-to-face meeting opportunities were crucial to the continued commitments and collaboration. Recognition of the AWCI and AfWCCI regional collaborative frameworks by the international GEO community has also been very important for ensuring these commitments from member-country policymakers. Under the AWCI and AfWCCI frameworks and in collaboration with DIAS, programs for system development and capacity-building have been launched in Asia and Africa. In these programs, global earth observations are effectively used to develop flood-control/drought-prevention measures, water environment improvements, and climate change adaptations.

One AWCI activity contributing to social issue solutions has involved assessing the impact of climate change on water resources, i.e. how precipitation patterns will change in future and how the frequency of floods and droughts will be affected as a result. This information and analysis is essential for drafting river basin management plans and adaptation measures. Implementation of this activity included:

The DIAS CMIP3 and CMIP5 tools have enabled such studies. Aside from pilot research and demonstration studies, comprehensive assessments of the impact of climate change on water resources have been requested in the Philippines (Pampanga River basin; Jaranilla-Sanchez et al., 2013; JICA Final Report, 2013), Japan (Yoshino River basin), and Indonesia (Musi River basin). These assessments have formed part of broader national studies on water resources and planning for the future in this regard.

Similarly, under the auspices of AfWCCI, government agencies and research institutions in African nations are promoting climate change analysis and adaptation. The GEOSS Joint Asia-Africa Water Cycle Symposium, held in Tokyo in November 2013, brought together approximately 200 people involved in AWCI and AfWCCI. Stakeholders had the opportunity to consider water-related issues in Asia and Africa, share knowledge and experience, and contribute to meaningful debate on integrated water resources management and watershed management.

Rapid urbanization and the effects of climate change mean there is urgent need for Myanmar to reduce the risk of floods in the country (Htut et al. 2014; Win et al. 2015; Kawasaki et al. 2017). These twin challenges will lead to a greater number of devastating floods if left unchecked. Scientific analysis and planning that considers the effects of current flood characteristics and climate change, and the ways in which the flood risk will change, is required. With such analysis, decision makers and government agencies can make informed and careful plans for development, while working to reduce the flood risk and adapt to future scenarios. In addition, local, regional and national disaster management systems in Myanmar should be formulated on a scientific basis, especially given the potential effects of climate change.

Climate change analysis requires the use, storage and processing of big data. Figure 6 shows the research framework used in an on-going climate change study (Myanmar). There are three main steps in climate change analysis:

Central to climate change analysis for flood risk reduction is the provision of high-resolution precipitation data. This data is necessary for understanding flood hazards in specific regions and river basins. High-resolution precipitation is produced using dynamic downscaling through the usage of a regional climate model (RCM), in this case the Weather Research and Forecasting model. Using GCM model outputs from the CMIP5 project in DIAS as forcings to RCMs, high-resolution past/present and future climate projection data for a 30 year period each. However, there are high computational needs and huge data storage needs (several GB of space to ~100 GB per 1 year simulation of the ~25 km grid resolution) in using RCMs to downscale precipitation. Also, the production of climate projection data face two main issues related to RCMs for downscaling and GCM outputs as inputs to RCMs.

Domain settings and sub-grid physics selection in RCMs, which previous studies have shown to be highly sensitive in simulating past events, are especially important in setting-up the model correctly. So, multi-observation comparison (Figure 6) with simulation results is needed to validate the high-resolution (downscaled) products. This ensures the model is capable of reproducing past simulation events, and its correct set-up. When not set properly, these two components can add to the uncertainty involved in climate projection outputs.

Before GCM outputs are used as forcing for producing high-resolution data, GCMs are selected using the DIAS CMIP5 tool, which analyzes the capability of the GCM to reproduce the historical run. This ensures that the selected GCMs have the ability to adequately reproduce the past/present climate over Myanmar region. GCMs are selected are based on a scoring procedure (Figure 7) for capturing the relative distribution and absolute value of the precipitation (including other variables defined important for Myanmar climate). The selection of GCMs limits the uncertainty in the climate projections data. Due to GCMs coarse resolution and usage of RCMs, the downscaled historical (past/present) runs can still have biases as compared to the observations. The biases of the high-resolution products are minimized through bias-correction methods. After setting up WRF model and selecting GCMs, high-resolution data (historical and future projections) are produced based on the Pseudo-Global-Warming (PGW) method of F. Kimura and A. Kitoh (2007).

On production of the 30-year historic and future climate projection runs, bias correction of products such as precipitation and temperature is needed before these can be used as inputs for impact assessment. This requires computational needs and storage, but not to the level of the previous step. Bias correction is needed due to the coarse resolution of the GCMs and inherent errors therein. Thereafter, hydrologic (Water and Energy Budget Distributed Hydrological Model, WEB-DHM) and inundation models can be used for local scale assessment: this is necessary in order to understand how floods in the past and floods in the future may differ.

The DIAS environment provides the model running and model evaluation procedure of the downscaling precipitation component of the framework. The DIAS environment analysis cluster is used to run multiple cases considered in the WRF model set-up and GCM selection procedure. Bias correction analysis, hydrologic modeling and inundation modeling are done on the analysis cluster and archives allow for validation. The DIAS environment links data storage needs, access to archived datasets, and computational demands, and as such provides the analytical backbone of the downscaling precipitation and basin scale impact assessment.

Essentially, the key to climate change analysis is to have a platform that can provide multi-observation and reanalysis data, as well as GCM outputs (CMIP5 climate projections), and handle storage needs and high computational demands. The DIAS environment meets all these requirements. Without such a platform, it would take more time and effort to assess the impact of climate change. The study in Myanmar is testament to this, and the effectiveness of the DIAS platform.

With regard to research on the global environment, DIAS is involved in all aspects of the data lifecycle—measurement, retrieval, analysis, archiving, and sharing of information. Although there are other specialized data measurement/management infrastructure systems, DIAS is distinctive in the way it integrates the infrastructure system, application development, and R&D community to facilitate interdisciplinary and transdisciplinary research and support solution development with a broad cross-sectorial approach. Over the 10 years of DIAS’ implementation, the DIAS team has learned a number of lessons that may be useful to those undertaking other similar initiatives:

Political decision-making based on data has a decisive role to play in future climate change analysis and disaster risk reduction. In this regard, funding for platforms such as DIAS must be ongoing. A commitment to DIAS and other integrated platforms will also lead to greater community understanding of climate change and its effects. DIAS combines analysis methods, data assimilation, and prediction models from across fields and disciplines. In this way, DIAS aims to initiate cooperation between a variety of stakeholders, and contribute to the creation of scientific knowledge and the goal of sustainable development.