How is data prepared for the models?
To apply each model, a “dataframe” needs to be developed; i.e. the data required for each model. Again, the intent is to develop the DIF to:
- Provide quantitative and geospatial baselines for land cover (vegetation structure, plant-based biodiversity and functional types), land use, land quality (soils, topography), and hydrology (stream flow network, flows) for the project areas. Attention needs to be paid to both historical conditions, and to what scenarios of future landuse will be of interest.
- Link the above DIF data layers via a functional distributed hydrology model to predict the impacts of climate, land cover, and land use changes on biodiversity, land, and water.
- Build local capacity to develop, calibrate, and use the models to achieve project goals and to then adapt and scale it up to national scale.
- For the task of elaborating the dataframe, the involvement of all partners is essential to identify the sources of data, and to decide on parameters, formats, access etc. (all of the “details”) that will be used
The VIC model is used for the application to the whole state. Inclusion of the upper Doce means (at least) two things. The first is that this complicates the selection of databases for model setup, because it is necessary to have the “same” data sources for the entire setup. Much of the data available in e.g. GEOBASE is just for the state, and doesn’t include the upper Doce. So other data sources had to be found. The second is that the interest in the work has broadened considerably, as the Rio Doce is a national-level asset, and it is not clear that an integrating model system, such as what is being done here, is available. The decision was made to focus first on the dataframe setup and model operation for VIC. Data tend to be more available at this scale from different sources’ not just local, which makes it faster to obtain and process. The experience of doing the VIC first then allows a more detailed scoping of what is available for DHSVM. The setup is for a daily time step, with a spatial scale of; 1/16o, ~6km).
Three primary issues arise in mobilizing the dataframe for DHSVM.
- There are clearly multiple overlapping sources of data/information for topography, soils and landuse, between the different agencies (INCAPER, IJSN, CESAN, IEMA). It is be necessary to sort out which is the most updated and highest-quality data, get access to it, then develop the data frames. In the end, very little data has been forthcoming to date from these sources. As a result, the UW team has had to develop the primary data.
- For the higher resolution climate forcing, it would be best to use time series data from local/regional observations, and/or regional models. But a ready source of such information is not yet available.
- Discharge data at gauging stations. As discussed, significant work is going into producing usable data records.
The dataframe for the DHSVM application for the Rios Jucu and Santa Maria da Vitória was being established with a time step of 3 hour, and a spatial scale of 150 m. A higher resolution topography is required for the sediments module, of 30m. Future applications for the Mangaraí will likely be a higher resolution, pending data availability.
These overall model requirements establish the basis for the data development described under topography, soils, landuse, and climate.