Topographic data at multiple scales from satellites and contours
Topographic data at multiple scales provides the base information for basin masks, basin ID, river network delineation, and flow paths. The “Curva de Nivel” data (below) is available for the State of Espírito Santo, but it doesn’t include the upper Rio Doce basin. To provide coverage for the whole model region, the 90-m SRTM (Shuttle Radar Topography Mission, top image) conditioned by HydroSheds is an efficient and proven product.
Espirito Santo State and Minas Gerais
SRTM data were acquired and masked to Espírito Santos and the Rio Doce. The VIC domain, as derived from the topographic data, was set at a basin scale of 1/16th degrees ~ 6km. Procedures were set up so that, as VIC basin outlets are chosen, separate basin routing models can be built. This allows the routing model to reflect the fact that flow from some cells will be shared by two or more basins. This is particularly important because of the small size of some important basins. There are ten major river basins in the VIC domain, with a mask of 2938 points. Cells were created by selecting the ES coastline, finding all the 3-second cells that drain it, and selecting all the 1/16th degree cells that contain any small cells. Some of those cells might contain just bits of coast, and are not in the soil polygons. Basin data includes “fraction files”, which indicate what proportion of a 1/16th degree cell lies in the basin.
SRTM masked (left) and DEM and domain of Espirito Santo and Minas Gerais with station locations and flow directions (right)
Bacias Rio Jucu and Rio Santa Maria da Vitoria
The DHSVM model setup (At 150-m resolution) for the Rios Jucu and Santa Maria da Vitória requires area, a basin mask (boundary-modified to be consistent with the DEM, 150 m), a stream network map (derived from high resolution DEM, a stream routing system (processed through complex script into stream segments with identified stream class and stream computational orders), and stream class attributes (channel hydraulic properties associated with stream class ID. Default values are hardwired in the program).
An evaluation of different data sources for the DEM was done. (1) The default choice would be to simply use the SRTM/Hydrosheds product, scaled up to the 150 m resolution of DHSVM. (2) A group at INPE reprocessed the SRTM data, for specific application across Brazil, as TOPODATA. (3) Local DEMs based on field topography measurements, such as the the “Curva Nivel,” typically provide greater spatial resolution than SRTM products. (4) The ASTER GDEM Version 2 produces a 1 arc-second (30 m) DEM. (5) The new very high resolution DEM is being produced, by HiParc, under contract with IEMA, wasn’t yet available. There were surprising differences among versions of SRTM, and the HydroSHEDS conditioned DEM was surprisingly poor for this region. Version 4 of the CGIAR (related to CIAT) SRTM DEM proved to be the best. This was combined with mapped rivers.
Rivers for the basins were clipped out of Base de Dados Vetorial - Municipios ES/Rio_Principias - ES.shp and topologically repaired. (Small gaps were closed, and some river segments were directed downhill.) These rivers were rasterized at 30-meter resolution. The DEM was reprojected to UTM at 30-meter resolution, and the rivers were "burned in". Flow directions were calculated and a special algorithm (http://gis.ess.washington.edu/projects/scaling) was used to optimize the upscaling of the data to 150 meters. The resulting flow directions were displayed over aerial imagery and a few small adjustments were made.
DEM and the flow network for the Jucu and SMV