Water balance analysis for Objective 3
Rafaela Flach
2022
Data sources
Here two sets of data are analyzed: the CAMELS-BR dataset (Chagas et al. 2020), and the dataset we built and organized based on the methods described by Ruhoff et al. (2022).
Dataset based on Ruhoff et al. (2022) methods
1- Hydrological data
The hydrological data was obtained from the Hydroweb portal from the Brazilian Water Agency (ANA) for 12 measurement stations in the Xingu, São Francisco and Tocantins basins. The microbasin database was downloaded from the ANA metadada portal and was used to delimitate the drainage basin for each station.
2- Precipitation
Xavier et al. (2021) - Daily gridded meteorological data for Brazil Obtained in .nc format from the authors.
CHIRPS Daily: Climate Hazards Group InfraRed Precipitation With Station Data (Version 2.0 Final). Obtained through Google Earth Engine.
3- Evapotranspiration
PML_V2 0.1.7: Coupled Evapotranspiration and Gross Primary Product (GPP)
MOD16A2: MODIS Global Terrestrial Evapotranspiration 8-Day Global 1km
4- Water storage
GRACE Monthly Mass Grids - Land
CAMELS-BR dataset
1- Hydrological data
The hydrological data was obtained from the Hydroweb portal from the Brazilian Water Agency (ANA). The total database contains 897 measuring stations. Out of these, we selected 68 stations localized within the bounds of the study area, and that have good quality and long-term data availability. The CAMELS-BR database also provides a shapefile of with the drainage basins for each of the measurement points.
2- Precipitation
The CAMELS-BR database includes precipitation data from CHIRPS v2.0 (Funk et al., 2015), CPC (NOAA, 2019a), and MSWEP v2.2 (Beck et al., 2019).
3- Evapotranspiration
The CAMELS-BR database includes actual evapotranspiration data from GLEAM v3.3a and MGB South America (Siqueira et al., 2018).
4- Water storage
The CAMELS-BR database does not contain water storage data.
Data proceessing
The Google Earth Engine scripts used for proceessing the remote
sensing data is available at
users/trasegis/traseh20/O3data.
The R script used for processing the hydrological data is available
at
trase\data\brazil\indicators\environmental\water\Traseh2obasins.R.
The R script used for processing the precipitation data is available
at
trase\data\brazil\indicators\environmental\water\precipitation_municipality.R.
The R script that organizes the datasets for use in this document is
available at
trase\data\brazil\indicators\environmental\water\waterbalanceBrazil.R.
Hydrological data
Stations and drainage basins of analysis
Methods based on Ruhoff et al (2022)
| Name | Code | Basin | Drainage_area_km2 |
|---|---|---|---|
| 01 ALTAMIRA | 18850000 | RIO AMAZONAS | 448000 |
| 02 BARRA | 46998000 | RIO SÃO FRANCISCO | 425000 |
| 03 BOA SORTE | 18460000 | RIO AMAZONAS | 210000 |
| 04 CACHOEIRA DA MANTEIGA | 42210000 | RIO SÃO FRANCISCO | 107000 |
| 05 DESCARRETO | 23700000 | RIO TOCANTINS | 297000 |
| 06 MIRACEMA DO TOCANTINS | 22500000 | RIO TOCANTINS | 183355 |
| 07 PIRAPORA BARREIRO | 41135000 | RIO SÃO FRANCISCO | 62200 |
| 08 SANTA MARIA DA BOA VISTA | 48290000 | RIO SÃO FRANCISCO | 535000 |
| 09 SÃO FRANCISCO | 44200000 | RIO SÃO FRANCISCO | 184000 |
| 10 SÃO ROMÃO | 43200000 | RIO SÃO FRANCISCO | 154000 |
| 11 TRAIPU | 49660000 | RIO SÃO FRANCISCO | 630000 |
| 12 UHE TUCURUÍ BARRAMENTO | 29680080 | RIO TOCANTINS | 764000 |
CAMELS-BR
Data availability for hydrological stations
Methods based on Ruhoff et al (2022)
CAMELS-BR
Precipitation data
For the precipitation data, two data sources were obtained through Google Earth Engine, and three through the CAMELS-BR dataset. The Chirps dataset is present in both sets.
Here the comparison between the mean monthly data points for the five datasets, for each basin of analysis, is presented.
First, the mean annual precipitation for all datasets is shown below.
And then, the mean monthly precipitation for all datasets is shown
below.
The scatterplot shows the relation between each dataset and the ensemble mean of all the datasets.
## `geom_smooth()` using formula 'y ~ x'
Since there are two sources of Chirps data, the graph below shows how our estimates obtained through Google Earth Engine compare with the CAMELS-BR database.
Finally, the graphs below show the difference between the mean
monthly data points for two datasets over time, expressed by
prec(Xavier et al) - prec(Chirps). It is evident that the
differences between them are larger in the more recent years.
Evapotranspiration data
Our evapotranspiration estimates were obtained through Google Earth Engine from two sources: PML and MODIS. The CAMELS-BR database contains data from GLEAM (Martens et al., 2017) and MGB-IPH model (Pontes et al., 2017). The two graphs below show the degree of similarity between them for each basin of analysis.
The first graph shows the monthly mean evapotranspiration for each basin of analysis, and each evapotranspiration product.
The second graph shows the annual evapotranspiration for each basin of analysis, and each evapotranspiration product.
The third graph shows the relationship between each evapotranspiration product, and the ensemble mean between them.
## `geom_smooth()` using formula 'y ~ x'
Water storage
The GRACE dataset of water storage presents three different spherical harmonic solutions - from GFZ (GeoForschungsZentrum), CSR (Center of Space Research from the Universityof Texas), and NASA JPL (Jet Propulsion Laboratory).
The graph below compares the results from three different methods with their simple average. The average is calculated by taking the mean of the three datasets for each point of analysis.
Considering that the results of the three solutions are very similar, in the water balance analysis only the average of the three datasets will be used.
Water balance analysis
An overall view of the water balance in the selected basins is shown below, with the monthly values for each of the variables.
And here, an overall view of the water balance in the selected basins through the annual cycle.
To analyze the water balance, and more especifically the Evapotranspiration Products, we estimated ET based on the following equation:
\[ ET = P - \frac{\delta S}{\delta t} - Streamflow \]
The scatterplots below shows the comparison of the measured ET values, with the values estimated though the water balance equation.
The ET measured values were obtained from GLEAM, MGB, MODIS, and PML. The ensemble mean of the ET values is also shown. The ET estimated values through the water balance was estimated with the use of different precipitation products, and the GRACE satellite data.
The first scatterplot shows the relationship between measured and estimated ETs for each basin. It is evident that the results are much more poor in some of the basins, the main example being the 18850000. This is the hydrological station at the end of the Xingu river; it has been discussed in previous studies that the evapotranspiration estimated by remote sensing tends to be in the amazon area tends to be more poorly matched to observations.
The following scatterplot and table shows this relationship, but for each precipitation and ET product combination. The table shows the RMSE and R2 values for each combination. The RMSE and R2 values are shown in the table.
“RMSE and R values for each combination of precipitation and ET products
|source_p |ensemble |GLEAM- CAMELSBR |MGB- CAMELSBR |MODIS |PML | |:—————–|:————|:—————|:————-|:————|:————| |Chirps |54.28 - 0.45 |54.16 - 0.45 |54.1 - 0.43 |57.09 - 0.36 |56.94 - 0.49 | |Chirps - CAMELSBR |56.37 - 0.44 |56.14 - 0.44 |56.1 - 0.42 |59.01 - 0.36 |59.19 - 0.49 | |mswep - CAMELSBR |62.55 - 0.43 |62.11 - 0.43 |61.92 - 0.42 |64.96 - 0.34 |65.68 - 0.47 | |cpc - CAMELSBR |53.07 - 0.3 |55.1 - 0.28 |52.96 - 0.3 |57.23 - 0.23 |52.2 - 0.35 | |Ensemble |55.26 - 0.41 |55.58 - 0.4 |54.87 - 0.4 |58.29 - 0.33 |57.38 - 0.46 | |Xavier |56.83 - 0.4 |57.28 - 0.39 |56.2 - 0.39 |59.72 - 0.32 |59.09 - 0.43 |
Reference list
Chagas, V. B. P., Chaffe, P. L. B., Addor, N., Fan, F. M., Fleischmann, A. S., Paiva, R. C. D., and Siqueira, V. A.: CAMELS-BR: hydrometeorological time series and landscape attributes for 897 catchments in Brazil, Earth Syst. Sci. Data, 12, 2075–2096, https://doi.org/10.5194/essd-12-2075-2020, 2020.
Martens, B., Miralles, D.G., Lievens, H., van der Schalie, R., de Jeu, R.A.M., Fernández-Prieto, D., Beck, H.E., Dorigo, W.A., and Verhoest, N.E.C.: GLEAM v3: satellite-based land evaporation and root-zone soil moisture, Geoscientific Model Development, 10, 1903–1925, doi: 10.5194/gmd-10-1903-2017, 2017.
Pontes, Paulo & Fan, Fernando & Fleischmann, Ayan & Cauduro Dias de Paiva, Rodrigo & Buarque, Diogo & Siqueira, Vinícius & Jardim, Pedro & Sorribas, Mino & Collischonn, Walter. (2017). MGB-IPH model for hydrological and hydraulic simulation of large floodplain river systems coupled with open source GIS. Environmental Modelling and Software. 94. 1-20. 10.1016/j.envsoft.2017.03.029.
Ruhoff, Anderson & Comini, Bruno & Laipelt, Leonardo & Fleischmann, Ayan & Siqueira, Vinícius & Moreira, Adriana & Barbedo, Rafael & Cyganski, Gabriele & Fernandez, Gabriel & Breda, Joao & Paiva, Rodrigo & Meller, Adalberto & Teixeira, Alexandre & Araújo, Alexandre & Fuckner, Marcus & Biggs, Trent. (2022). Global Evapotranspiration Datasets Assessment Using Water Balance in South America. Remote Sensing. 14. 2526. 10.3390/rs14112526.