The National Water Agency dataset of Resevoir Transpiration

Estimation method

The net evaporation is estimated as

EΔ = EL - ER

Where:

EΔ is the net water body evaporation; EL is the evaporation in the surface of the water body; and ER is the real evapotranspiration that would occur in the same area, if the water body was absent.

The EL value was estimated from the MOD16A/MODIS product based on the Penman–Monteith equation (Mu et al.,2011).

The ANA database of water masses

the Shapefile of ANA-mapped water masses (geoft_bho_massa_dagua_v2019.shp) is integrated currently into MapBiomas. The ANA dataset of water masses contains rivers, lakes, reservoirs, and others - any water body that can be mapped as a polygon, rather than a line or point.

The water body dataset has a unique identifier esp_cd, which identifies each water mass. This key does not have a complex codification method, and it is only a number that goes from 1 to 244539. In the water masses dataset there are fields for the municipality and ottobacia where the water mass is located, but these fields are mostly empty. The municipality and ottobacia of each impoundment would therefore have to be inferred topologically.

The following methodology was used for mapping the water masses, according to the Nota Técnica Nº 52/2020/SPR:

“A metodologia utilizada envolveu a agregação de planos de informação da base de dados Global Surface Water (GSW) que utiliza imagens Landsat para o bioma Caatinga e na região conhecida como MATOPIBA (divisa dos Estados do Maranhão, Tocantins, Piauí e Bahia), do mapeamento de massas d’água a partir de imagens RapidEye para o Cadastro Ambiental Rural (CAR) do Serviço Florestal Brasileiro (SFB) executado pela Fundação Brasileira para o Desenvolvimento Sustentável (FBDS) para os biomas Mata Atlântica e Cerrado (exceto Maranhão e Piauí), além do emprego da metodologia proposta e adaptada por Ferreira (2017) utilizando imagens RapidEye na plataforma Google Earth Engine para o bioma Pampa. No caso das imagens RapidEye foi considerado, após testes e avaliação da metodologia, um limiar de 0,5 ha de área superficial das feições para inserção na base e, no caso das imagens Landsat, de 4 hectares.”

A brief description of the Ferreira (2017) method:

“No âmbito do Programa de Formação Avançada da ANA, o autor adaptou a metodologia de Tetteh e Schonert (2015) para a plataforma de processamento em nuvem Google Earth Engine baseada em linguagem JavaScript num experimento com 19 imagens RapidEye distribuídas por diferentes regiões do Brasil, mais ou menos adensadas em termos da presença de massas d’água de origem natural ou artificial.”

Some stats

Impoundment type

The total water area mapped by ANA including all types of water bodies is of 173750 km^2.

The area covered by the 173707 bodies of water presented in the evp_liquida_reservatorios_anualOK2 dataset is distributed among the following main uses:

As we can see, a lot of this data is still not categorized among the classes of reservoir water use. The class with larger amount of water bodies at this point is “NA”. Additionally, the classes have some overlap between each other.

It is likely that the hydroelectric and “human supply” (urban water supply) reservoirs are well categorized, but otherwise this information is not entirely easy to trust at face value. The bulk of water bodies that are of interest to us are not categorized, and would have to be categorized through analysis of surrounding land use.

Impoundment size

In the evp_liquida_reservatorios_anualOK2.xlsx file, there are evaporation values for a total of 173707 impoundments. The largest impoundment mapped is the UHE Sobradinho, with 313 thousand hectares.

About 84% of the impoundments are smaller than 5 hectares in size, and 92% are smaller than 10ha. The distribution of impoundment area for impoundments below 5 ha is shown below.

Start of operation

Another important information that exists in this dataset is when the reservoir was built or started to be operated.