In situ ground observation measurement of precipitation is difficult in sparsely populated areas with trying access conditions, as is the case in many countries in Africa. The use of remote sensors installed in satellites can be very useful in overcoming this challenge, enabling the improvement of the spatial variability description of this variable and the extension of data series. A number of standard products offering precipitation estimates on a regular basis is now available and may be used for water planning and management purposes. The present study examines the performance of four of these products in Angola, namely the Tropical Rainfall Measuring Mission (TRMM) 3B43 (version 6), Global Precipitation Climatology Project (GPCP) Combined Precipitation Data Set (version 2.2), Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) and the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Centre (CPC) Morphing Technique (CMORPH), by comparing annual and monthly precipitation estimates with ground observation measurements. The data set of precipitation ground observation measurements was collected by the authors from different sources in Angola and Portugal, and is the result of an intense effort to gather hydrological records from Angola. It is believed to be one of the most complete data sets of monthly precipitation data from Angola. The four remote-sensing products are able to describe the main features of the spatial and temporal variability of annual and monthly precipitation in Angola. The results also show that the estimates from the TRMM are more accurate than the estimates offered by the other products, a conclusion which is in line with previous studies and which may be explained by the fact that this is the first product to incorporate measurements from precipitation radar. The estimation bias of TRMM is also more consistent which means that the results presented in the present study can be used in an operational environment to reduce the precipitation estimation error.

The study aims to understand when the effort required to run a distributed and detailed water balance model leads to effective better results, when compared with a lumped, data-parsimonious, and easier-to-apply hydrological model. The study compares the results obtained from a lumped hydrological model with the results from soil water assessment tool (SWAT) when used to estimate stream flow values from precipitation and climate data. Both models use the curve number (CN) concept, determined from land use, soil hydrologic group, and antecedent soil moisture conditions, and were run with a daily time step. The lumped model input variables and parameters are considered to be uniformly distributed throughout the watershed, while the SWAT model requires an extensive and detailed spatial data, including weather data (rainfall, temperature, humidity, wind, and solar radiation), physical characteristics of the basin (topography and drainage network), as well as soil and land use data. The effort required to fully understand SWAT formulation, to obtain the needed data and to calibrate the model, is time consuming and does not lead to significant better results, when compared with the lumped model results. The results confirm that in many situations, the simpler lumped model leads to reliable values of hydrological variables, both at a daily and monthly scale. The models’ predictability power is higher at a monthly time step, because both models fail to reproduce the daily flow values fully in a consistent way, although they are able to reproduce the overall variability. Lack of data to describe the precipitation accurately may explain the poorer model performance at the daily time step. The study concludes that when available data fail to represent the spatial distribution of the input variables and model parameters, there is little advantage in using the more complex distributed model.