This thesis validates three different satellite products (TRMM, GPM, GOES) for the Paute river basin. Ground truthing implies that the correlation between gauged and satellite precipitation estimates can be best explained in a spatial context.
Comparison of multiple satellite derived precipitation estimates in a mountainous basin. A case study in the Paute river basin – Ecuador.
In many remote parts of the world and particularly in developing countries and mountain areas, rain gauges are sparse or non-existent. The "Prediction in Ungauged Basins" (PUB) initiative was launched by the IAHS to improve hydrological predictions worldwide. An interesting area for PUB is the Andes of Ecuador, as an example of a sparsely gauged area. These mountains are covered by the very vulnerable páramo ecosystem, which provide drinking water for many people. Reliable precipitation estimates are needed for good hydrological predictions and water resource management in the Paute river basin in Ecuador. Satellite products provide a great addition to precipitation estimates worldwide but the large spatial variability in precipitation in mountainous regions is a great challenge. More information is needed about the accuracy of satellite precipitation estimates and the hydrological function of the páramo. Therefore the main goal of this thesis is to validate three different satellite precipitation products for the Paute basin with ground truthing. Data collection took place during November 2014-January 2015 in the Tomebamba subcatchment of the Paute river basin. Fieldwork was done to investigate small scale spatial variability in precipitation and changes in soil moisture of the páramo. Reference precipitation data of 0.1 mm tipping buckets for the period September 2013-January 2015 were compared with daily and monthly satellite products of the Tropical Rainfall Measuring Mission (TRMM) and the Global Precipitation Measurement (GPM) mission. Also a historical dataset of the years 2005-2006 was compared with TRMM for the Paute basin. The third satellite (GOES) was excluded from further research during the analysis phase because of the high amount of missing daily products and unreliable data. Field measurements showed a high spatial variation in precipitation and a strong correlation between precipitation and soil moisture of the páramo (average R2 of 0.90). The reference precipitation data showed that a division in rain gauges can be made corresponding to the rainfall regimes. There were hardly any dry days in the páramo region. Dry days did occur in the valley deep below. GPM and TRMM were compared for the Tomebamba subcatchment, showing that GPM has the tendency to overestimate the amount of precipitation when the rainfall intensity is low. TRMM shows less overestimation, but is better in estimating the precipitation at higher intensity rainfall events. The gauged and satellite precipitation estimates in mm/day were compared. GPM detected on average 3.3 mm/day, TRMM 2.1 mm/day and the tipping buckets 2.9 mm/day. Based on the high spatial variation in precipitation, a satellite product with a high resolution (GPM) is preferred. The validation of TRMM for the Paute river basin showed that the correlation was best on top of the mountain system (average R2 of 0.79) and deep below in the valley (average R2 of 0.75). The lowest R2 values (average of 0.52) could be found along the mountain slopes, which could possibly be explained by the complex cloud systems along the mountain slopes. Co-Kriging with a digital elevation model was done to look at the differences in regional estimates between TRMM and in situ precipitation measurements. For further studies it is recommended to analyse a longer dataset of TRMM and GPM to strengthen the conclusions of this thesis. Taking in to account the relatively short time series for analysis, it can be concluded that precipitation estimates of satellite images are reasonably accurate for the Paute river basin.