In the last decade, satellite precipitation estimation techniques have reached significant improvement in quantitative description of rainfall intensity and distribution, and even higher performances are expected from the full exploitation of Global Precipitation Measurement Mission. Parallel to the development of new techniques the need of accurate and reliable ground reference fields is also growing, for both calibrating and validating satellite algorithms. In the frame of the EUMETSAT’s Satellite Application Facility for the Support to Operational Hydrology and Water Management (H-SAF) different satellite estimation techniques are developed with the aim to provide instantaneous to 24h cumulated products to assist hydrological implementations, supported by a careful validation activity. Following the EUMETSAT guidelines, raingauges (and/or radar) have to be used as reference for satellite techniques validation. The different satellite and raingauge views of the precipitation field, however, pose several problems when they have to be compared: several factors cumulate each other throughout the matching process, resulting in large discrepancies between the two rainfall fields. In this work, we evaluate the impact of two of these factors in the satellite estimation validation process, taking advantage of the availability of locally-dense, high-resolution (1 minute) raingauge data over Italy. We first estimate the error due to the different time and spatial sampling between raingauges and satellite products (often referred as representativeness error), and then we evaluate the impact of raingauge density variations on the interpolated maps obtained by different techniques. Results show that a careful selection of the raingauge network density, time sampling, and raingauge interpolation algorithms can greatly reduce such matching errors that, however, can be very high. A Fractional Standard Error ranging from 100% to 300% between satellite estimate and raingauge reference value can be due to different raingauge sampling and interpolation strategy, while a reduction in the raingauge density of a factor of 2 leads to a degradation of the quality of the interpolation of about 40%
On the uncertainties in validating satellite instantaneous rainfall estimates with raingauge operational network
PORCU', Federico;MILANI, Lisa;Petracca, Marco
2014
Abstract
In the last decade, satellite precipitation estimation techniques have reached significant improvement in quantitative description of rainfall intensity and distribution, and even higher performances are expected from the full exploitation of Global Precipitation Measurement Mission. Parallel to the development of new techniques the need of accurate and reliable ground reference fields is also growing, for both calibrating and validating satellite algorithms. In the frame of the EUMETSAT’s Satellite Application Facility for the Support to Operational Hydrology and Water Management (H-SAF) different satellite estimation techniques are developed with the aim to provide instantaneous to 24h cumulated products to assist hydrological implementations, supported by a careful validation activity. Following the EUMETSAT guidelines, raingauges (and/or radar) have to be used as reference for satellite techniques validation. The different satellite and raingauge views of the precipitation field, however, pose several problems when they have to be compared: several factors cumulate each other throughout the matching process, resulting in large discrepancies between the two rainfall fields. In this work, we evaluate the impact of two of these factors in the satellite estimation validation process, taking advantage of the availability of locally-dense, high-resolution (1 minute) raingauge data over Italy. We first estimate the error due to the different time and spatial sampling between raingauges and satellite products (often referred as representativeness error), and then we evaluate the impact of raingauge density variations on the interpolated maps obtained by different techniques. Results show that a careful selection of the raingauge network density, time sampling, and raingauge interpolation algorithms can greatly reduce such matching errors that, however, can be very high. A Fractional Standard Error ranging from 100% to 300% between satellite estimate and raingauge reference value can be due to different raingauge sampling and interpolation strategy, while a reduction in the raingauge density of a factor of 2 leads to a degradation of the quality of the interpolation of about 40%I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
