![]() The PERSIANN-CDR algorithm uses the existing PERSIANN algorithm as its backbone model. The Precipitation Estimation from Remotely Sensed Information using Artificial Neural Network – Climate Data Record (PERSIANN-CDR) provides long-term ( to delayed present) near-global (60°S-N latitude and 0°-360° longitude bands) precipitation data at daily and 0.25° temporal and spatial resolutions. For a more detailed comparison of PERSIANN- CDR, TRMM 3B42, GPCP-1DD and other precipitation estimates see Gehne et al. For climate change studies PERSIANN CDR is better suited than TRMM 3B42. TRMM 3B42 is a high-resolution precipitation product and thus does not have the same homogeneity goals as the GPCP-1DD or PERSIANN CDR. It will not be extended past the current end date. TRMM 3B42 is of higher temporal resolution (3 hourly), and available from 1998-2014. Correlations between the time series of PERSIANN CDR and GPCP-1DD averaged over continental areas are above 0.9 for daily, monthly, and annual averages (Fig. 8 and 9, while daily correlations over continental regions are higher than 0. Even though the monthly means are adjusted to the monthly GPCPv2.2 product, the daily distributions of the GPCP-1DD product and PERSIANN CDR have very different behavior as can be seen below in Figs. With the main differences between GPCP-1DD and PERSIANN CDR in the long term monthly means are over the tropical oceans (Figs. For large scale averages GPCP-1DD and PERSIANN CDR show very similar variability (Fig. GPCP-1DD is of comparable quality, though it has slightly coarser spatial resolution (1 degree ) and only starts in 1996. More studies concerned with evaluation and verification of this data set are necessary to further identify potential issues and possible limitations.Ĭomparable data sets include the GPCP-1DD and TRMM 3B42 products. At the point of writing the PERSIANN CDR has been available to users for only a relatively short period of time. The adjustment to GPCPv2.2 monthly means (which include rain gauge measurements) likely alleviates this problem to some extent, but spurious trends in GPCPv2.2 will also be found in this data set. Spurious trends in these data are likely to translate to the precipitation rates estimated by the PERSIANN al- gorithm. PERSIANN CDR is heavily based on infrared satellite retrievals from multiple satellites. The data set also provides users with long time series of precipitation rates at up to daily temporal resolution. 2015)), and estimating precipitation distributions over regional and continental areas. This data set is of high enough spatial resolution for studies of high-impact precipitation events (e.g. With more than 30 years in the data record the long-term, consistent, data set is intended for climatological studies of the hydrological component of the climate system. While the direct input data for PERSIANN-CDR are GridSat-B1 infrared data (and training of the ANN model is done on the NCEP stage IV hourly precipitation data), the adjustment to GPCPv2.2 monthly means indirectly includes rain gauge measurements as well. Details of the algorithm can be found in Ashouri et al. ![]() 5 degree and temporally to monthly scales match the GPCPv2.2 monthly values. 5 degree grid box for each month and each year, so that the PERSIANN-CDR values averaged spatially over 2. The high-resolution values of PERSIANN-CDR are corrected in each 2. Bias adjustment of the PERSIANN-CDR precipitation estimates are applied by matching the Global Precipitation Climatology Project (GPCP) monthly product version 2.2 (GPCPv2.2) at 2. For the retrospective estimation of rainfall rates the nonlinear regression parameters of the ANN model remain fixed after training. ![]() The artificial neural network (ANN) model is trained to associate variations in the GridSat-B1 brightness temperature of cold-cloud pixels and their surroundings with the surface rain rate from the National Centers for Environmental Predic- tion (NCEP) stage IV hourly precipitation data. 25 degrees in the latitude band 60 S - 60 N from 1983 to the delayed present. The Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks- Climate Data Record (PERSIANN-CDR) provides daily rainfall estimates at a spatial resolution of 0.
0 Comments
Leave a Reply. |