MSWEP
Multi-Source Weighted-Ensemble Precipitation
Overview
Accurate
Based on state-of-the-art gauge, satellite, and reanalysis data sources
Gauge-corrected
Daily corrections using observations from over 77,000 gauges
Near real-time
Near real-time availability to support time-critical applications
Transparent
Methods published in open-access peer-reviewed scientific journals
Performance
Reference: Stage-IV gauge-radar data
Values represent the mean daily temporal R2 over the US
Adapted from Beck et al. (2019)
Reference: 75,540 gauges worldwide
Values represent the mean 3‑day temporal R2
*Version without gauge corrections
Publication in preparation
Methodology
- Beck, H. E., Wood, E. F., Pan, M., Fisher, C. K., Miralles, D. M., van Dijk, A. I. J. M., McVicar, T. R., and Adler, R. F. MSWEP V2 global 3‑hourly 0.1° precipitation: methodology and quantitative assessment Bulletin of the American Meteorological Society 100(3), 473–500, 2019
Data license
Frequently asked questions
- to merge gauge, satellite, and reanalysis precipitation estimates to enhance the performance in densely gauged, convection-dominated, and frontal-dominated weather regimes, respectively;
- to incorporate daily gauge observations and account for gauge reporting times (which minimizes temporal mismatches between the satellite-reanalysis estimates and the gauge observations);
- with compatible medium-range forecast ensembles (updated daily) and seasonal forecast ensembles (updated monthly; see MSWX).
The ‘Past’ and ‘Past_nogauge’ variants represent the historical satellite-reanalysis merge including and excluding gauge corrections, respectively. The ‘NRT’ variant represents the near real-time extension of the historical record to the present (with a latency of ~3 hours). We recommend using the ‘Past_nogauge’ variant in precipitation product performance evaluations using gauge observations as reference, and the ‘Past’ variant for any other purpose.
- Download and install rclone.
- Link rclone to your Google account by following the steps in this video.
- Access the MSWEP shared folder by visiting it via your browser. Check if the shared folder is listed under “Shared with me” on your Google Drive page.
- Confirm that rclone can find the shared folder:
$ rclone lsd --drive-shared-with-me GoogleDrive: -1 2021-02-03 10:14:35 -1 MSWEP_V280
- Download daily and monthly MSWEP data from the shared folder to your local drive:
If the download is interrupted, the command can be run again, and files that already exist will be skipped.$ rclone sync -v --exclude 3hourly/ --drive-shared-with-me GoogleDrive:/MSWEP_V280 c:/temp/MSWEP_V280 2021/02/03 11:09:02 INFO : Past/Monthly/202007.nc: Copied (new) 2021/02/03 11:09:02 INFO : Past/Monthly/202002.nc: Copied (new) 2021/02/03 11:09:02 INFO : Past/Monthly/202005.nc: Copied (new) ...
Medium-range and seasonal forecast data from GloH2O’s Multi-Source Weather (MSWX) product is compatible with MSWEP. MSWX can thus be used to extend MSWEP into the future. Note, however, that some inconsistencies between MSWEP and MSWX may be present due to the inclusion of gauge and satellite data in MSWEP.
rclone sync -v --drive-shared-with-me GoogleDrive:/MSWEP_V280/Past/Daily/2020116.nc ./
The data are read and plot using MATLAB as follows:
global_precip = ncread('2020116.nc','precipitation')';
imagesc(global_precip,[0 30]);
colorbar
title('Precipitation on April 25, 2020 (mm/day)')
The same data are read and plot using Python as follows:
from netCDF4 import Dataset
import matplotlib.pyplot as plt
dataset = Dataset('2020116.nc','r')
global_precip = dataset.variables['precipitation'][:]
dataset.close()
plt.plot(global_precip,vmin=0,vmax=30)
plt.colorbar()
plt.title("Precipitation on April 25, 2020 (mm/day)")
plt.show()
We did not correct for gauge under-catch in the latest version of MSWEP (unlike in previous versions) for two main reasons: (i) in our experience, users tend to give preference to precipitation estimates that match gauge observations as closely as possible; and (ii) under-catch corrections are generally subject to considerable uncertainty. To obtain a bias-corrected version of MSWEP — important for hydrological modeling — we recommend using GloH2O’s PBCOR product.
MSWEP V1 lacks cumulative distribution function (CDF) corrections, and therefore exhibits significant drizzle and less reliable trends due to changes in the data sources through time. Conversely, MSWEP V2 matches the CDF of each dataset combination to a single reference CDF, resulting in less drizzle and a more homogeneous record. See the technical documentation for the full list of changes.
Precipitation product evaluations carried out at the daily time scale should account for the reporting times of gauges (i.e., the end times of the daily accumulations). Disregarding reporting times in daily evaluations can result in significant temporal mismatches between the product and the gauge, which can affect the results of the evaluation. Accounting for reporting times requires the use of sub-daily data to compute new daily accumulations that correspond to the reporting times of the gauges. Alternatively, the evaluation can be carried out at a 3‑day (instead of daily) time scale, which reduces the impact of potential mismatches. See Beck et al. (2019) for maps of reporting times for GHCN‑D and GSOD gauges.
There is currently no other way to download MSWEP except using rclone.