GPM IMERG rainfall data organization
Ries Plescher
rplesch@umich.edu
May 31, 2024


This resource contains GPM IMERG data for central Nepal, as well as a "scaled" product produced by scaling the GPM IMERG to match physical gauge records. The process is described in more detail below. The GPM IMERG data is contained in a MATLAB variable, which can be accessed using the function latlon2rainfall.m. Unscaled and Scaled GPM IMERG is also available in .csv files corresponding to each FRES, DHM, and ICIMOD station.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

RAW DATA

File Structure:  

The raw data consists of 13 folders containing V07A GPM IMERG Final Run data downloaded in November and December 2023 from a file subset generated using NASA GES DISC. Each folder contains one year of 30m netCDF files (e.g. 365*48=17520 files, 17568 for a leap year). These file folders are too large to include on Hydroshare, so a processed MATLAB 3D array is included, described below.

Time Bounds: 1/1/2010 - 12/31/2022
Latitude Bounds: 26.7 - 29.2 Degrees North
Longitude Bounds: 82.8 - 88.3 Degrees East


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

PROCESSED DATA:

The MATLAB struct array has the following fields:

•	rainfall: a 3D matrix of precipitation intensity (mm/hr) with dimensions [lat x lon x time]
•	lat_bnds: upper and lower latitude bounds corresponding to dimension 1 of the rainfall array
•	lon_bnds: upper and lower longitude bounds corresponding to dimension 2 of the rainfall array
•	time_bnds: upper and lower temporal bounds corresponding to dimension 3 of the rainfall array, in seconds since Jan. 6, 1980
•	date_bnds: equal to time_bnds, but converted to a MATLAB datetime matrix for convenience
•	year: year of rainfall data

These structs are saved in the file GPM_IMERG_master_struct.mat, as an array of 13 structs, one for each year 2010-2022.

GPM IMERG data for a pair of coordinates can be accessed using the companion function latlon2rainfall.m, which takes a GPM struct as in input as well as a pair of coordinates and interpolates rainfall between the four surrounding 0.1x0.1 degree IMERG cells. Detail on interpolation can be found in IMERG_interpolation_method.png.

latlon2rainfall is used to save a GPM IMERG rainfall record for each FRES, DHM, and ICIMOD station used in the FRES project for all 13 years. Any value under a 0.2mm/hr noise threshold recommended by NASA is set to 0.

File Structure: ___GPM_IMERG_processed.csv

Column Name	Description
datetime_start_UTC		Datetime of beginning of rainfall period, UTC, format M/D/YYYY H:MM:SS
datetime_end_UTC		Datetime of end of rainfall period, UTC, format M/D/YYYY H:MM:SS
year_end			Year of the end of the rainfall period, UTC
month_end			Month of the end of the rainfall period, UTC
day_end				Day of the end of the rainfall period, UTC
time_end_UTC			Time of the end of the rainfall period, UTC, format H:MM:SS
interpolated_GPM_IMERG_mm	Millimeters of rainfall measured in the 30 minute period between datetime_start_UTC and datetime_end_UTC


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

SCALED DATA:

A "scaled" GPM IMERG product is also calculated for each DHM, FRES, and ICIMOD station, as the IMERG algorithm is known to have error in high relief areas such as Nepal. To calculate the scaled GPM IMERG, a 24hr moving sum is calculated from both the gauge station* and the corresponding interpolated GPM IMERG timeseries. The scaling factor is calculated as the ratio between the gauge sum and the GPM IMERG sum, giving a distinct scaling factor for every 30 minute period. The scaled GPM IMERG product is the multiplication of the scaling factor and the original GPM IMERG timeseries.

*Stations with daily totals are upscaled to a 30 minute time series before the moving sum is calculated, using the assumption of uniform rainfall so that each 30 minute period is exactly 1/48th of the daily total.

File Structure: ___GPM_IMERG_scaled.csv

Column Name	Description
datetime_start_UTC		Datetime of beginning of rainfall period, UTC, format M/D/YYYY H:MM:SS
datetime_end_UTC		Datetime of end of rainfall period, UTC, format M/D/YYYY H:MM:SS
year_end			Year of the end of the rainfall period, UTC
month_end			Month of the end of the rainfall period, UTC
day_end				Day of the end of the rainfall period, UTC
time_end_UTC			Time of the end of the rainfall period, UTC, format H:MM:
scaled_GPM_IMERG_mm		Millimeters of scaled rainfall in the 30 minute period between datetime_start_UTC and datetime_end_UTC