floodpeople/KazFlow85_dataset

Dataset Documentation

Overview

Dataset Name: KazFlow85_dataset

Short Description: This dataset consists of meteorological (time series) and geophysical (catchment attributes) data of 85 basins of Kazakhstan. It is intended for use in weather forecasting or modeling, as well as flood prediction based on the attributes provided.

Long Description: We developed basin scale hydrometeorological forcing data for 85 basins in the conterminous Kazakhstan basin subset. Retrospective model forcings are computed from ERA5-Land forcing data run from 1 Jan 2000 to 31 Dec 2022. Model timeseries output is available for the same time periods as the forcing data.

Topographic characteristics (e.g. elevation and slope) were retrieved from MERIT data. Climatic indices (e.g., aridity and frequency of dry days) and hydrological signatures (e.g., mean annual discharge and baseflow index) were computed using the time series provided by Newman et al. (2015). Soil characteristics (e.g., porosity and soil depth) were characterized using the soilgrids-isric and HiHydroSoilv2_0 dataset. Vegetation characteristics (e.g. the leaf area index and the rooting depth) were inferred using MODIS data.

---

Shapefiles

The shapefiles folder contains subfolders for each basin, with each subfolder named using the basin’s unique identifier basin_id. Within every subfolder, five files describe the basin’s spatial data: .cpg (character encoding), .dbf (attribute data), .prj (projection information), .shp (geometry), and .shx (shape index). These files collectively define the basin’s geographical boundaries and associated metadata. The shapefiles are later used to retrieve attributes, such as solar radiation or elevation, from Google Earth Engine (GEE) by overlaying the basin geometries onto GEE’s datasets for spatial analysis.

Folder Structure

The dataset is organized into the following folders:

• attributes/: The collection geophysical data (or catchment attributes)

  • Contains 4 CSV files kazflow85_clim, kazflow85_topo, kazflow85_soil, and kazflow85_vege.
  • Sources: MODIS, MERIT, ESA, HiHydroSoilv2_0, soilgrids-isric datasets using Google Earth Engine.

• mean_basin_forcing/: Only meteorological data with daily temporal resolution

  • Contains 85 CSV files with the format [id].csv (e.g. 11001.csv, 11129.csv), where [id] stands for basin id.
  • Sources: "ECMWF/ERA5_LAND/DAILY_AGGR", "JAXA/GPM_L3/GSMaP/v6/operational", "UCSB-CHG/CHIRPS/DAILY".

• streamflow/: Hydro data feature, particularly discharge

  • Contains 85 CSV files with the format [id].csv (e.g. 11001.csv, 11129.csv), where [id] stands for basin id.
  • Sources: KazHydroMet website [link]

• time_series/: The merge of previous two data mean_basin_forcing and streamflow stored as .nc formatted files.

  • Contains 85 NetCDF files with the format [id].nc (e.g. 11001.nc, 11129.nc), where [id] stands for basin id.

---

Features

Dynamic (daily) meteorological attributes (mean_basin_forcing/)

Column Name	Description	Unit	Datatype
date	Date of observation	-	DateTime
prcp_{era/mswep/gsmap/chirps}	Daily Precipitation (basin-averaged)	mm/d	Float
t_mean	Daily average temperature of air at 2m above the underlying surface	ºC	Float
t_min	Daily minimum temperature of air at 2m above the underlying surface	ºC	Float
t_max	Daily maximum temperature of air at 2m above the underlying surface	ºC	Float
dew_mean	Temperature to which the air would have to be cooled for saturation to occur	ºC	Float
wind_speed	Wind speed at a height of 10m above the surface	m/s	Float
vp1	Vapor pressure computed using dew_mean	Pa	Float
vp2	Vapor pressure computed using dew_mean	Pa	Float
srad	Solar radiation adjusted by daylight hours (in seconds)	W/m²	Float

Dynamic (daily) hydrological attributes (streamflow/)

Column Name	Description	Unit	Datatype
date	Date of observation	-	DateTime
discharge	Daily volume of water flowing through river hydropost per drainage area	mm/d	Float

Static geophyscial catchment attributes (attributes/)

Column Name	Description	Unit	Datatype
	kazflow85_clim.csv
basin_id	Unique identifier for each basin	-	Integer
p_mean_{era/mswep/gsmap/chirps}	Mean daily precipitation (basin-averaged)	mm/d	Float
pet_mean	Mean daily potential evapotranspiration	mm/d	Float
aridity_{era/mswep/gsmap/chirps}	Ratio of mean precipitation to potential evapotranspiration	-	Float
p_seasonality_{era/mswep/gsmap/chirps}	Seasonality and timing of precipitation estimated using sine curve	-	Float
frac_snow_daily_{era/mswep/gsmap/chirps}	Fraction of precipitation as snow	-	Float
high_prec_freq_{era/mswep/gsmap/chirps}	Frequency of high precipitation events (days per year)	d/year	Float
high_prec_dur_{era/mswep/gsmap/chirps}	Average duration of high precipitation events	d	Float
low_prec_freq_{era/mswep/gsmap/chirps}	Frequency of low precipitation events (days per year)	d/year	Float
low_prec_dur_{era/mswep/gsmap/chirps}	Average duration of low precipitation events	d	Float
	kazflow85_soil.csv
basin_id	Unique identifier for each basin	-	Integer
soil_conductivity	Saturated soil hydraulic conductivity	cm/hr	Float
max_water_content	Maximum soil water holding capacity	m	Float
sand_frac	Fraction of sand in soil	%	Float
silt_frac	Fraction of silt in soil	%	Float
clay_frac	Fraction of clay in soil	%	Float
	kazflow85_topo.csv
basin_id	Unique identifier for each basin	-	Integer
elev_mean	Mean elevation of the basin	m	Float
slope_mean	Mean slope of the basin	m/km	Float
area_gages2	Basin area (from GAGES-II dataset)	km²	Float
	kazflow85_vege.csv
basin_id	Unique identifier for each basin	-	Integer
forest_frac	Fraction of basin covered by forest	-	Float
lai_max	Maximum monthly mean of the leaf area index	-	Float
lai_diff	Difference between the maximum and mimumum monthly mean of the leaf area index	-	Float
gvf_max	Maximum monthly mean of the green vegetation fraction	-	Float
gvf_diff	Difference between the maximum and mimumum monthly mean of the green vegetation fraction	-	Float

Data Collection and Preprocessing

Collection

• Data was collected from "ECMWF/ERA5_LAND/DAILY_AGGR", "JAXA/GPM_L3/GSMaP/v6/operational", "UCSB-CHG/CHIRPS/DAILY", KazHydroMet (meteo data) and MODIS, MERIT, ESA, HiHydroSoilv2_0, soilgrids-isric datasets (catchment attributes) using Google Earth Engine.
• Timeframe: [Jan 2000 - Dec 2022].

Preprocessing

• Missing values: All missing and invalid values were replaced by np.nan.
• Normalization: Discharge data was normalized by area of the basin and stored in mm/d (instead of m^3/s).

---

Support Contact

Madina Abdrakhmanova
ISSAI - Institute of Smart Systems and Artificial Intelligence, Astana, KZ
madina.abdrakhmanova@nu.edu.kz

References

Kratzert, F., Klotz, D., Shalev, G., Klambauer, G., Hochreiter, S., and Nearing, G.: Towards learning universal, regional, and local hydrological behaviors via machine learning applied to large-sample datasets, Hydrol. Earth Syst. Sci., 23, 5089–5110, https://doi.org/10.5194/hess-23-5089-2019, 2019.

All of the derivation function and code computations can be found via this GitHub link.

by Flood People