leharris3/ccrfcd-mrms-hrrr-env-2021-2025

1H gauge accumulation + MRMS/HRRR zarr dataset for the Desert Southwest

50+ MRMS+HRRR variables; 220+ gauges; 400k samples

---

NOTE: work in-progress.

This is a dataset for training and evaluating synthetic quantitative precipiation estimation (QPE) models. Given some input context (e.g., radar fields, envionrmental parameters), predict how much rain fell at a rain gauge site over some period of time. Concretely, we've gather and QC'd data from 220 tipping bucket gauges through Nevada for the years 2021-2026. These readings are then used as ground-truth for ~400k samples, temporally and spatially aligned to 13 MRMS radar feilds and 39 HRRR envionrmental parameters. This dataset can therefore support training diverse QPE models (point -> point, local-spatial context -> point, etc.)

Region of Interest

---

• Lattitude: 34.751857˚N, 37.103662˚N
• Longitude: 113.792819˚W, 116.146925˚W

File Structure

---

data/
├── ccrfcd_rain_gauge_metadata.csv               # gauge site metadata (lat, lon, name, id)
├── gauge_feature_cache.csv                      # cached MRMS product values at gauge locations
├── gauge_feature_cache_meta.json
├── norm_stats_train.json                        # per-channel mean/std for normalization
├── 2021-01-01_2025-07-25_gt.csv                 # ground-truth accumulation samples
├── 2021-01-01_2026-03-23_gt.csv
├── 2021-01-01_2025-07-25_hrrr_env_final_v8.csv  # point-sample dataset (v8)
├── 2021-01-01_2025-07-25_hrrr_env_final_v9.csv  # point-sample dataset (v9)
│
├── ccrfcd_gauges_raw_data/                      # raw per-gauge CSVs from CCRFCD
│   └── <YYYYMMDD-YYYYMMDD>-<id>_<name>_RAIN.csv
│
├── dem/                                         # digital elevation model on HRRR grid
│   ├── dem_hrrr.npy                             # elevation (235, 236) float32, meters
│   ├── dzdx_hrrr.npy                            # east-west slope
│   ├── dzdy_hrrr.npy                            # north-south slope
│   └── srtm_tiles/                              # raw SRTM source tiles
│
├── hrrr_zarr/<product>/<YYYYMMDD>.zarr/         # raw HRRR fields (22 products)
│   ├── data/       (N, 235, 236) float32
│   ├── timestamps/ (N,) float64 unix seconds
│   └── zarr.json
│
├── hrrr_zarr_derived/<product>/<YYYYMMDD>.zarr/ # derived HRRR fields (18 products)
│   ├── data/       (N, 235, 236) float32
│   ├── timestamps/ (N,) float64 unix seconds
│   └── zarr.json
│
└── mrms_zarr/<product>/<YYYYMMDD>.zarr/         # MRMS radar fields (13 products)
    ├── data/       (N, 235, 236) float32
    ├── timestamps/ (N,) float64 unix seconds
    └── zarr.json

Samples

---

Variable List

---

Name	Variable	Units	Point	Spatial	Temp. Resolution	Source	Notes
Gauge ID	gauge_idx	int	✅	–––	–––	CCRFCD	unique gauge id assigned by CCRFCD
Start Time	start_datetime_utc	datetime	–––	–––	2-min	MRMS	intervals for which MRMS 1H radar QPE are available; UTC
End Time	end_datetime_utc	datetime	–––	–––	2-min	MRMS	start time + 01:00:00
Gauge 1H Accumulation	gauge_acc_in	in	✅	–––	2-min	CCRFCD	rain gauge 1H accumulation; aligned to nearest MRMS 2-min interval
MRMS Radar 1H QPE	mrms_q3evap_qpe	in	✅	❌	2-min	MRMS	MRMS operational 1H radar-only QPE
Latitude	lat	degrees	✅	–––	–––	CCRFCD	latitude of rain gauge
Longitude	lon	degrees	✅	–––	–––	CCRFCD	longitude of rain gauge
10m U-Wind	10m_above_ground_UGRD	m/s	❌	✅	1-hour	HRRR Zarr
850mb U-Wind	850mb_UGRD	m/s	✅	✅	1-hour	HRRR Zarr
700mb U-Wind	700mb_UGRD	m/s	✅	✅	1-hour	HRRR Zarr
10m V-Wind	10m_above_ground_VGRD	m/s	❌	✅	1-hour	HRRR Zarr
850mb V-Wind	850mb_VGRD	m/s	✅	✅	1-hour	HRRR Zarr
700mb V-Wind	700mb_VGRD	m/s	✅	✅	1-hour	HRRR Zarr
850-700mb Mean Wind	850_700mb_mean_wind	m/s	✅	✅	1-hour	HRRR Zarr*	magnitude of mean u-v vector
Surface Pressure	surface_PRES	Pa	✅	✅	1-hour	HRRR Zarr
850mb Geopot. Height	850mb_HGT	gpm	✅	✅	1-hour	HRRR Zarr
700mb Geopot. Height	700mb_HGT	gpm	✅	✅	1-hour	HRRR Zarr
Freezing Level	highest_tropospheric_ freezing_level_HGT	m	✅	✅	1-hour	HRRR Zarr
LCL Height	level_of_adiabatic_ condensation_ from_sfc_HGT	gpm	✅	✅	1-hour	HRRR Zarr
Warm Cloud Depth	convective_warm _cloud_depth	m	✅	✅	1-hour	HRRR Zarr*
2m Temperature	2m_above_ground_TMP	K	✅	✅	1-hour	HRRR Zarr
925mb Temperature	925mb_TMP	K	✅	✅	1-hour	HRRR Zarr
850mb Temperature	850mb_TMP	K	✅	✅	1-hour	HRRR Zarr
700mb Temperature	700mb_TMP	K	✅	✅	1-hour	HRRR Zarr
500mb Temperature	500mb_TMP	K	✅	✅	1-hour	HRRR Zarr
2m Dewpoint	2m_above_ground_DPT	K	✅	✅	1-hour	HRRR Zarr
925mb Dewpoint	925mb_DPT	K	✅	✅	1-hour	HRRR Zarr
850mb Dewpoint	850mb_DPT	K	✅	✅	1-hour	HRRR Zarr
700mb Dewpoint	700mb_DPT	K	✅	✅	1-hour	HRRR Zarr
500mb Dewpoint	500mb_DPT	K	✅	✅	1-hour	HRRR Zarr
925-700mb DPT Diff	925mb_700mb_DPT	K	✅	❌	1-hour	HRRR Zarr*
925-700mb DPT Lapse	925mb_700mb_DPT_lapse	K	❌	✅	1-hour	HRRR Zarr*	DPT(925) - DPT(700); layer difference
925-700mb Mean DPT	925mb_700mb_mean_DPT	K	❌	✅	1-hour	HRRR Zarr*
850-700mb DPT Diff	850mb_700mb_DPT	K	✅	❌	1-hour	HRRR Zarr*
850-700mb DPT Lapse	850mb_700mb_DPT_lapse	K	❌	✅	1-hour	HRRR Zarr*	DPT(850) - DPT(700); layer difference
850-700mb Mean DPT	850mb_700mb_mean_DPT	K	❌	✅	1-hour	HRRR Zarr*
850-500mb DPT Diff	850mb_500mb_DPT	K	✅	❌	1-hour	HRRR Zarr*
850-500mb DPT Lapse	850mb_500mb_DPT_lapse	K	❌	✅	1-hour	HRRR Zarr*	DPT(850) - DPT(500); layer difference
850-500mb Mean DPT	850mb_500mb_mean_DPT	K	❌	✅	1-hour	HRRR Zarr*
Sfc Equiv. Pot. Temp	surface_theta_e	K	✅	✅	1-hour	HRRR Zarr*
Low-Level Mixing Ratio	lowest_100mb_mean_ mixing_ratio	kg/kg	✅	✅	1-hour	HRRR Zarr*
LCL-LFC Rel. Humidity	LCL_LFC_RH	%	✅	✅	1-hour	HRRR Zarr*	relative humidity between LCL and LFC
0-3km Rel. Humidity	0-3km_RH	%	✅	✅	1-hour	HRRR Zarr*	0-3 km (mean) relative humidity
0-5km Rel. Humidity	0-5km_RH	%	✅	✅	1-hour	HRRR Zarr*	0-5 km (mean) relative humidity
Downdraft CAPE	DCAPE	J/kg	✅	❌	1-hour	HRRR Zarr*	calculated w/ sparse pressure levels: {src, 925, 850, 700, 500}; clip$\text{DCAPE} \in [0, 2000]$ j/kg
Low-Level Lapse Rate	low_level_lapse_rate	K/km	✅	✅	1-hour	HRRR Zarr*
3hr Lapse Rate Change	3hr_lapse_rate_change	K/km	✅	❌	1-hour	HRRR Zarr*
Sfc-850mb PWAT	sfc_850_pw	mm	✅	✅	1-hour	HRRR Zarr*	pwat, surface to 850mb
Sfc-700mb PWAT	sfc_700_pw	mm	✅	✅	1-hour	HRRR Zarr*	pwat, surface to 700mb
Total Column PWAT	entire_atmosphere_ single_layer_PWAT	kg/m^2	✅	✅	1-hour	HRRR Zarr
Moisture Flux Conv.	moisture_flux_ convergence	s⁻¹	❌	✅	1-hour	HRRR Zarr*
Orographic Vert. Vel.	w_oro	m/s	❌	✅	1-hour	HRRR Zarr*
Bright Band Bottom	BrightBandBottom Height_00.00	m	✅	✅	2-min	NOAA MRMS
Bright Band Top	BrightBand TopHeight_00.00	m	✅	✅	2-min	NOAA MRMS
1hr Max Composite Refl.	CREF_1HR_MAX_00.50	dBZ	✅	✅	2-min	NOAA MRMS
Radar QPE Quality Idx	RadarAccumulation QualityIndex_01H_00.00	non-dim	✅	✅	2-min	NOAA MRMS
Refl. at -5C	Reflectivity_-5C_00.50	dBZ	✅	✅	2-min	NOAA MRMS
Refl. at -10C	Reflectivity_-10C_00.50	dBZ	✅	✅	2-min	NOAA MRMS
Refl. at -15C	Reflectivity_-15C_00.50	dBZ	✅	✅	2-min	NOAA MRMS
Refl. at -20C	Reflectivity_-20C_00.50	dBZ	✅	✅	2-min	NOAA MRMS
Refl. at 0C	Reflectivity_0C_00.50	dBZ	✅	✅	2-min	NOAA MRMS
Lowest Alt. Refl.	ReflectivityAt LowestAltitude_00.50	dBZ	✅	✅	2-min	NOAA MRMS
Seamless Hybrid Scan Refl.	SeamlessHSR_00.00	dBZ	✅	✅	2-min	NOAA MRMS
Seamless HSR Height	SeamlessHSRHeight_00.00	km	✅	✅	2-min	NOAA MRMS
Vert. Integrated Liquid	VIL_00.50	kg/m^2	✅	✅	2-min	NOAA MRMS
*Fields derived from Zarr HRRR

Alignment

---

Temporal

Spatial

Miscillaneous Notes

---

All MRMS/HRRR fields are stored in Zarray (zarr) format and are condense via Blosc compression.

CODEC = BloscCodec(cname="zstd", clevel=3, shuffle="bitshuffle")