Climate Data Analysis with ERA5 & CHIRPS

New in v1.0.0 🌡️

This tutorial showcases the new climate reanalysis capabilities introduced in v1.0.0, including ERA5-Land (47 variables, 1950-present) and CHIRPS precipitation (1981-present).

Overview

Ndvi2Gif v1.0.0 extends beyond vegetation monitoring to comprehensive climate analysis. You can now analyze:

• 🌡️ Temperature (ERA5): Air, surface, soil temperatures with daily min/max

• 🌧️ Precipitation (ERA5 & CHIRPS): Multiple high-quality rainfall datasets

• 💧 Soil Moisture (ERA5): 4 depth layers

• ☀️ Radiation (ERA5): Solar radiation and heat flux

• 💨 Wind & Pressure (ERA5): Wind components and atmospheric pressure

• ❄️ Snow (ERA5): Snow depth and snowfall

Dataset Comparison

Dataset	Variables	Resolution	Coverage	Best For
ERA5-Land	47 climate vars	~11 km	1950-present, global	Multi-variable climate analysis
CHIRPS	Precipitation	~5.5 km	1981-present, 50°S-50°N	High-res rainfall, drought monitoring

ERA5-Land: Temperature Analysis

Basic Temperature Analysis

import ee
from ndvi2gif import NdviSeasonality
from ndvi2gif.timeseries import TimeSeriesAnalyzer

ee.Initialize()

# Define ROI (e.g., Doñana National Park, Spain)
roi = ee.Geometry.Point([-6.48, 37.13]).buffer(10000)  # 10km buffer

# Monthly mean temperature (in Celsius)
temp = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='temperature_2m_celsius',  # Automatic Kelvin→Celsius conversion
    periods=12,                      # Monthly composites
    start_year=2020,
    end_year=2023,
    key='mean'                       # Average daily temperature
)

# Generate composite
composite = temp.get_year_composite()
print(f"Generated {len(temp.imagelist)} annual composites")

Temperature Extremes (Min/Max)

# Daily maximum temperatures (summer heat analysis)
temp_max = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='temperature_2m_max_celsius',  # Daily maximum
    periods=12,
    start_year=2020,
    end_year=2023,
    key='max'  # Maximum of daily maxima (hottest day of month)
)

# Daily minimum temperatures (frost analysis)
temp_min = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='temperature_2m_min_celsius',  # Daily minimum
    periods=12,
    start_year=2020,
    end_year=2023,
    key='min'  # Minimum of daily minima (coldest night of month)
)

Time Series Analysis

# Extract temperature time series
analyzer = TimeSeriesAnalyzer(temp)
df = analyzer.extract_time_series(point=(-6.48, 37.13), scale=11000)

print(df.head())
#         date      value  year    period  doy  season  month
# 0 2020-01-16  11.2       2020   january   16  winter      1
# 1 2020-02-14  12.5       2020  february   45  winter      2
# 2 2020-03-16  15.8       2020     march   75  spring      3

# Analyze temperature trends
trends = analyzer.analyze_trend(df=df, method='mann_kendall')
print(f"Temperature trend: {trends['interpretation']}")

# Comprehensive dashboard (shows climate stats, not phenology)
fig = analyzer.plot_comprehensive_analysis()
fig.savefig('temperature_analysis.png', dpi=300)

Precipitation Analysis

ERA5-Land Precipitation

# Monthly precipitation totals (in L/m² = mm)
precip_era5 = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='total_precipitation_sum_lm2',  # Automatic m→L/m² conversion
    periods=12,
    start_year=2020,
    end_year=2023,
    key='sum'  # Sum for monthly totals
)

# Extract and analyze
analyzer = TimeSeriesAnalyzer(precip_era5)
df_precip = analyzer.extract_time_series(scale=11000)

print(f"Annual precipitation 2020-2023:")
for year in range(2020, 2024):
    yearly_precip = df_precip[df_precip['year'] == year]['value'].sum()
    print(f"  {year}: {yearly_precip:.1f} mm")

CHIRPS Precipitation (Higher Resolution)

# CHIRPS: Higher spatial resolution, station-calibrated
precip_chirps = NdviSeasonality(
    roi=roi,
    sat='CHIRPS',
    index='precipitation',  # mm/day
    periods=12,
    start_year=2020,
    end_year=2023,
    key='sum'  # Monthly totals
)

# Compare ERA5 vs CHIRPS
analyzer_chirps = TimeSeriesAnalyzer(precip_chirps)
df_chirps = analyzer_chirps.extract_time_series(scale=5566)

# Correlation analysis
import numpy as np
from scipy.stats import pearsonr

# Align time series
common_dates = set(df_precip['date']).intersection(df_chirps['date'])
era5_vals = df_precip[df_precip['date'].isin(common_dates)]['value'].values
chirps_vals = df_chirps[df_chirps['date'].isin(common_dates)]['value'].values

r, p = pearsonr(era5_vals, chirps_vals)
print(f"ERA5 vs CHIRPS correlation: r={r:.3f}, p={p:.4f}")

Drought Monitoring Example

# Combine precipitation and evapotranspiration for drought index
# Step 1: Get precipitation
precip = NdviSeasonality(
    roi=roi,
    sat='CHIRPS',
    index='precipitation',
    periods=12,
    start_year=2015,
    end_year=2023,
    key='sum'
)

# Step 2: Get potential evapotranspiration
pet = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='potential_evaporation_sum_lm2',
    periods=12,
    start_year=2015,
    end_year=2023,
    key='sum'
)

# Step 3: Calculate water balance
analyzer_precip = TimeSeriesAnalyzer(precip)
analyzer_pet = TimeSeriesAnalyzer(pet)

df_precip = analyzer_precip.extract_time_series(scale=5566)
df_pet = analyzer_pet.extract_time_series(scale=11000)

# Simple drought indicator: P - PET
df_precip['water_balance'] = df_precip['value'] - df_pet['value'].abs()
df_precip['drought'] = df_precip['water_balance'] < -50  # Deficit > 50mm

print(f"Drought months (2015-2023): {df_precip['drought'].sum()}")

Soil Moisture Analysis

# Root zone soil moisture (0-100cm)
soil_moisture = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='volumetric_soil_water_layer_3',  # 28-100cm depth
    periods=12,
    start_year=2020,
    end_year=2023,
    key='mean'
)

analyzer = TimeSeriesAnalyzer(soil_moisture)
df_soil = analyzer.extract_time_series(scale=11000)

# Identify dry/wet periods
dry_threshold = df_soil['value'].quantile(0.25)
wet_threshold = df_soil['value'].quantile(0.75)

df_soil['condition'] = 'normal'
df_soil.loc[df_soil['value'] < dry_threshold, 'condition'] = 'dry'
df_soil.loc[df_soil['value'] > wet_threshold, 'condition'] = 'wet'

print(df_soil['condition'].value_counts())

Combined Climate-Vegetation Analysis

# Correlate vegetation (NDVI) with climate variables

# 1. Get NDVI time series
ndvi = NdviSeasonality(
    roi=roi,
    sat='S2',
    index='ndvi',
    periods=12,
    start_year=2020,
    end_year=2023,
    key='median'
)

# 2. Get temperature
temp = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='temperature_2m_celsius',
    periods=12,
    start_year=2020,
    end_year=2023,
    key='mean'
)

# 3. Get precipitation
precip = NdviSeasonality(
    roi=roi,
    sat='CHIRPS',
    index='precipitation',
    periods=12,
    start_year=2020,
    end_year=2023,
    key='sum'
)

# 4. Extract and merge
from ndvi2gif.timeseries import TimeSeriesAnalyzer
import pandas as pd

df_ndvi = TimeSeriesAnalyzer(ndvi).extract_time_series(scale=20)
df_temp = TimeSeriesAnalyzer(temp).extract_time_series(scale=11000)
df_prec = TimeSeriesAnalyzer(precip).extract_time_series(scale=5566)

# Merge on date
df_combined = df_ndvi[['date', 'value']].rename(columns={'value': 'ndvi'})
df_combined = df_combined.merge(
    df_temp[['date', 'value']].rename(columns={'value': 'temperature'}),
    on='date'
)
df_combined = df_combined.merge(
    df_prec[['date', 'value']].rename(columns={'value': 'precipitation'}),
    on='date'
)

# Correlation matrix
print(df_combined[['ndvi', 'temperature', 'precipitation']].corr())

# Lagged correlations (vegetation response to climate)
df_combined['precip_lag1'] = df_combined['precipitation'].shift(1)
df_combined['temp_lag1'] = df_combined['temperature'].shift(1)

print("\nLagged correlations (1-month lag):")
print(df_combined[['ndvi', 'precip_lag1', 'temp_lag1']].corr())

Export Climate Data

# Export temperature rasters to Google Drive
temp = NdviSeasonality(
    roi=roi,
    sat='ERA5',
    index='temperature_2m_celsius',
    periods=12,
    start_year=2023,
    end_year=2023,
    key='mean'
)

# Generate composites
temp.get_year_composite()

# Export all 12 monthly images
for i, image in enumerate(temp.imagelist[0]):
    task = ee.batch.Export.image.toDrive(
        image=image,
        description=f'temp_2023_month_{i+1:02d}',
        folder='ERA5_Temperature',
        region=roi,
        scale=11000,
        crs='EPSG:4326'
    )
    task.start()
    print(f"Started export: month {i+1}")

Available ERA5 Variables

Temperature (24 variables)

• temperature_2m, dewpoint_temperature_2m, skin_temperature, soil_temperature_level_1

• Add _min or _max for daily extremes (8 variables)

• Add _celsius for Celsius conversion (12 variables)

Precipitation & Water Balance (11 variables)

• total_precipitation_sum, total_evaporation_sum, potential_evaporation_sum

• runoff_sum, surface_runoff_sum, snowfall_sum

• Add _lm2 for L/m² units (6 variables)

Soil Moisture (4 variables)

• volumetric_soil_water_layer_1 (0-7 cm)

• volumetric_soil_water_layer_2 (7-28 cm)

• volumetric_soil_water_layer_3 (28-100 cm)

• volumetric_soil_water_layer_4 (100-289 cm)

Radiation (3 variables)

• surface_solar_radiation_downwards_sum

• surface_net_solar_radiation_sum

• surface_latent_heat_flux_sum

Wind & Pressure (3 variables)

• u_component_of_wind_10m, v_component_of_wind_10m

• surface_pressure

Snow (2 variables)

• snow_depth_water_equivalent, snowfall_sum

Key Differences: Climate vs Vegetation Data

Aspect	Vegetation (S2, Landsat)	Climate (ERA5, CHIRPS)
Statistics	max, median, percentile (cloud-free)	mean, sum, min, max (all valid)
Time Series	Phenology metrics (SOS, EOS, POS)	Seasonal climate statistics
Temporal Res	5-16 days	Daily (aggregated to periods)
Spatial Res	10-30m	5.5-11 km
Coverage	2013-present (S2)	1950-present (ERA5)

Tips & Best Practices

Best Practices

1. Use sum for precipitation/flux variables (total accumulation)

2. Use mean for temperature/pressure (average conditions)

3. Use min/max for extremes (frost risk, heat stress)

4. CHIRPS for precipitation (better station calibration)

5. ERA5 for multi-variable analysis (temperature + precip + soil moisture)

6. Match scales: ERA5 ~11km, CHIRPS ~5.5km

7. Climate dashboards automatically replace phenology panels

References

• ERA5-Land: Muñoz Sabater, J. (2019). DOI:10.24381/cds.e2161bac

• CHIRPS: Funk et al. (2015). Scientific Data. DOI:10.1038/sdata.2015.66

---

See also

• time_series - Advanced time series analysis

• Statistical_methods - Statistical reducers explained

• Datasets Reference - Complete dataset documentation