Module 7: Climate Monitoring from Space

Satellites provide the only truly global, consistent observing system for climate variables. This module covers sea ice concentration retrieval from passive microwave radiometry, ice sheet mass balance from GRACE gravity measurements, atmospheric composition monitoring with TROPOMI, and sea surface temperature anomaly detection.

1. Sea Ice Concentration from Passive Microwave

Passive microwave radiometers (SSMIS, AMSR2) measure the brightness temperature$T_B$ of the surface at frequencies from 6 to 89 GHz. Sea ice and open water have dramatically different microwave emissivities: ice emissivity is ~0.9 at 37 GHz (V-pol), while open water is ~0.4. This contrast enables sea ice concentration (SIC) retrieval regardless of cloud cover or polar darkness.

The Bootstrap algorithm, one of the two standard NASA algorithms, estimates SIC using a linear mixing model at 37 GHz vertical polarization:

$$SIC = \frac{T_B(37V) - T_{OW}}{T_{ICE} - T_{OW}} \times 100\%$$

where $T_{OW}$ is the brightness temperature tie-point for 100% open water (~180 K at 37 GHz V-pol) and $T_{ICE}$ is the tie-point for 100% sea ice (~250 K for first-year ice). The NASA Team algorithm uses polarization ratios and gradient ratios to distinguish ice types, while newer algorithms like ARTIST Sea Ice (ASI) use the 89 GHz channel for higher resolution (~5 km vs ~25 km).

Arctic Sea Ice Record

Continuous satellite monitoring since 1979 reveals a decline of ~13% per decade in September minimum extent. The record low was set in 2012 (3.39 million km²). The passive microwave record is the longest continuous satellite climate data record.

Ice Thickness from Altimetry

CryoSat-2 and ICESat-2 measure freeboard (ice above water). Combined with snow depth estimates, hydrostatic equilibrium gives thickness:$h_{ice} = \frac{\rho_w}{\rho_w - \rho_i}f_b + \frac{\rho_s}{\rho_w - \rho_i}h_s$where $f_b$ is freeboard.

Weather Filtering

Atmospheric water vapor and liquid water path can increase $T_B$ over open ocean, creating false ice signals. Weather filters using the 19 GHz and 22 GHz channels correct for this effect. The gradient ratio$GR = (T_B(37V) - T_B(19V))/(T_B(37V) + T_B(19V))$ identifies contaminated pixels when it exceeds empirical thresholds.

2. GRACE Gravity & Ice Sheet Mass Balance

The GRACE and GRACE-FO missions measure changes in Earth's gravity field by tracking the distance between twin satellites (~220 km apart) to micrometer precision. Mass redistribution on Earth's surface (ice loss, groundwater depletion, ocean mass changes) alters the gravity field, which changes the satellite separation.

The gravity field is expressed as spherical harmonic coefficients $C_{lm}$and $S_{lm}$. Changes in these coefficients are converted to equivalent water height (EWH) — the thickness of a water layer that would produce the same gravity change:

$$\Delta h_{EWH} = \frac{R^2}{M_E} \sum_{l,m} \frac{2l+1}{1+k_l} \Delta C_{lm} \bar{P}_{lm}$$

where $R$ is Earth's radius, $M_E$ is Earth's mass,$k_l$ are Love numbers (accounting for Earth's elastic response),$\Delta C_{lm}$ are the monthly spherical harmonic coefficient anomalies, and $\bar{P}_{lm}$ are normalized associated Legendre polynomials.

Key GRACE Results

●Greenland: Losing ~270 Gt/year (2002–2023), equivalent to ~0.75 mm/year sea level rise.
●Antarctica: Losing ~150 Gt/year, predominantly from West Antarctica and the Antarctic Peninsula.
●Groundwater: GRACE revealed severe aquifer depletion in India, California's Central Valley, and the Middle East.

Glacial Isostatic Adjustment (GIA)

The Earth's mantle is still rebounding from the last ice age. This GIA signal must be subtracted from GRACE data to isolate present-day ice mass changes. GIA models (e.g., ICE-6G) contribute the largest source of uncertainty in Antarctic mass balance estimates, particularly for East Antarctica.

3. Atmospheric Composition: TROPOMI & OCO

The Sentinel-5P TROPOMI instrument provides daily global maps of NO₂, SO₂, CO, O₃, CH₄, and HCHO at 3.5 km × 5.5 km resolution. The Differential Optical Absorption Spectroscopy (DOAS) technique retrieves trace gas column densities by fitting the observed spectrum to reference cross-sections in the UV-visible range.

For CO₂, NASA's Orbiting Carbon Observatory (OCO-2/3) and ESA's upcoming CO2M mission retrieve column-averaged CO₂ (XCO₂) from shortwave infrared absorption at 1.6 and 2.0 μm. The retrieval precision of ~1 ppm enables detection of regional CO₂ enhancements from cities and power plants.

TROPOMI NO₂ Applications

Air quality monitoring: Tropospheric NO₂ columns map combustion sources (traffic, power plants, shipping lanes). COVID-19 lockdowns produced dramatic 20–50% reductions visible from space.

Emission inventories: Top-down emission estimates invert satellite NO₂ maps with atmospheric transport models to constrain national emission reports.

Lightning detection: NOx produced by lightning creates stratospheric NO₂ enhancements detectable by TROPOMI, providing an independent lightning climatology.

4. Climate Data Visualizations

The following simulation generates three key climate monitoring visualizations: Arctic sea ice extent decline (from synthetic data matching observed trends), a global NO₂ distribution map (synthetic TROPOMI-like data), and sea surface temperature anomaly detection.

Arctic Sea Ice, TROPOMI NO2, and SST Anomaly Visualization

Python

script.py165 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

np.random.seed(42)

fig, axes = plt.subplots(2, 2, figsize=(14, 11))

# ─── Panel 1: Arctic Sea Ice Extent Trend ───
ax = axes[0, 0]
years = np.arange(1979, 2026)
n = len(years)

# September minimum extent (million km2) - synthetic matching observed trend
# Linear decline + acceleration + interannual variability
trend = 7.5 - 0.08 * (years - 1979) - 0.001 * (years - 1979)**2
noise = np.random.normal(0, 0.3, n)
# Add known events
extent = trend + noise
extent[2012 - 1979] = 3.39  # record low
extent[2020 - 1979] = 3.74
extent = np.clip(extent, 2.5, 8.5)

# Linear fit
z = np.polyfit(years, extent, 1)
trend_line = np.polyval(z, years)

ax.fill_between(years, extent, alpha=0.3, color='#38bdf8')
ax.plot(years, extent, 'o-', color='#38bdf8', markersize=3, linewidth=1.5, label='September minimum')
ax.plot(years, trend_line, '--', color='#ef4444', linewidth=2, label=f'Trend: {z[0]*10:.2f} M km2/decade')
ax.scatter([2012], [3.39], color='#ef4444', s=100, zorder=5, edgecolors='white', linewidths=1.5)
ax.annotate('2012 record\n3.39 M km2', xy=(2012, 3.39), xytext=(1998, 3.0),
           fontsize=9, color='#ef4444', arrowprops=dict(arrowstyle='->', color='#ef4444'))
ax.set_xlabel('Year', fontsize=11)
ax.set_ylabel('Extent (million km2)', fontsize=11)
ax.set_title('Arctic Sea Ice September Minimum', fontsize=12, fontweight='bold', color='white')
ax.legend(fontsize=9)
ax.grid(True, alpha=0.3)
ax.set_facecolor('#0a0a0a')

print(f"Sea ice trend: {z[0]*10:.3f} million km2 per decade")
print(f"Total decline 1979-2025: {extent[0] - extent[-1]:.2f} million km2")

# ─── Panel 2: Synthetic TROPOMI NO2 Map ───
ax2 = axes[0, 1]
lats = np.linspace(-60, 70, 130)
lons = np.linspace(-180, 180, 360)
LON, LAT = np.meshgrid(lons, lats)

# Background tropospheric NO2 (low over oceans, moderate over land)
no2 = np.ones_like(LON) * 0.5  # baseline 0.5e15 molec/cm2

# Major hotspots
hotspots = [
    (40, -74, 'New York', 12),
    (51, 7, 'Europe', 15),
    (40, 116, 'Beijing', 18),
    (35, 137, 'Japan', 10),
    (28.5, 77, 'Delhi', 16),
    (23, 113, 'Guangdong', 14),
    (-23, -46, 'Sao Paulo', 8),
    (6, 3, 'Lagos', 7),
    (37, 127, 'Seoul', 9),
]
for lat0, lon0, name, intensity in hotspots:
    dist = np.sqrt((LAT - lat0)**2 + (LON - lon0)**2)
    no2 += intensity * np.exp(-dist**2 / 80)

# Shipping lanes
shipping = np.exp(-((LAT - 30)**2) / 50) * 2 * (np.abs(LON) < 160).astype(float)
no2 += shipping * 0.3

# Add noise
no2 += np.random.normal(0, 0.3, no2.shape)
no2 = np.clip(no2, 0, 20)

im = ax2.pcolormesh(LON, LAT, no2, cmap='hot_r', vmin=0, vmax=16, shading='auto')
plt.colorbar(im, ax=ax2, label='NO2 (10^15 molec/cm2)', shrink=0.8)
ax2.set_title('Tropospheric NO2 (TROPOMI-like)', fontsize=12, fontweight='bold', color='white')
ax2.set_xlabel('Longitude', fontsize=11)
ax2.set_ylabel('Latitude', fontsize=11)
ax2.set_facecolor('#0a0a0a')

# ─── Panel 3: SST Anomaly Map ───
ax3 = axes[1, 0]
lats_sst = np.linspace(-60, 60, 120)
lons_sst = np.linspace(-180, 180, 360)
LON_S, LAT_S = np.meshgrid(lons_sst, lats_sst)

# Generate SST anomaly field (synthetic El Nino pattern)
sst_anomaly = np.zeros_like(LON_S)
# El Nino warm tongue in equatorial Pacific
enso = 2.5 * np.exp(-LAT_S**2 / 100) * np.exp(-((LON_S + 120)**2) / 2000)
sst_anomaly += enso
# Warm blob in NE Pacific
blob = 1.5 * np.exp(-((LAT_S - 45)**2 + (LON_S + 140)**2) / 300)
sst_anomaly += blob
# Cool anomaly in North Atlantic
cool = -1.2 * np.exp(-((LAT_S - 50)**2 + (LON_S + 30)**2) / 400)
sst_anomaly += cool
# Random variability
sst_anomaly += np.random.normal(0, 0.3, sst_anomaly.shape)
def gaussian_smooth(data, sigma):
    size = int(4 * sigma + 1)
    x = np.arange(-size, size + 1)
    kernel_1d = np.exp(-x**2 / (2 * sigma**2))
    kernel_1d /= kernel_1d.sum()
    smoothed = np.apply_along_axis(lambda row: np.convolve(row, kernel_1d, mode='same'), 1, data)
    smoothed = np.apply_along_axis(lambda col: np.convolve(col, kernel_1d, mode='same'), 0, smoothed)
    return smoothed
sst_anomaly = gaussian_smooth(sst_anomaly, 3)

im3 = ax3.pcolormesh(LON_S, LAT_S, sst_anomaly, cmap='RdBu_r', vmin=-3, vmax=3, shading='auto')
plt.colorbar(im3, ax=ax3, label='SST Anomaly (K)', shrink=0.8)
ax3.set_title('Sea Surface Temperature Anomaly', fontsize=12, fontweight='bold', color='white')
ax3.set_xlabel('Longitude', fontsize=11)
ax3.set_ylabel('Latitude', fontsize=11)
ax3.set_facecolor('#0a0a0a')

# ─── Panel 4: SST Anomaly Time Series (Nino 3.4) ───
ax4 = axes[1, 1]
months = np.arange(0, 360)  # 30 years monthly
dates = 1994 + months / 12.0

# Synthetic ENSO index (Nino 3.4 SST anomaly)
# Quasi-periodic oscillation with 3-7 year period
enso_ts = np.zeros(len(months))
for period in [3.5, 5.2, 7.1]:  # multiple modes
    phase = np.random.uniform(0, 2*np.pi)
    enso_ts += np.random.uniform(0.3, 0.8) * np.sin(2*np.pi*months/(period*12) + phase)
enso_ts += np.random.normal(0, 0.15, len(months))
# Smooth
kernel = np.ones(5) / 5
enso_ts = np.convolve(enso_ts, kernel, mode='same')

colors_ts = ['#ef4444' if v > 0.5 else '#3b82f6' if v < -0.5 else '#6b7280' for v in enso_ts]
ax4.bar(dates, enso_ts, width=1/12, color=colors_ts, alpha=0.8)
ax4.axhline(y=0.5, color='#ef4444', linestyle='--', linewidth=1, alpha=0.5, label='El Nino threshold')
ax4.axhline(y=-0.5, color='#3b82f6', linestyle='--', linewidth=1, alpha=0.5, label='La Nina threshold')
ax4.axhline(y=0, color='white', linewidth=0.5, alpha=0.3)
ax4.set_xlabel('Year', fontsize=11)
ax4.set_ylabel('Nino 3.4 SST Anomaly (K)', fontsize=11)
ax4.set_title('ENSO Index (Synthetic)', fontsize=12, fontweight='bold', color='white')
ax4.legend(fontsize=9, loc='upper right')
ax4.grid(True, alpha=0.3)
ax4.set_facecolor('#0a0a0a')
ax4.set_xlim(1994, 2024)

# Count El Nino / La Nina events
el_nino_months = np.sum(enso_ts > 0.5)
la_nina_months = np.sum(enso_ts < -0.5)
print(f"\nEl Nino months (>0.5K): {el_nino_months} ({100*el_nino_months/len(months):.1f}%)")
print(f"La Nina months (<-0.5K): {la_nina_months} ({100*la_nina_months/len(months):.1f}%)")

for ax in axes.flat:
    ax.tick_params(colors='white')
    for spine in ax.spines.values():
        spine.set_color('#334155')

plt.tight_layout()
plt.savefig('output.png', dpi=130, bbox_inches='tight', facecolor='#0a0a0a')
plt.close()
print("\n4-panel climate monitoring visualization saved.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

5. Sea Surface Temperature & Marine Heatwaves

SST is measured by both infrared sensors (MODIS, VIIRS, SLSTR) with ~1 km resolution and microwave radiometers (AMSR2) with ~25 km resolution but all-weather capability. Multi-sensor SST analyses (e.g., OSTIA, MUR) blend these sources into daily gap-free products at 1–5 km resolution.

Marine Heatwave Detection

A marine heatwave (MHW) is defined as a period when SST exceeds the 90th percentile of the climatological distribution for at least 5 consecutive days. The intensity categories follow Hobday et al. (2018):

●Moderate: SST between the 90th percentile and 2× the difference above the climatological mean.
●Strong: 2× to 3× above; associated with coral bleaching.
●Severe: 3× to 4× above; mass mortality events in marine ecosystems.
●Extreme: >4× above; unprecedented conditions, catastrophic ecosystem impacts.

ENSO & Global SST Teleconnections

The El Niño–Southern Oscillation (ENSO) is the dominant mode of interannual SST variability. The Niño 3.4 index (SST anomaly in 5°S–5°N, 170°W–120°W) defines El Niño (>+0.5 K for 5 consecutive months) and La Niña (<−0.5 K). Satellite SST records since 1981 have captured 12 El Niño events, including the extreme 1997–98 and 2015–16 episodes.

6. Essential Climate Variables (ECVs)

The Global Climate Observing System (GCOS) defines 54 Essential Climate Variables that are critical for monitoring the climate system. Satellites contribute to the majority of these ECVs. Key satellite-observed ECVs include:

Atmosphere

Temperature (upper air)
Water vapor
Ozone
Aerosol properties
CO₂, CH₄
Cloud properties
Earth radiation budget

Ocean

Sea surface temperature
Sea level
Sea ice
Ocean colour
Surface currents
Sea state
Salinity

Land

Land cover
Fire disturbance
Soil moisture
Snow cover / SWE
Glaciers & ice caps
Albedo
Land surface temperature

Climate Data Records

ESA's Climate Change Initiative (CCI) and NOAA's Climate Data Records program produce long-term, intercalibrated satellite datasets spanning multiple missions. These Fundamental Climate Data Records (FCDRs) and Thematic Climate Data Records (TCDRs) undergo rigorous homogenization to remove inter-sensor biases and ensure continuity across instrument transitions.

7. Key Takeaways

✓Passive microwave radiometry provides the 45+ year Arctic sea ice record, revealing a decline of ~13% per decade in September extent.

✓GRACE/GRACE-FO gravity measurements quantify ice sheet mass loss at Greenland (~270 Gt/yr) and Antarctica (~150 Gt/yr) with monthly temporal resolution.

✓TROPOMI provides daily global maps of air pollutants (NO₂, SO₂, CO, O₃) at unprecedented 3.5 km resolution, enabling emission monitoring from space.

✓Multi-sensor SST analyses detect marine heatwaves and ENSO events critical for fisheries, coral reef management, and seasonal climate prediction.

✓GCOS defines 54 Essential Climate Variables; satellites are the primary observing system for the majority of these ECVs.