Uncertainties in OCO-2 satellite retrievals of XCO₂ limit diagnosis of transport model simulation uncertainty

Abstract. Estimating regional CO₂ sources and sinks is challenging due to limited data and uncertainties in transport models. Orbiting Carbon Observatory-2 (OCO-2) overcomes measurement limits, providing CO₂ variations beyond in-situ networks. This study analyses altitude-wise model-observation CO₂ differences from surface to upper troposphere using aircraft observations from ATom, Amazon, and CONTRAIL campaigns over OCO-2 total column CO₂ (XCO₂) sampling location to characterise sources of uncertainty in MIROC4-ACTM. We show model aligns better with ATom tropospheric columns (0.03 ± 0.03 ppm) than OCO-2 XCO₂ (0.2 ± 0.5 ppm), especially over oceans, highlighting the need for expanded profile measurements to characterise errors robustly. Altitude-wise comparisons reveal this differences primarily occur in the lower troposphere (0–2 km), likely due to ACTM's near-surface land CO₂ flux errors. In contrast, ACTM better matches aircraft CO₂ in the middle (2–5 km) and upper (5–8 km) troposphere, likely due to accurate large-scale transport representation. Over the Amazon, CO₂ differences with aircraft and OCO-2 differ, likely due to a lack of regional surface sites for inversion and insufficient high-altitude profile (~4 km) not representative of XCO2. Over Asian megacity airports, which are significant emission hotspots, the model shows a large negative difference with CONTRAIL than OCO-2. This discrepancy likely hints that MIROC4-ACTM is unable to capture urban fossil CO₂ emission signals at airports due to coarse resolution (~2.8° x 2.8°) and higher resolution of OCO-2 limits ability to fully capture actual emission footprints.