EAA: Eastern Asia; SAA: South Asia; SEA: Southeast Asia; EUR: Europe; ENAM: Eastern North America; WNAM: Western North America; SAM: South America; NAME: North Africa and the Middle East; SAF: South Africa; OCE: Oceania.
On a global scale, the ART-GCT-GEE model demonstrates superior overall accuracy, attaining a high CV-R value of 0.902, with the median bias approaching zero and maintaining average RMSE and MAE values of 0.086 and 0.050, respectively. Overall, approximately 84.38% and 52.11% of the Landsat AOD retrievals conform to the EE and GCOS criteria (Figure 4a). Our model also demonstrates a strong capability to predict AOD at new locations in the absence of ground monitors (Figure 4b) and on dates when ground measurements are unavailable (Figure 4c), as evidenced by average CV-R values of 0.826 and 0.871, MAE values of 0.066 and 0.057, and RMSE values of 0.111 and 0.097, respectively. Notably, around 74.71% (41.68%) and 80.43% (47.34%) of the AOD predictions align with the EE and GCOS anticipated deviations, respectively. Furthermore, our proposed model also exhibits great performance at regional scales (Tables 1, S4, and S5), especially in East and South Asia (e.g., CV-R = 0.83–0.93, \(f_{\text{EE}}\) = 67–84%) and South America (e.g., R = 0.82–0.90, \(f_{\text{EE}}\) = 81–89%).