Fig. 1. Misjudged isolated clouds

Fig. 2. Comparison of different backscatter signals. (a) Simulated cloud-free signal under uniform atmospheric extinction; (b) normalizd relative backscatter signal without range correction; (c) normalizd relative backscatter signal with range correction; (d) logarithmic-corrected backscatter signal

Fig. 3. Flow chart of cloud detection algorithm of backscatter signal based on bidirectional reconstruction

Fig. 4. Forward reconstruction of backscatter signal. (a) Screening results of forward interpolation points; (b) B_A1 obtained by forward interpolation points

Fig. 5. Definition of candidate cloud peak interval. (a) Cloud peaks screened by adaptive threshold; (b) range of candidate cloud peak determined by left and right typical points

Fig. 6. Back reconstruction of backscatter signal. (a) Screening results of back interpolation points; (b) B_A2 obtained by back interpolation points

Fig. 7. Cloud layer information retrieved by B_A and B_LN. (a) Comparison between B_A and B_LN; (b) inversion result of first-layer cloud; (c) inversion result of second-layer cloud; (d) inversion result of third-layer cloud

Fig. 8. Comparison of different cloud top definitions

Fig. 9. Spatiotemporal distribution of preprocessed backscattering signals

Fig. 10. Detection result of cloud value-added product from SGP station

Fig. 11. Comparison of cloud layer detection results obtained by different algorithms. (a) IDZC; (b) VDE ;(c) SMS-SF ; (d) BRBS

Fig. 12. Comparison of cloud base height inversion performance of different algorithms from 00:00 to 24:00 on January 3, 2020. (a) IDZC; (b) VDE ;(c) SMS-SF; (d) BRBS

Fig. 13. Comparison of cloud top height inversion performance of different algorithms from 00:00 to 24:00 on January 3, 2020. (a) IDZC; (b) VDE ;(c) SMS-SF; (d) BRBS

Fig. 14. Error analysis of different algorithms for data inversion results of SGP station in 2020