The Tibetan Observation and Research Platform (TORP) is a project aimed at research on the land surface process over the TP, which comprises various comprehensive observation and research stations and other additional observational sites (Ma et al., 2008). The locations of the comprehensive observation and research stations used in this study are marked in Fig. 1, and their detailed information is listed in Table 1. The TORP in-situ measurements, with a temporal resolution of 30 minutes, were used to develop the retrieval algorithms and carry out the cross validation.
Figure 1. Locations of the Tibetan Plateau (black thick line) and in-situ meteorological stations (red stars) from TORP.
Station Longitude (°E) Latitude (°N) Altitude (m) Heat fluxes measurements for ground validation NADORS 79.7035 33.39056 4270 SWD, SWU, LWD, LWU, Rn NAMORS 90.9636 30.7699 4730 H, LE QOMS 86.949638 28.358209 4276 SWD, SWU, LWD, LWU, Rn, H SETORS 94.7417 29.7622 3326 SWD, SWU, LWD Notes: SWD, downwelling shortwave radiation flux; SWU, upwelling shortwave radiation flux; LWD, downwelling longwave radiation flux; LWU, upwelling longwave radiation flux; Rn, net radiation flux; H, sensible heat flux; LE, latent heat flux.
Table 1. Information on the TORP stations for ground validation.
The Fengyun-4 geostationary satellite system, the successor of Fengyun-2, is the second generation of the Chinese geostationary meteorological satellite system. Launched on 11 December 2016, Fengyun-4A (FY-4A) is the first test satellite of the Fengyun-4 system. The Level 1 data of FY-4A started on 12 March 2018. FY-4A provides two modes of data telemetering: the China region mode, and the full disk mode, which covers the Asia and Oceania region. The temporal resolution of FY-4A Level 1 data can reach 15 minutes (Min et al., 2017). The meteorological instruments onboard FY-4A include the Advanced Geostationary Radiation Imager (AGRI), the Geostationary Interferometric Infrared Sounder (GIIRS), the Lightning Mapping Imager (LMI), and the Space Environment Monitoring Instrument Package (SEP) (Yang et al., 2017). The band list of FY-4A/AGRI is displayed in Table 2. Compared with FY-2/VISSR, the imaging observation bands of multichannel scan imagery radiometer FY-4A/AGRI have been expanded from 5 channels to 14 channels. Of the total 14 spectral bands of FY-4A/AGRI, channels 1–3 cover the VIS to NIR bands, channels 4–6 cover the SWIR bands, channels 7–8 cover the MWIR bands, channels 9–10 are the water vapor bands, and channels 11–14 cover the LWIR bands. The spatial resolutions of the VIS to NIR, SWIR, MWIR and LWIR bands are 0.5–1 km, 2 km, 2–4 km and 4 km, respectively. In this study, channels 2–3 were used to calculate the NDVI, channels 1–6 were used for broadband albedo retrieval, and channels 12–13 were applied for estimation of LST (Fig. 2). For the spatial resolution differences of FY-4A/AGRI bands, in order to reduce the errors due to resolution mismatch, the finer spatial resolution bands should be upscaled to the same resolution as those bands with coarser resolution, i.e., from 0.5–2 km to 4 km resolution. Fortunately, for each band of FY-4A/AGRI, data with coarser resolution of 4 km were already provided on the official website. Therefore, the 4 km resolution data for all bands were uniformly used for subsequent surface characteristic parameter retrieval and turbulent heat flux estimation in this study.
1 0.45–0.49 0.47 1 2 0.55–0.75 0.65 0.5 3 0.75–0.90 0.83 1 4 1.36–1.39 1.37 2 5 1.58–1.64 1.61 2 6 2.10–2.35 2.22 2 7 3.50–4.00 3.72 2 8 3.50–4.00 3.72 4 9 5.80–6.70 6.25 4 10 6.90–7.30 7.10 4 11 8.00–9.00 8.50 4 12 10.3–11.3 10.8 4 13 11.5–12.5 12.0 4 14 13.2–13.8 13.5 4
Table 2. Band list of FY-4A/AGRI.
Figure 2. Flowchart of the retrieval method of the land surface characteristic parameters and heat fluxes by combining FY-4A/AGRI and CLDAS-V2.0 data (α1–α6, narrowband albedos of FY-4A/AGRI spectral bands 01–06; T12 and T13, brightness temperatures monitored in band 12 and 13 of FY-4A/AGRI, respectively; SWD, downwelling shortwave radiation flux; p, near-surface air pressure; q, specific humidity at 2 m, Tair, air temperature at 2 m; u, wind speed at 10 m; NDVI, Normalized Difference Vegetation Index; α, surface broadband albedo; LST, land surface temperature; WVC, atmospheric water vapor content; LWD, downwelling longwave radiation flux; Pv, proportion of vegetation; SWU, upwelling shortwave radiation flux; Rn, net radiation flux; H, sensible heat flux; ε12 and ε13, narrowband emissivities of band 12 and 13 of FY-4A/AGRI, respectively; ε, land surface broadband emissivity; LWU, upwelling longwave radiation flux; G, soil heat flux; LE, latent heat flux).
For the clear detection on FY-4A/AGRI, the threshold method (Oku et al., 2007) was used in this study. For bands 12–13, brightness temperature values below 250 K were not used in the LST retrieval. The threshold of 250 K was determined based on TORP in-situ measurements to acclimate the general climatic condition over the TP in spring. Due to the presence of cloud, the cloud-top temperature instead of the surface temperature will be detected, which yields an underestimation in brightness temperature and thus a lower LST. The underestimation in LST will further induce underestimation in LWU (upwelling longwave radiation flux) and H (sensible heat flux). Additionally, areas covered by clouds will also exhibit higher surface broadband albedo than those under clear-sky conditions. The overestimation in surface broadband albedo will result in overestimation in SWU. Therefore, the abnormal low brightness temperature data points were considered as cloudy and then eliminated based on this threshold. However, several defects also exist in the threshold method. First, the selection of the threshold is difficult to determine, for this one-size-fits-all approach may induce some uncertain missing reports and false alarms. Second, due to the land cover heterogeneity over the TP, the threshold method is not sufficient to meet the unique hydrometeorological conditions over the whole TP. It should be noted that the FY-4A Level-2 CLM (cloud mask) product is also available for clear detection. However, the areal proportion of the clear region over the whole TP detected via the FY-4A CLM product is too small (< 20%) for most observation times, from which it is difficult to derive the spatial distributions of the surface characteristic parameters. Thus, the threshold method was used for the clear detection in this study.
For estimation of the radiation flux components and turbulent heat fluxes, meteorological forcing data are needed as inputs. In this study, the CLDAS (China Land Data Assimilation System) meteorological forcing dataset, CLDAS-V2.0, was used in combination with FY-4A satellite data for flux estimation. CLDAS-V2.0 has a spatial resolution of 0.0625° × 0.0625° and a temporal resolution of 1 h, and covers the whole East Asia region (0°–65°N, 60°–160°E) from 19 January 2017. The near-surface air pressure p, specific humidity q measured at 2 m, air temperature Tair measured at 2 m, horizontal wind speed u measured at 10 m, and surface downwelling shortwave radiation flux (SWD) were used as the input data. For integration of the CLDAS-V2.0 with FY-4A satellite data, the FY-4A data were upscaled from the original 4 km × 4 km normalized projection grid to the 0.0625° × 0.0625° equal latitude and longitude grid of CLDAS-V2.0 using bilinear spatial interpolation.
|Station||Longitude (°E)||Latitude (°N)||Altitude (m)||Heat fluxes measurements for ground validation|
|NADORS||79.7035||33.39056||4270||SWD, SWU, LWD, LWU, Rn|
|QOMS||86.949638||28.358209||4276||SWD, SWU, LWD, LWU, Rn, H|
|SETORS||94.7417||29.7622||3326||SWD, SWU, LWD|
|Notes: SWD, downwelling shortwave radiation flux; SWU, upwelling shortwave radiation flux; LWD, downwelling longwave radiation flux; LWU, upwelling longwave radiation flux; Rn, net radiation flux; H, sensible heat flux; LE, latent heat flux.|