Evaluation Of Air-sea Turbulent Heat Flux Based On CYGNSS Data Product

The sea-air heat flux is an important way to study the sea-air interaction,which controls the energy exchange between the ocean and the atmosphere and plays a pivotal role in the ocean dynamics processes and climate system changes at different spatial and temporal scales.The heat flux products developed based on CYGNSS（Cyclone Global Navigation Satellite System）wind speed have high spatial and temporal resolution and provide a new platform for the study of sea-air heat fluxes under extreme sea conditions.In this paper,the measured wind speed and turbulent heat flux from KEO（Kuroshio Extension Observatory）buoy and TAO（Tropical Atmosphere Ocean）buoy array are used as validation datasets to evaluate the accuracy and applicability of the CYGNSS heat flux product in the northwestern and equatorial Pacific Ocean.The results show that the CYGNSS heat flux is in good agreement with the measured heat flux of KEO buoy with the correlation coefficient as high as 0.90;there are relatively large errors in the heat fluxes between CYGNSS and TAO buoy array;statistical metrics of 25 km spatial matching heat fluxes are better than those of 50 km matching generally.The validity of the CYGNSS heat flux product in the mid-latitude sea area is verified;in the absence of in-situ observation data in the low-latitude sea area,the CYGNSS heat flux product has some reference significance.Then,based on the COARE（Coupled Ocean-Atmosphere Response Experiment）3.5 turbulent heat flux algorithm and the bulk aerodynamic formulas,we designed sensitivity tests combining the CYGNSS satellite wind speed product,MERRA-2（Modern-Era Retrospective Analysis for Research and Applications Version 2）reanalysis data and GTMBA（Global Tropical Moored Buoy Array）buoy data to obtain a total of six turbulent heat flux results.The heat fluxes obtained from CYGNSS and MERRA-2 wind speeds are in good agreement with the buoy heat flux.Based on this,the performance of each experiment in the three oceans,as well as the performance and error distribution within the ocean basin are analyzed.The errors of heat flux derived from CYGNSS wind speed are mainly distributed in the warm pool near the equator in the central-eastern Pacific Ocean,the Arabian Sea and the Bay of Bengal in the Indian Ocean,and the stations near the Gulf of Guinea in the Atlantic Ocean;the RMSE of latent heat flux reaches the maximum in the Bay of Bengal（90°E,12°N）,about 85W·m^-2.In addition,the performance of these results among different temporal scales is performed.Based on the hourly scale,wavelet coherence analysis was performed to give the correlations of turbulent heat flux of different experiments in the time-frequency domain.CYGNSS heat flux reliably represents intraseasonal and seasonal variability.However,more specific regional investigations are needed to reveal variations at synoptic scales or shorter time periods.EEMD（Ensemble Empirical Mode Decomposition）was performed on the differences in heat fluxes between experiments at the daily scale to analyze the contribution of input variables on each mode.CYGNSS wind speed may contribute to 50%of the total error in heat flux calculations for periods shorter than a week.Finally,based on the monthly time series,it was found that heat fluxes fluctuate most frequently and have the largest amplitude in the Atlantic Ocean.The Indian and Atlantic Oceans exhibit an approximately annual cycle of variability.Sea surface temperature anomalies may explain the drastic changes in heat flux in the Atlantic Ocean and large intra-ocean basin variability.The other input variables of MERRA-2 also fail to respond in time to the strong fluctuations in heat fluxes caused by air temperature.