3 Results
3.1 CYGNSS and GISS-E2.1 agree: ARs decrease LHF beneath their footprint
The spatial patterns and amplitudes of the climatological latent heat fluxes (LHF) for GISS-E2.1+MERRA2u/v and CYGNSS are the same (Fig. 1). Simulated biases are never larger than 10% of the climatological values (not shown), which is of the order of CYGNSS LHF uncertainty. The model has negative biases just off the equator in the Western Pacific, Caribbean Sea, and Indian Oceans and positive biases in the coastal Eastern Equatorial South Pacific, tropical Atlantic, and South Atlantic. The differences between CYGNSS and GISS LHF may be related to biases in MERRA-2, the GISS planetary boundary layer scheme, or GISS simulated near surface atmospheric specific humidity.
Additionally, even in regions with biases, the temporal variability of simulated LHF matches CYGNSS. We calculate the temporal correlation between the GISS and CYGNSS LHFs centered on the Eastern Equatorial South Pacific region. The Pearson correlation coefficient (e.g., Wilks, 2011, p.50) when using a single grid-point is 0.7. Using time-series created from a 10˚x10˚ area-average for the region, the correlation is 0.83. For regions of low bias (e.g., the center of the North Pacific, Fig. 1c), the correlation for area-averaged data is 0.92. Therefore, spatial and temporal LHF distributions and variability appear realistic in GISS-E2.1.