Comparisons of plots of H and LE vs latitude along the flight track with the slope of a least-squares best-fit straight line in scatter plot relating H to LE along the track reveals that the slope DLE/DH is a good measure of the ratio of the LE to the H amplitude of horizontal variability and provides justification for ignoring uncorrelated smaller-scale excursions in the H and LE plots when comparing HRLDAS results to observations.
The “control” HRLDAS run with default parameters departed significantly from the observations, with H overestimated, LE underestimated, and a slope (-0.9) that was shallower than observed (-1.6), and none of the observed south-south trends.. In order to replicate the magnitudes and horizontal trends of all three variables, it was necessary to (a) adjust the constant in the equation relating the roughness lengths for momentum and heat/moisture from 0.1 to 0.5, and (b) adjust the near-surface soil moisture values to reflect the northward decrease. Adjusting green vegetation fraction had only a minor effect, but the range of values used was small since vegetation was sparse. We speculate that accounting for puddles in the soaked southern part of the track would improve agreement even more, as would allowing for variation in soil albedo.
Supplementary URL: