Abstract. In eastern East Africa (the southern Ethiopia, eastern Kenya and southern Somalia region), poor boreal spring (long wet season) rains in 1999, 2000, 2004, 2007, 2008, 2009, and 2011 contributed to severe food insecurity and high levels of malnutrition. Predicting rainfall deficits in this region on seasonal and decadal time frames can help decision makers implement disaster risk reduction measures while guiding climate-smart adaptation and agricultural development. Building on recent research that links more frequent East African droughts to a stronger Walker circulation, resulting from warming in the Indo-Pacific warm pool and an increased east-to-west sea surface temperature (SST) gradient in the western Pacific, we show that the two dominant modes of East African boreal spring rainfall variability are tied to SST fluctuations in the western central Pacific and central Indian Ocean, respectively. Variations in these two rainfall modes can thus be predicted using two SST indicesthe western Pacific gradient (WPG) and central Indian Ocean index (CIO), with our statistical forecasts exhibiting reasonable cross-validated skill (r cv ≈ 0.6). In contrast, the current generation of coupled forecast models show no skill during the long rains. Our SST indices also appear to capture most of the major recent drought events such as 2000, 2009 and 2011. Predictions based on these simple indices can be used to support regional forecasting efforts and land surface data assimilations to help inform early warning and guide climate outlooks.

[1] The turbulent heat fluxes play a pivotal role in the exchange of energy between the atmosphere and ocean. The calculation of these fluxes over the global oceans requires the use of bulk aerodynamic or flux-gradient methods that rely on estimates of the sea surface temperature (SST), near-surface wind speed, air temperature, and specific humidity. Errors in current methodologies of satellite retrievals of near-surface properties have been shown to be the main sources of error for calculation of the fluxes. A new neural network technique is presented here that significantly improves the error characteristics of the air temperature and specific humidity compared to previous methods. Improvements in predicting near-surface wind speed and SST are also seen. Additional improvements are also made by accounting for the effects of high cloud liquid water contents, the effects of which can be mitigated through the use of regime-specific linear and nonlinear retrieval methods. The use of a first-guess SST is shown to result in significant improvement in retrieval accuracy.

No abstract

Motivated by the question of whether recent interannual to decadal climate variability and a possible “climate shift” may have affected the global water balance, we examine precipitation minus evaporation (P – E) variability integrated over the global oceans and global land for the period 1979–2010 from three points of view—remotely sensed retrievals and syntheses over the oceans, reanalysis vertically integrated moisture flux convergence (VMFC) over land, and land surface models (LSMs) forced with observations-based precipitation, radiation, and near-surface meteorology. Over land, reanalysis VMFC and P − evapotranspiration (ET) from observationally forced LSMs agree on interannual variations (e.g., El Niño/La Niña events); however, reanalyses exhibit upward VMFC trends 3–4 times larger than P − ET trends of the LSMs. Experiments with other reanalyses using reduced observations show that upward VMFC trends in the full reanalyses are due largely to observing system changes interacting with assimilation model physics. The much smaller P − ET trend in the LSMs appears due to changes in frequency and amplitude of warm events after the 1997/98 El Niño, a result consistent with coolness in the eastern tropical Pacific sea surface temperature (SST) after that date. When integrated over the global oceans, E and especially P variations show consistent signals of El Niño/La Niña events. However, at scales longer than interannual there is considerable uncertainty especially in E. This results from differences among datasets in near-surface atmospheric specific humidity and wind speed used in bulk aerodynamic retrievals. The P variations, all relying substantially on passive microwave retrievals over ocean, also have uncertainties in decadal variability, but to a smaller degree.

The recently produced Modern Era Retrospective-Analysis for Research and Applications (MERRA; Rienecker et al. 2011) provides a high-resolution dataset that can be used to examine components of the Earth's surface energy and water balance. Latent and sensible heat exchanges between the ocean and atmosphere are fundamental components of these balances and are the focus of this study. The primary objectives are to characterize the MERRA surface energy fluxes with respect to their:1. Accuracy against direct measurements; 2. Large scale spatio-temporal variability and representation of extremes; 3. Connection to forcing by the data assimilation system. Summary Points1. MERRA produces estimates of the turbulent fluxes that agree very well with observational estimates for average conditions; however, it is distinct in amplitude with a particularly weak representation of the surface heat fluxes over the western boundary currents and in conditions of very weak and very strong near-surface stratification. A weaker covariability between wind speed and temperature/moisture stratification than observed exists.1. MERRA has slightly weaker seasonal variability of the latent and sensible heat fluxes compared to an observational ensemble estimate. It tends to under-represents the occurrence of strong, episodic events compared with observations in the Northern Hemisphere mid-latitudes.1. Data assimilation, as expected, tends to drive the analysis closer to the observational ensemble; the impact on near-surface variables contains a systematic response to the changing observing system and could introduce artificial trends into the analysis. Observational DatabaseHigh-quality, direct in situ measurements of the turbulent latent and sensible heat fluxes and near-surface variables serve as a standard against which the veracity of turbulent flux products are compared. The SEAFLUX (Curry et al. 2004) program has compiled a large dataset of these measurements and are utilized in this study for validation purposes. The spatial and temporal distribution of these observations are characterized below ( Fig. 1). Evaluating Large Scale VariabilityValidating surface heat flux estimates at large spatial and temporal scales relies on intercomparisons between multiple estimates and the use of physically based constraints. Further support is provided through the use of local or regional comparisons to direct observations. This study makes use of three additional products to characterize the large scale variability of the MERRA surface turbulent fluxes. These products and their primary data sources are: The ensemble mean of these products is used to characterize the annual mean (Fig. 3, right) estimate from MERRA. The differences between MERRA and observationally-based estimates show that MERRA captures the major patterns; however MERRA tends to underestimate the latent and sensible heat flux over the western boundary currents by 50Wm -2 and 15Wm -2 , respectively. There are outside the range (hatching) of any of the available observationally-base...