Abstract

Skillful seasonal forecasts can provide useful information for decision-makers, particularly in regions heavily dependent on agriculture, such as East Africa. We analyze prediction skill for seasonal East African rainfall and temperature one to four months ahead from two seasonal forecasting systems: the U.S. National Centers for Environmental Prediction (NCEP) Coupled Forecast System Model, version 2 (CFSv2), and the Met Office (UKMO) Global Seasonal Forecast System, version 5 (GloSea5). We focus on skill for low or high temperature and rainfall, below the 25th or above the 75th percentile, respectively, as these events can have damaging effects in this region. We find skill one month ahead for both low and high rainfall from CFSv2 for December–February in Tanzania, and from GloSea5 for September–November in Kenya. Both models have higher skill for temperature than for rainfall across Ethiopia, Kenya, and Tanzania, with skill two months ahead in some cases. Performance for rainfall and temperature change in the two models during certain El Niño–Southern Oscillation (ENSO) and Indian Ocean dipole (IOD) phases, the impacts of which vary by country, season, and sometimes by model. While most changes in performance are within the range of uncertainty due to the relatively small sample size in each phase, they are significant in some cases. For example, La Niña lowers performance for Kenya September–November rainfall in CFSv2 but does not affect skill in GloSea5.

Abstract

Subseasonal heatwave-driven concurrent hot and dry extreme events (HDEs) can cause substantial damage to crops, and hence to lives and livelihoods. However, the physical processes that lead to these devastating events are not well understood. Based on observations and reanalysis data for 1979–2016 over China, we show that HDEs occur preferentially over central and eastern China (CEC) and southern China (SC), with a maximum of three events per year along the Yangtze Valley. The probability of longer-lived and potentially more damaging HDEs is larger in SC than in CEC. Over SC the key factors of HDEs—positive anomalies of surface air temperature and evapotranspiration, and negative anomalies of soil moisture—begin two pentads before maximizing at the peak of the HDEs. These anomalies occur south of a positive height anomaly at 200 hPa, associated with a large-scale subsidence anomaly. The processes over CEC are similar to those for SC, but the anomalies begin one pentad before the peak. HDE frequency is strongly related to the Silk Road pattern and the boreal summer intraseasonal oscillation. Positive phases of the Silk Road pattern and suppressed phases of the boreal summer intraseasonal oscillation are associated with positive height anomalies over CEC and SC, increasing HDE frequency by about 35%–54% relative to the climatological mean. Understanding the effects of subseasonal and seasonal atmospheric circulation variability, such as the Silk Road pattern and boreal summer intraseasonal oscillation, on HDEs is important to improve HDE predictions over China.

Abstract

Anomalies in Siberian snow cover have been shown to affect Eurasian winter climate through the North Atlantic Oscillation (NAO). The existence of a teleconnection between North American snow cover and the NAO is far less certain, particularly for limited, regional snow cover anomalies. Using three ensembles of the Community Atmosphere Model, version 2 (CAM2), the authors examined teleconnections between persistent, forced snow cover in the northern Great Plains of the United States and western Eurasian winters. One ensemble allowed the model to freely determine global snow cover, while the other two forced a 72-cm snowpack centered over Nebraska. Of the forced ensembles, the “early-season” (“late season”) simulations initiated the snowpack on 1 November (1 January). The additional snow cover generated lower (higher) sea level pressures and geopotential heights over Iceland (the Azores) and warmer (cooler) temperatures over northern and western (eastern and southeastern) Europe, which suggests the positive NAO phase.

Differences between the free-snow-cover and early-season ensembles were never significant until January, which implied either that the atmospheric response required a long lag or that the late-winter atmosphere was particularly sensitive to Great Plains snow. The authors rejected the former hypothesis and supported the latter by noting similarities between the early- and late-season ensembles in late winter for European 2-m temperatures, transatlantic circulation, and an NAO index. Therefore, a regional North American snow cover anomaly in an area of high inter- and intra-annual snow cover variability can show a stronger teleconnection to European winter climate than previously reported for broader snow cover anomalies. In particular, anomalous late-season snow in the Great Plains may shift the NAO toward the positive phase.

Abstract

This paper reports a consistent seesaw relationship between interdecadal precipitation variability over North China and the Southwest United States, which can be found in observations and simulations with several models. Idealized model simulations suggest the seesaw could be mainly driven by the interdecadal Pacific oscillation (IPO), through a large-scale circulation anomaly occupying the entire northern North Pacific, while the Atlantic multidecadal oscillation (AMO) contributes oppositely and less. Modulation of precipitation by the IPO tends to be intensified when the AMO is in the opposite phase, but weakened when the AMO is in the same phase. The warm IPO phase is associated with an anomalous cyclone over the northern North Pacific; consequently, anomalous southwesterly winds bring more moisture and rainfall to the Southwest United States, while northwesterly wind anomalies prevail over North China with negative rainfall anomalies. The east–west seesaw of rainfall anomalies reverses sign when the circulation anomaly becomes anticyclonic during the cold IPO phase. The IPO-related tropical SST anomalies affect the meridional temperature gradient over the North Pacific and adjacent regions and the mean meridional circulation. In the northern North Pacific, the atmospheric response to IPO forcing imposes an equivalent barotropic structure throughout the troposphere. An important implication from this study is the potential predictability of drought-related water stresses over these arid and semiarid regions, with the progress of our understanding and prediction of the IPO and AMO.

Abstract

There is still no consensus about the best methodology for attributing observed changes in climate or climate events. One widely used approach relies on experiments in which the time periods of interest are simulated using an atmospheric general circulation model (AGCM) forced by prescribed sea surface temperatures (SSTs), with and without estimated anthropogenic influences. A potential limitation of such experiments is the lack of explicit atmosphere–ocean coupling; therefore a key question is whether the attribution statements derived from such studies are in fact robust. In this research the authors have carried out climate model experiments to test attribution conclusions in a situation where the answer is known—a so-called perfect model approach. The study involves comparing attribution conclusions for decadal changes derived from experiments with a coupled climate model (specifically an AGCM coupled to an ocean mixed-layer model) with conclusions derived from parallel experiments with the same AGCM forced by SSTs derived from the coupled model simulations. Results indicate that attribution conclusions for surface air temperature changes derived from AGCM experiments are generally robust and not sensitive to air–sea coupling. However, changes in seasonal mean and extreme precipitations, and circulation in some regions, show large sensitivity to air–sea coupling, notably in the summer monsoons over East Asia and Australia. Comparison with observed changes indicates that the coupled simulations generally agree better with observations. These results demonstrate that the AGCM-based attribution method has limitations and may lead to erroneous attribution conclusions in some regions for local circulation and mean and extreme precipitation. The coupled mixed-layer model used in this study offers an alternative and, in some respects, superior tool for attribution studies.

Abstract

The Silk Road pattern (SRP) teleconnection manifests in summer over Eurasia, where it is associated with substantial temperature and precipitation anomalies. The SRP varies on interannual and decadal scales; reanalyses show an increase in its decadal variability around the mid-1970s. Understanding what drives this decadal variability is particularly important, because contemporary seasonal prediction models struggle to predict the phase of the SRP. Based on analysis of observations and multiple targeted numerical experiments, this study proposes a mechanism for decadal SRP variability. Causal effect network analysis confirms a positive feedback loop between the eastern portion of the SRP pattern and vertical motion over India on synoptic time scales. Anomalies over a larger region of subtropical South Asia can reinforce a background state that projects onto the positive or negative SRP through this mechanism. This effect is isolated and confirmed in targeted numerical simulations. The transition from weak to strong decadal variability in the mid-1970s is consistent with more spatially coherent interannual precipitation variability over subtropical South Asia. Furthermore, results suggest that oceanic variability does not directly force the SRP. Nevertheless, sea surface temperatures in the North Atlantic and the North Pacific may indirectly affect the SRP by modulating South Asian rainfall on decadal time scales.

Abstract

The tropical west Pacific Ocean and the Philippines are often affected by tropical cyclones (TCs), with threats to human life and of severe economic damage. The performance of the Met Office global operational forecasts at predicting TC-related precipitation is examined between 2006 and 2017, the first time total TC rainfall has been analyzed in a long-term forecast dataset. All precipitation falling within 5° of a TC track point is assumed to be part of the TC rainbands. Forecasts are verified against TC tracks from the JRA-55 reanalysis and precipitation from TRMM 3B42. In composites from the forecasts, the total precipitation (TC and non-TC) is too high and the TC-related precipitation is too low, over both ocean and the Philippines. These biases exist all year-round and generally worsen with lead time, but have improved in recent years with upgrades to the forecasting system. Biases in TC-related precipitation in the Philippines are attributable mainly to TC lifetime being too short over land and ocean and (over land) possibly to individual TCs producing too little rain. There are considerable biases in predicted large-scale conditions related to TC intensification, particularly too little lower-troposphere relative humidity and too strong vertical wind shear. The shear appears to have little impact on the amount of TC precipitation, but dry biases in humidity are consistent with dry biases in TC rainfall. The forecast system accurately reproduces the impact of the MJO on TC precipitation, relative to the forecasts’ own climatology, potentially providing the opportunity for predictability out to several weeks.

Abstract

Canopy interception of incident precipitation is a critical component of the forest water balance during each of the four seasons. Models have been developed to predict precipitation interception from standard meteorological variables because of acknowledged difficulty in extrapolating direct measurements of interception loss from forest to forest. No known study has compared and validated canopy interception models for a leafless deciduous forest stand in the eastern United States. Interception measurements from an experimental plot in a leafless deciduous forest in northeastern Maryland (39°42′N, 75°50′W) for 11 rainstorms in winter and early spring 2004/05 were compared to predictions from three models. The Mulder model maintains a moist canopy between storms. The Gash model requires few input variables and is formulated for a sparse canopy. The WiMo model optimizes the canopy storage capacity for the maximum wind speed during each storm. All models showed marked underestimates and overestimates for individual storms when the measured ratio of interception to gross precipitation was far more or less, respectively, than the specified fraction of canopy cover. The models predicted the percentage of total gross precipitation (P_G ) intercepted to within the probable standard error (8.1%) of the measured value: the Mulder model overestimated the measured value by 0.1% of P_G ; the WiMo model underestimated by 0.6% of P_G ; and the Gash model underestimated by 1.1% of P_G . The WiMo model’s advantage over the Gash model indicates that the canopy storage capacity increases logarithmically with the maximum wind speed. This study has demonstrated that dormant-season precipitation interception in a leafless deciduous forest may be satisfactorily predicted by existing canopy interception models.

Abstract

While the Indian monsoon exhibits substantial variability on interannual time scales, its intraseasonal variability (ISV) is of greater magnitude and hence of critical importance for monsoon predictability. This ISV comprises a 30–50-day northward-propagating oscillation (NPISO) between active and break events of enhanced and reduced rainfall, respectively, over the subcontinent. Recent studies have implied that coupled general circulation models (CGCMs) were better able to simulate the NPISO than their atmosphere-only counterparts (AGCMs). These studies have forced their AGCMs with SSTs from coupled integrations or observations from satellite-based infrared sounders, both of which underestimate the ISV of tropical SSTs.

The authors have forced the 1.25° × 0.83° Hadley Centre Atmospheric Model (HadAM3) with a daily, high-resolution, observed SST analysis from the United Kingdom National Center for Ocean Forecasting that contains greater ISV in the Indian Ocean than past products. One ensemble of simulations was forced by daily SSTs, a second with monthly means, and a third with 5-day means. The ensemble with daily SSTs displayed significantly greater variability in 30–50-day rainfall across the monsoon domain than the ensemble with monthly mean SSTs, variability similar to satellite-derived precipitation analyses. Individual ensemble members with daily SSTs contained intraseasonal events with a strength, a propagation speed, and an organization that closely matched observed events. When ensemble members with monthly mean SSTs displayed power in intraseasonal rainfall, the events were weak and disorganized, and they propagated too quickly. The ensemble with 5-day means had less intraseasonal rainfall variability than the ensemble with daily SSTs but still produced coherent NPISO-like events, indicating that SST variability at frequencies higher than 5 days contributes to but is not critical for simulations of the NPISO.

It is concluded that high-frequency SST anomalies not only increased variance in intraseasonal rainfall but helped to organize and maintain coherent NPISO-like convective events. Further, the results indicate that an AGCM can respond to realistic and frequent SST forcing to generate an NPISO that closely resembles observations. These results have important implications for simulating the NPISO in AGCMs and coupled climate models, as well as for predicting tropical ISV in short- and medium-range weather forecasts.

Abstract

The northward-propagating intraseasonal (30–40 day) oscillation (NPISO) between active and break monsoon phases exerts a critical control on summer-season rainfall totals over India. Advances in diagnosing these events and comprehending the physical mechanisms behind them may hold the potential for improving their predictability. While previous studies have attempted to extract active and break events from reanalysis data to elucidate a composite life cycle, those studies have relied on first isolating the intraseasonal variability in the record (e.g., through bandpass filtering, removing harmonics, or empirical orthogonal function analysis). Additionally, the underlying physical processes that previous studies have proposed have varied, both among themselves and with studies using general circulation models.

A simple index is defined for diagnosing NPISO events in observations and reanalysis, based on lag correlations between outgoing longwave radiation (OLR) over India and over the equatorial Indian Ocean. This index is the first to use unfiltered OLR observations and so does not specifically isolate intraseasonal periods. A composite NPISO life cycle based on this index is similar to previous composites in OLR and surface winds, demonstrating that the dominance of the intraseasonal variability in the monsoon climate system eliminates the need for more complex methods (e.g., time filtering or EOF analysis) to identify the NPISO. This study is also among the first to examine the NPISO using a long-period record of high-resolution sea surface temperatures (SSTs) from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager. Application of this index to those SSTs demonstrates that SST anomalies exist in near quadrature with convection, as suggested by recent coupled model studies. Analysis of the phase relationships between atmospheric fields and SSTs indicates that the atmosphere likely forced the SST anomalies. The results of this lag-correlation analysis suggest that the oscillation serves as its own most reliable—and perhaps only—predictor, and that signals preceding an NPISO event appear first over the Indian subcontinent, not the equatorial Indian Ocean where the events originate.