The compensating effect of glaciers: Characterizing the relation between interannual streamflow variability and glacier cover

Meltwater from glaciers is not only a stable source of water but also affects downstream streamflow dynamics. One of these dynamics is the interannual variability of streamflow. Glaciers can moderate streamflow variability because the runoff in the glacierized part, driven by temperature, correlates negatively with the runoff in the non-glacierized part of a catchment, driven by precipitation, thereby counterbalancing each other. This is also called the glacier compensation effect (GCE), and the effect is assumed to depend on relative glacier cover. Previous studies found a convex relationship between streamflow variability and glacier cover of different glacierized catchments, with lowest streamflow variability at a certain optimum glacier cover. In this study, we aim to revisit these previously found curves to find out if a universal relationship between interannual streamflow variability and glacier cover exists, which could potentially be used in a space-for-time substitution analysis. Moreover, we test the hypothesis that the dominant climate drivers (here precipitation and temperature) switch around the suggested optimum of the curve. First, a set of virtual nested catchments, with the same absolute glacier area but varying non-glacierized area, were modelled to isolate the effect of glacier cover on streamflow variability. The modelled relationship was then compared with a multicatchment data set of gauged glacierized catchments in the European Alps. In the third step, changes of the GCE curve over time were analysed. Model results showed a convex relationship and the optimum in the simulated curve aligned with a switch in the dominant climate driver. However, the multicatchment data and the time change analyses did not suggest the existence of a universal convex relationship. Overall, we conclude that GCE is complex due to entangled controls and changes over time in glacierized catchments. Therefore, care should be taken to use a GCE curve for estimating and/or predicting interannual streamflow variability in glacierized catchments.     

Estimating thermohaline variability of the equatorial Pacific Ocean from satellite altimetry

The vertical thermohaline structure in the western equatorial Pacific is examined with a Gravest Empirical Mode (GEM) diagnosis of in-situ mooring measurements. The poor GEM performance in estimating deep thermohaline variability from satellite altimetry confirms a lack of vertical coherence in the equatorial ocean. Mooring observation reveals layered equatorial water with phase difference up to 6 months between thermocline and sub-thermocline variations. The disjointed layers reflect weak geostrophy and resemble pancake structures in non-rotating stratified turbulence. A coherency theorem is then proved, stating that traditional stationary GEM represents in-phase coherent structure and can not describe vertically out-of-phase variability. The fact that stationary GEM holds both spatial and temporal coherence makes it a unique tool to diagnose vertical coherent structure in geophysical flows. The study also develops a non-stationary GEM projection that captures more than 40% of the thermohaline variance in the equatorial deep water.     

Characterizing the spatial variability of groundwater quality using the entropy theory: I. Synthetic data

This paper, the first in a series of two, applies the entropy (or information) theory to describe the spatial variability of synthetic data that can represent spatially correlated groundwater quality data. The application involves calculating information measures such as transinformation, the information transfer index and the correlation coefficient. These measures are calculated using discrete and analytical approaches. The discrete approach uses the contingency table and the analytical approach uses the normal probability density function. The discrete and analytical approaches are found to be in reasonable agreement. The analysis shows that transinformation is useful and comparable with correlation to characterize the spatial variability of the synthetic data set, which is correlated with distance. Copyright © 2004 John Wiley & Sons, Ltd.     

Regional and interannual variability of ecosystem dynamics in the Southern Ocean

To investigate regional and interannual variability of the ecosystem in the Southern Ocean, a coupled circumpolar ice–ocean–plankton model has been developed. The ice–ocean component (known as BRIOS-2) is based on a modified version of the s-coordinate primitive equation model (SPEM) coupled to a dynamic–thermodynamic sea-ice model. The biological model (BIMAP) comprises two biogeochemical cycles – silica and nitrogen – and a prognostic iron compartment to include possible effects of micronutrient limitation. Simulations with the coupled ice–ocean–plankton model indicate that the physical–biological interaction is not limited to the effect of a varying surface mixed-layer depth. In the Pacific sector, large anomalies in winter mixed-layer depth cause an increased iron supply and enhance primary production and plankton biomass in the following summer, whereas in the Atlantic sector variability in primary production is caused mainly by fluctuations of oceanic upwelling. Thus, the Antarctic Circumpolar Wave (ACW) induces regional oscillations of phytoplankton biomass in both sectors, but not a propagating signal. Furthermore, interannual variability in plankton biomass and primary production is strong in the Coastal and Continental Shelf Zone and the Seasonal Ice Zone around the Antarctic continent. Interannual variability induced by the ACW has large effects on the regional scale, but the associated variability in biogenic carbon fluxes is small compared to the long-term carbon sequestration of the Southern Ocean.     

Global ENSO-streamflow teleconnection,streamflow forecasting and interannual variability

Abstract

El Niño Southern Oscillation (ENSO) has been linked to climate anomalies throughout the world. This paper presents an overview of global ENSO-streamflow teleconnection and identifies regions where the relationship may be exploited to forecast streamflow several months ahead. The teleconnection is investigated by fitting a first harmonic to 24-month El Niño streamflow composites from 581 catchments worldwide and the potential for forecasting is investigated by calculating the lag correlation between streamflow and two indicators of ENSO. The analyses indicate clear ENSO-streamflow teleconnections in many catchments, some of which are consistent across large geographical regions. Strong and regionally consistent ENSO-streamflow teleconnections are identified in Australia and New Zealand, South and Central America, and weaker signals are identified in some parts of Africa and North America. The results suggest that the ENSO-streamflow relationship and the serial correlation in streamflow can be used to successfully forecast streamflow. The streamflow forecasts can be used to help manage water resources, particularly in systems with high interannual variability in Australia, southern and drier parts of Africa and some areas of North America.     

Characterizing the variability of transit time distributions and young water fractions in karst catchments using flux tracking

Hydrological and biogeochemical processes in karst environments are strongly controlled by heterogeneous fracture-conduit networks. Quantifying the spatio-temporal variability of water transit time and young water fractions in such heterogeneous hydrogeological systems is fundamental to linking discharge and water quality dynamics in the karst critical zone. We used a tracer-aided conceptual hydrological model to track the fate of each hour of rain input individually. Using this approach, the variability of transit time distributions and young water fraction were estimated in the main landscape units in a karst catchment of Chenqi in Guizhou Province, Southwest China. The model predicted that the mean young water (i.e., <~2 months old) fraction of ground conduit flow is 0.31. Marked seasonal variabilities in water storage and hydrological connectivity between the conduit network and fractured matrix, as well as between hillslopes and topographic depression, drive the dynamics of young water fraction and travel time distributions in each landscape unit. Especially, the strong hydrological connectivity between the land surface and underground conduits caused by the direct infiltration through large fractures and sinkholes, leads the drastic increasement in young water fraction of runoff after heavy rain. Even though the contribution of young water to runoff is greater, the strong mixing and drainage of small fractures accelerate the old water release during high flows during the wet season. It is notable that the young water may sometimes be the most contaminated component contributing to the underground conduit network in karst catchments, because of the direct transfer of contaminants from the ground surface with rain water via large fractures and sinkholes.     

Controls on the interannual variability of hypoxia in a subtropical embayment and its adjacent waters in the Guangdong coastal upwelling system,northern South China Sea

Coastal embayments located downwind of large rivers under an upwelling-favorable wind are prone to develop low-oxygen or hypoxic conditions in their bottom water. One such embayment is Mirs Bay, off the Guangdong coast, which is affected by upwelling and the Pearl River Estuary (PRE) plume during summer. The relative importance of physical and biochemical processes on the interannual variability of hypoxia in Mirs Bay and its adjacent waters was investigated using statistical analyses of monthly hydrographic and water quality monitoring data from 2001 to 2015. The results reveal that the southwesterly wind duration and the PRE river discharge together explain 49% of the interannual variability in the size of the hypoxic area, whereas inclusion of the nutrient concentrations inside Mirs Bay and phytoplankton on the shelf explains 75% of the interannual variability in the size of the hypoxic area. This finding suggests that the interannual variability of hypoxia in Mirs Bay is regulated by coupled physical and biochemical processes. Increase of the hypoxic area under a longer-lasting southwesterly wind is caused by increased stratification, extended bottom water residence time, and onshore transport of a low-oxygen water mass induced by stable upwelling. In contrast, a reduction in the size of the hypoxic area may be attributed to a decrease in the surface water residence time of the particulate organic matter outside Mirs Bay due to increased discharge from the PRE. The results also show that the effects of allochthonous particulate organic matter outside Mirs Bay on bottom hypoxia cannot be neglected.     

Assessing the impact of interannual variability of precipitation and potential evaporation on evapotranspiration

The impact of interannual variability of precipitation and potential evaporation on the long-term mean annual evapotranspiration as well as on the interannual variability of evapotranspiration is studied using a stochastic soil moisture model within the Budyko framework. Results indicate that given the same long-term mean annual precipitation and potential evaporation, including interannual variability of precipitation and potential evaporation reduces the long-term mean annual evapotranspiration. This reduction effect is mostly prominent when the dryness index (i.e., the ratio of potential evaporation to precipitation) is within the range from 0.5 to 2. The maximum reductions in the evaporation ratio (i.e., the ratio of evapotranspiration to precipitation) can reach 8–10% for a range of coefficient of variation (CV) values for precipitation and potential evaporation. The relations between the maximum reductions and the CV values of precipitation and potential evaporation follow power laws. Hence the larger the interannual variability of precipitation and potential evaporation becomes, the larger the reductions in the evaporation ratio will be. The inclusion of interannual variability of precipitation and potential evaporation also increases the interannual variability of evapotranspiration. It is found that the interannual variability of daily rainfall depth and that of the frequency of daily rainfall events have quantitatively different impacts on the interannual variability of evapotranspiration; and they also interact differently with the interannual variability of potential evaporation. The results presented in this study demonstrate the importance of understanding the role of interannual variability of precipitation and potential evaporation in land surface hydrology under a warming climate.     

Seasonal and interannual thermohaline variability in the Guaymas Basin of the Gulf of California

Temperature and salinity data for the years 1939–1983 are used to investigate seasonal and interannual scales of the hydrographic variability across the Guaymas Basin, which is located between 27° and 28°N in the Central Gulf of California. Winter conditions extend from December to April and summer conditions from June to October, with transition periods in May and November. Sea surface temperature increases from about 16°C in February–March to 31°C in August. No clear seasonal cycle in surface salinity was found. Typical values are above 35.1‰ even in winter, and up to 35.5‰ in November.Relatively cold and low salinity near-surface waters observed in June 1957 and in June 1982, suggest advection of California Current Water to the Guaymas Basin. Subtropical Subsurface Water may occur around the year, but is obscured by vertical mixing with Gulf Water mainly during winter, when vertical stratification is weaker. The Intermediate and Deep Pacific Water masses successively fill the Guaymas Basin to the bottom (2000 m), showing very stable T-S characteristics.Positive sea level anomalies at Guaymas increases during El Nin˜o years, and anomalous low salinity and high temperature at the surface indicate the presence in the Guaymas Basin of water from the south. Observed differences reached 0.4‰ in surface salinity and 3°–5°C in surface temperature. There is evidence that the observed low salinities could not be due to abundant precipitation. An additional effect is a deepening of the winter pycnocline down about 200 m, compared to the usual depth of <100 m. In summer, this effect is not as clear as in winter, due to the strong stratification. The effects of the very strong 1982–1983 and 1957–1958 ENSO episodes may have lasted for one and two years, respectively. It is argued that during an ENSO event the Transition Water of the California Current meet and mix near the Gulf entrance with the Tropical Surface Water of the Costa Rica Coastal Current. This mixed water could have been carried north into the Gulf by local surface circulation. In 1983 it was found at the surface in the Guaymas Basin above the southeastward flow of the colder and saltier Gulf Water.     

Two interannual variability modes of the Northwestern Pacific Subtropical Anticyclone in boreal summer

Using the reanalysis data and 20th century simulation of coupled model FGOALS_gl developed by LASG/IAP, we identified two distinct interannual modes of Northwestern Pacific Subtropical Anticyclone (NWPAC) by performing Empirical Orthogonal Function (EOF) analysis on 850 hPa wind field over the northwestern Pacific in summer. Based on the associated anomalous equatorial zonal wind, these two modes are termed as "Equatorial Easterly related Mode" (EEM) and "Equatorial Westerly related Mode" (EWM), respectively. The formation mechanisms of these two modes are similar, whereas the maintenance mechanisms, dominant periods, and the relationships with ENSO are different. The EEM is associated with El Ni o decaying phase, with the anomalous anticyclone established in the preceding winter and persisted into summer through local positive air-sea feedback. By enhancing equatorial upwelling of subsurface cold water, EEM favors the transition of ENSO from El Ni o to La Ni a. The EWM is accompanied by the El Ni o events with long persistence, with the anomalous anticyclone formed in spring and strengthened in summer due to the warm Sea Surface Temperature anomalies (SSTA) forcing from the equatorial central-eastern Pacific. The model well reproduces the spatial patterns of these two modes, but fails to simulate the percentage variance accounted for by the two modes. In the NCEP reanalysis (model result), EEM (EWM) appears as the firstmode, which accounts for 35.6% (68.2%) of the total variance.     

Characterizing the spatial variability of groundwater quality using the entropy theory: II. Case study from Gaza Strip

This paper, the second in the series, uses the entropy theory to describe the spatial variability of groundwater quality data sets. The application of the entropy theory is illustrated using the chloride observations obtained from a network of groundwater quality monitoring wells in the Gaza Strip, Palestine. The application involves calculating information measures, such as transinformation, the information transfer index and the correlation coefficient. These measures are calculated using a discrete approach, in which contingency tables are used. An exponential decay fitting approach was applied to the discrete models. The analysis shows that transinformation, as a function of distance, can be represented by the exponential decay curve. It also indicates that, for the data used in this study, the transinformation model is superior to the correlation model for characterizing the spatial variability. Copyright © 2004 John Wiley & Sons, Ltd.     

Temperature trends and interannual variability in the Strait of Georgia,British Columbia

A continuous 36 year long record of semi-monthly temperature profiles from the central Strait of Georgia, British Columbia is used to examine low frequency variability and trends through the water column. Decomposition of temperature anomalies into empirical orthogonal functions shows that the dominant mode accounts for 78% of the variance, while the principal component associated with this mode (PC1) is dominated by fluctuations on interannual time scales. To relate the variability within the Strait to that occurring over the northeast Pacific, PC1 is compared with anomalies in local air temperature, sea surface temperatures off the west coast of Vancouver Island, and upper ocean temperatures along Line-P. These comparisons suggest that much of the interannual variability observed in the Strait of Georgia occurs in response to large-scale atmospheric forcing over the northeast Pacific. However, following tropical El Niño events there are significant anomalies associated with processes occurring along the coastal oceanic wave guide. The strongest event in the entire record, the remarkable negative temperature anomaly of winter 1978/1979, appears to be associated with a deep water intrusion that was forced locally.     

欧亚地区夏季大气环流年际变化的关键区及亚洲夏季风的关联信号

本文利用资料分析和数值模拟方法研究了欧亚地区夏季大气环流的相关性及其与亚洲夏季风的关联信号,以期为欧亚地区的气候变异及可预测性研究提供科学依据.结果表明:欧亚区域同期(JJA)500 hPa高度场年际变化的关键区包括热带区、中纬度的贝加尔湖和巴尔喀什湖之间以及欧洲地中海附近地区;表面气温的关键区主要位于热带海洋;海平面气压的关键区包括热带的海洋性大陆区域、印度洋和非洲大陆赤道附近部分区域、中高纬的贝加尔湖与巴尔喀什湖之间的地区.另外,夏季大气环流年际变化的春季关键区明显西移/南退,特别是表面气温(其西太平洋区不再是关键区).公用气候系统模式CCSM4.0的大气模式在给定海温年际变化的情况下对于上述大气环流相关场及其关键区的模拟基本合理,其中500 hPa高度场的模拟结果较好,海平面气压场的结果逊之;对于同期和前期的结果,模式都有夸大西太平洋海温影响的倾向.对于东亚夏季风指数与大气环流的同期年际变化信号而言,其空间分布基本表现为以30°N为界呈西南东北向的波列状分布;其春季前期信号中,30°N以南的显著区几乎都位于海洋,30°N以北主要位于欧洲、巴尔喀什湖与贝加尔湖之间的地区.南亚夏季风指数的前期显著相关区比同期明显西移/南退.总之,模式的模拟结果和观测结果相当吻合,但其同期模拟结果比前期的更好一些.这些结果说明:模式对于大气环流年际变化的耦合变化信息的刻画是基本合理的,这为利用气候模式进行有关可预测性研究和降尺度预测研究奠定了基础.     

Characterizing and understanding the variability of streamflow drought indicators within the USA

ABSTRACT

Low streamflow conditions can have adverse consequences for society and river ecology. The variability and drivers of streamflow drought indicators within the USA were investigated using observed streamflow records from 603 gauges across the USA. The analysis was based on two main approaches: (i) low-flow magnitude indicators, and (ii) streamflow deficit indicators. First, we examined how streamflow drought indicators vary spatially across the USA. Second, we used a data-driven clustering method to identify spatial clusters for each indicator. Finally, we assessed the association with regional climate drivers. The results show that the spatial variability of low-flow magnitude indicators is significantly different from the deficit indicators. Further, our clustering approach identifies regions of spatial homogeneity, which can be linked to the extreme regional climate drivers and land–atmosphere interactions. The influence of regional climate on streamflow drought indicators varies more between clusters than between indicators.     

Cadmium and manganese distributions in the Hudson River estuary: interannual and seasonal variability

Surface waters collected along the salinity gradient of the Hudson River estuary in four cruises between 1995 and 1997 were size-fractionated into particulate (>0.45 μm), ‘dissolved' (<0.45 μm), colloidal (10 kDa, 0.45 μm) and low molecular weight (<10 kDa) phases. Dissolved Cd concentrations (range: 0.11–1.19 nM) in surface waters of the estuary appear to have decreased fourfold (from an average of 2.36 to 0.61 nM) over a 23-year period, since the initial analysis of samples collected in the 1970s by Klinkhammer and Bender [Estuar. Coastal Shelf Sci. 12 (1981) 629–643]. This interannual decline reflects improvement in sewage treatment and the elimination of industrial Cd sources to the Hudson River estuary. In contrast, dissolved Mn levels (range: 0.033–1.46 μM) have remained relatively constant over the same period of time, suggesting that anthropogenic sources have very limited impact on Mn concentrations in the estuary. The concentrations of both Cd and Mn appeared to strongly depend on the season and/or river discharge. The highest concentrations were detected under low freshwater discharge, implying that limited hydraulic flushing allows a build-up of metals in the water column. Although the decline in Cd levels within the estuary reflects a reduction in the magnitude of anthropogenic inputs, mass balance estimates indicated that current sources of Cd to the estuary include sewage discharges (in the lower estuary around Manhattan) and diagenetic remobilization from industrial Cd deposited in sediments nearly 2 decades ago (in the upper estuary near Foundry Cove). Moreover, under low river discharge, the sources considered in our model (sewage, riverine input, atmospheric deposition, and benthic fluxes) could account for no more than 60% of the Cd exported from the lower estuary to the ocean. This suggests that undefined sources such as ground water and inputs from other watersheds (e.g., Long Island Sound and Newark Bay) may potentially influence the water quality of the New York Harbor. The size-fractionated metal concentrations indicated that most of the traditionally defined ‘dissolved' Cd and Mn consisted of <10 kDa molecular weight species. High molecular weight colloidal species of Mn accounted for about 50% of the dissolved fraction at the riverine end-member and <5% at intermediate and high salinities. Colloidal Cd accounted for <6% of the dissolved phase throughout the estuary. Unlike the non-conservative excess (relative to ideal dilution of river water and seawater) of dissolved Mn observed along the estuary, high molecular weight colloidal Mn appeared to be removed at the head of the estuary. The small contribution of colloidal Cd and Mn to the ‘dissolved' phase suggests that remobilization from suspended particulate phases and/or from sediments occurs through the formation of small molecular weight species.     

The boreal spring stratospheric final warming and its interannual and interdecadal variability

Based on the daily NCEP/DOE reanalysis II data,dates of the boreal spring Stratospheric Final Warming(SFW) events during 1979–2010 are defined as the time when the zonal-mean zonal wind at the central latitudes(65°–75°N) of the westerly polar jet drops below zero and never recovers until the subsequent autumn.It is found that the SFW events occur successively from the mid to the lower stratosphere and averagely from the mid to late April with a temporal lag of about 13 days from 10 to 50 hPa.Over the past 32 years,the earliest SFW occurs in mid March whereas the latest SFW happens in late May,showing a clear interannual variability of the time of SFW.Accompanying the SFW onset,the stratospheric circulation transits from a winter dynamical regime to a summertime state,and the maximum negative tendency of zonal wind and the strongest convergence of planetary-wave are observed.Composite results show that the early/late SFW events in boreal spring correspond to a quicker/slower transition of the stratospheric circulation,with the zonal-mean zonal wind reducing about 20/5 m s-1 at 30 hPa within 10 days around the onset date.Meanwhile,the planetary wave activities are relatively strong/weak associating with an out-of-/in-phase circumpolar circulation anomaly before and after the SFW events in the stratosphere.All these results indicate that,the earlier breakdown of the stratospheric polar vortex(SPV),as for the winter stratospheric sudden warming(SSW) events is driven mainly by wave forcing;and in contrast,the later breakdown of the SPV exhibits more characteristics of its seasonal evolution.Nevertheless,after the breakdown of SPV,the polar temperature anomalies always exhibit an out-of-phase relationship between the stratosphere and the troposphere for both the early and late SFW events,which implies an intimate stratosphere–troposphere dynamical coupling in spring.In addition,there exists a remarkable interdecadal change of the onset time of SFW in the mid 1990s.On average,the SFW onset time before the mid 1990s is 11 days earlier than that afterwards,corresponding to the increased/decreased planetary wave activities in late winter-early spring before/after the 1990s.     

欧亚地区夏季大气环流年际变化的关键区及亚洲夏季风的关联信号

本文利用资料分析和数值模拟方法研究了欧亚地区夏季大气环流的相关性及其与亚洲夏季风的关联信号,以期为欧亚地区的气候变异及可预测性研究提供科学依据.结果表明:欧亚区域同期(JJA)500 hPa高度场年际变化的关键区包括热带区、中纬度的贝加尔湖和巴尔喀什湖之间以及欧洲地中海附近地区;表面气温的关键区主要位于热带海洋;海平面气压的关键区包括热带的海洋性大陆区域、印度洋和非洲大陆赤道附近部分区域、中高纬的贝加尔湖与巴尔喀什湖之间的地区.另外,夏季大气环流年际变化的春季关键区明显西移/南退,特别是表面气温(其西太平洋区不再是关键区).公用气候系统模式CCSM4.0的大气模式在给定海温年际变化的情况下对于上述大气环流相关场及其关键区的模拟基本合理,其中500 hPa高度场的模拟结果较好,海平面气压场的结果逊之;对于同期和前期的结果,模式都有夸大西太平洋海温影响的倾向.对于东亚夏季风指数与大气环流的同期年际变化信号而言,其空间分布基本表现为以30°N为界呈西南东北向的波列状分布;其春季前期信号中,30°N以南的显著区几乎都位于海洋,30°N以北主要位于欧洲、巴尔喀什湖与贝加尔湖之间的地区.南亚夏季风指数的前期显著相关区比同期明显西移/南退.总之,模式的模拟结果和观测结果相当吻合,但其同期模拟结果比前期的更好一些.这些结果说明:模式对于大气环流年际变化的耦合变化信息的刻画是基本合理的,这为利用气候模式进行有关可预测性研究和降尺度预测研究奠定了基础.     

中国降水年际和年代际变率对空间尺度的敏感性

利用中国740站45年降水资料按5种分辨率分气候区计算了降水年际和年代际变率. 降水年际和年代际变率对空间尺度的敏感性分析表明,中国各气候区降水年际变率对空间尺度的敏感性都随空间尺度的增加而逐渐减小,且存在明显的季节变化,而年代际变率对空间尺度的敏感性却随空间尺度的增加而增大,但不存在季节变化;由于中国各气候区降水的特殊性,各气候区降水年际和年代际变率对空间尺度的敏感程度存在不可忽视的差异.在年际和年代际尺度上,西南地区降水变率对空间尺度都是最敏感的,因而该区域降水年际和年代际变率信号的检测最困难.而华南地区在年际尺度上比较敏感,年代际尺度却不敏感,但华南地区在年际和年代际尺度上区域内降水分布的非均匀程度对空间尺度的敏感性都最大.     

Vegetation response to rainfall seasonality and interannual variability in tropical dry forests

Projected changes in rainfall seasonality and interannual variability are expected to have severe impacts on arid and semi‐arid tropical vegetation, which is characterized by a fine‐tuned adaptation to extreme rainfall seasonality. To study the response of these ecosystems and the related changes in hydrological processes to changes in the amount and seasonality of rainfall, we focused on the caatinga biome, the typical seasonally dry forest in semi‐arid Northeast Brazil. We selected four sites across a gradient of rainfall amount and seasonality and analysed daily rainfall and biweekly Normalized Difference Vegetation Index (NDVI) data for hydrological years 2000 to 2014. Rainfall seasonal and interannual statistics were characterized by recently proposed metrics describing duration, timing and intensity of the wet season and compared to similar metrics of NDVI time series. The results show that the caatinga tends to have a more stable response with longer and less variable growing seasons (3.1 ± 0.1 months) compared to the duration wet seasons (2.0 ± 0.5 months). The ecosystem ability to buffer the interannual variability of rainfall is also evidenced by the stability in the timing of the growing season compared to the wet season, which results in variable delays (ranging from 0 to 2 months) between the peak of the rainfall season and the production of leaves by the ecosystem. The analyses show that the shape and size of the related hysteresis loops in the rainfall–NDVI relations are linked to the buffering effects of soil moisture and plant growth dynamics. Finally, model projections of vegetation response to different rainfall scenarios reveal the existence of a maximum in ecosystem productivity at intermediate levels of rainfall seasonality, suggesting a possible trade‐off in the effects of intensity (i.e. amount) and duration of the wet season on vegetation growth and related soil moisture dynamics and transpiration rates. Copyright © 2016 John Wiley & Sons, Ltd.