气候模拟日降水量的统计误差订正分析——以上海为例

对区域气候模式系统PRECIS在SRES A1B情景下模拟的上海日降水输出按季节进行了统计误差订正。该方法首先对降水日数进行比率订正,以消除模式产生的微小值降水。然后利用Γ分布拟合日降水量的累计概率分布,采用整体和分段拟合两种方法构建传递函数TF（Transfer Function）进行订正。选取1962年12月—1992年11月作为控制时段,构建TF并将其应用于验证时段（1992年12月—2002年11月）。该订正方案消除了模式产生的微小值降水,解决了模拟的小降水值偏多的问题,频率误差保持在1%以下,分段拟合订正相比整体拟合订正具有更强的对极端降水的订正能力;对冬、春季的订正效果比夏、秋季更显著。该方案不仅有效消除了平均值的漂移,而且显著订正了变率,同时提高了极端降水事件的再现能力,是一种相对完善的订正方案。      

基于分位数映射法的黑河上游气候模式降水误差订正

区域气候模式降水弥补了高寒山区气象站点稀少的缺陷,是水文模拟的重要驱动变量。然而,高寒山区模式输出降水的总量和频率都存在较大不确定性。因此,改进了用于降水频率纠正的分位数映射法(Quantile Mapping,QM),对中尺度数值预报模式(Weather Research and Forecasting model,WRF)模拟的黑河上游日降水输出数据进行误差订正。选取第95分位和第98分位降水量为阈值,选择2004-2009年为建模时段,2010-2013年为验证时段,使用分段拟合的方法建立传递函数,侧重于对极端降水进行单独订正。研究结果表明:该方法不仅对降水空间分布有明显的改善,对极端降水也有很好的订正效果。订正前模式模拟日降水与台站之间的均方根误差为3.41 mm·d^-1,绝对偏差为115.67 mm·y^-1,订正后均方根误差减少为3.11 mm·d^-1,绝对偏差有明显改善,为60.3 mm·y^-1。订正后流域内年降水空间分布更加合理,年降水量也更接近于观测降水插值结果,其空间相关系数由0.74改善为0.94。春、夏季订正效果优于秋、冬季,其中夏季订正效果较为明显,订正前降水偏差百分比在-0.1~0.1以内的区域面积仅占流域总面积的28%,而订正后占比增加至66%。同时,该方法对极端降水有较好的订正效果,减小了日降水强度(SDII)和极强降水量(R99p)的模拟偏差,订正后的第95分位模拟降水与观测降水插值的相关系数由0.15提高到0.48。本研究为站点稀少的黑河上游提供了一种更有效的误差订正方案,有利于为寒区水文研究获取更精确的降水数据。     

基于DERF2.0的华南前汛期降水订正

针对我国华南前汛期（4—6月）降水,基于国家气候中心第2代月动力延伸模式（DERF2.0）结果,利用非参数百分位映射方法将模式预测结果转化为概率预报,并进行概率订正。分别选用交叉建模与独立样本建模两种订正方法,并利用偏差、偏差百分率、时间相关系数、均方根误差等统计方法检验订正效果。结果表明:订正方法对预报技巧的改善与起报时间无显著相关,且具有误差稳定性,其订正效果受预报误差影响较小;与订正前模式预测降水落区的范围和平均强度相比,订正后结果与观测更接近;按百分位区间统计的不同强度降水订正预报均有明显改进;预测时段的订正效果与回报时段的订正效果基本一致。     

CWRF模式极端降水模拟误差订正

基于1980—2015年6—8月CWRF模式（Climate-Weather Research and Forecasting model）14种方案的模拟结果和全国逐日降水观测资料,对比了Q-lin,Q-tri,RQ-lin,RQ-tri,SSP-lin和CDFt 6种误差订正方法对CWRF模式控制化方案（C1）模拟中国东部夏季日极端降水的订正效果,以CWRF模式14种方案日极端降水的模拟效果排名为基础,对比了模拟效果较好的4种方案集合、模拟较差的4种方案集合以及14种方案集合的订正效果,选出相对较好的订正方案进一步评估其成员集合后订正和成员分别订正后再集合的订正效果,结果表明:采用6种误差订正方法均可明显减少日极端降水模拟误差,其中RQ-lin方法订正效果最佳。CWRF模式对中国东部的极端降水指数均表现出较好的模拟能力,不同参数化集合方案得到14种方案成员先订正再集合与观测日极端降水平均值最为接近,研究结果对于改进模拟结果、提高其预测能力有重要应用价值。     

分位数映射法在RegCM4中国气温模拟订正中的应用

将一种分位数映射法RQUANT,应用到一个区域气候模式（RegCM4）所模拟中国气温的误差订正中。从气候平均态、年际变率、极端气候及农业气候等多方面,评估了该方法对日平均气温、日最高气温和日最低气温模拟的订正效果。结果表明,该订正方法对模式模拟的日平均、日最高和最低气温气候平均态的订正效果都非常明显,中国大部分地区的订正结果与观测的偏差在±0.5℃之间。在降低极端气温指数和农业气候相关指数的模拟误差方面也有显著的效果,但对气温年际变率的订正效果有限。结合以往对降水订正的评估分析,该方法对模式模拟结果有较好的订正效果,可以应用于区域气候模式的气候变化模拟预估中,为气候变化及相关影响评估研究提供更适用和可靠的数据。     

QM和QDM方法对中国极端气候的高分辨率气候变化模拟的误差订正对比

采用分位数映射（Quantile Mapping, QM）和delta分位数映射（Quantile Delta Mapping, QDM）两种误差订正方法对区域气候模式RegCM4在中国区域内模拟的逐日气温和降水数据进行订正。模式数据是5种不同全球气候模式驱动下的区域模式气候变化模拟结果。计算订正前后的极端气候指数进行对比分析,包括日最高气温极大值（TXx）、日最低气温极小值（TNn）、连续干旱日数（CDD）和最大日降水量（RX1day）。结果表明,5组模拟结果和其集合平均（ensR）都显示气温指数的模拟效果高于降水指数,其中对TXx模拟最好,对CDD的模拟最差;经过订正后,针对不同模式的两种订正结果都能够有效地减小模式与观测的偏差并提高了空间相关系数,且两种方法的订正效果无明显差别。对RCP4.5情景下未来变化的分析中,QM在一定程度上改变了模式模拟的未来变化幅度和空间分布特征,QDM则能够有效地保留所有极端指数的气候变化信号。从全国平均来看,除CDD外,所有指数未来都呈现增加趋势,且QDM订正结果与订正前模式模拟的变化趋势更为接近。建议在气候变化模拟的误差订正中采用QDM方法。     

概率调整法在气候模式模拟降水量订正中的应用

应用概率调整法订正区域气候模式系统PRECIS在SRES A1B情景下模拟的各季节全国日降水量。以第95百分位降水量为阈值,利用Γ分布分段拟合1962年12月—1972年11月的模拟值,构建传递函数,得到1991年12月—2001年11月的订正值。结果表明:全国平均日降水量空间分布的模拟改善明显,偏差百分率高于100%的格点比例从23.5%降低到1.0%;对各地区平均降水月循环的模拟结果改善,冷季降水较暖季更接近观测,提高拟合优度是改进订正方法的关键;多数地区连续干日数、连续5 d最大降水量及极端降水贡献率的空间强度、概率分布与空间相关性的订正效果显著。总体来说,该方法对模拟中国区域降水的平均态与极端降水均有明显改善,有助于气候评估工作的展开。     

基于概率匹配的西南区域模式定量降水订正试验

利用2014—2015年5—10月地面观测降水资料和同时段的西南区域模式降水预报资料,基于概率匹配方法,采取分区及点对点匹配两种方案对2016年6—8月降水集中时段逐12h累积降水进行订正试验。结果表明:(1)订正后的模式预报相比订正前而言,平均(绝对)误差有所减小,降水落区的范围和平均强度与实况更加接近;(2)量级偏差越大,运用该方法的订正效果越好,夜间降水订正效果优于白天;(3)分区统计方案对模式系统性偏差的订正效果优于点对点方案,合理的区域划分增加统计样本量可以提高订正效果。     

基于CWRF模式夏季日降水量的订正技术对比

误差订正是提高模式模拟和预报性能的有效方法。基于CWRF(regional Climate-Weather Research and Forecasting model)25套不同物理参数化方案的日降水量模拟资料, 对比仅进行降水日数订正(OCD)、仅进行降水量订正(OCM)和先订正降水日数再订正降水量(COR)三种订正方法, 先订正再等权重集成和先等权重集成再订正两种订正思路, 重点对1997—2015年华中和华南地区夏季日降水进行订正效果的对比。结果表明:(1)降水日的订正是必要的, 综合而言COR方法对CWRF模式日降水的订正效果更佳, 尤其是小量级降水, 但降水强度的表现不如OCM; (2)先集成后订正的效果更好; (3) CWRF模式不同参数化方案对日降水的模拟能力有显著差别, 经过订正后模拟能力均有所提升, 但对于不同的模拟方案, 其订正效果也不同。表明, 误差订正确实能有效提高模式模拟及预报性能, 但其效果存在不确定性。提高模式的预报性能, 关键还是提高模式对真实大气动力学的表述能力。      

基于旋转经验正交分解的流域降水气候预测误差订正方法

流域尺度的降水短期气候预测水平对流域的防灾减灾具有重要价值。为了进一步提高中国科学院大气物理研究所新一代大气环流模式IAP AGCM 4.1在淮河和长江流域夏季降水预测效果,利用旋转经验正交分解（Rotated Empirical Orthogonal Function, REOF）方法对两个流域夏季降水区域特征进行分析的基础上,建立了一个适用于流域的分区经验正交分解（Empirical Orthogonal Function, EOF）订正方案,并利用IAP AGCM 4.1气候预测系统在两个流域的夏季降水共30年（1981～2010年）的集合回报试验结果进行了订正试验。结果表明分区订正方法明显改进了模式对淮河流域的夏季降水预测水平,淮河流域的流域平均相关系数从0.03提高到了0.22。对长江流域的季度降水预报也有显著的改进效果,平均相关系数从-0.05提高到0.24。分区订正结果明显优于流域整体订正方案,证明了基于REOF分析确定降水具有强局地性特征的订正区域,能够很好地提高EOF订正方法的效果稳定性,这对其他流域降水预测的订正研究具有很好的借鉴意义。     

Influence of SST biases on future climate change projections

We use a quantile-based bias correction technique and a multi-member ensemble of the atmospheric component of NCAR CCSM3 (CAM3) simulations to investigate the influence of sea surface temperature (SST) biases on future climate change projections. The simulations, which cover 1977?C1999 in the historical period and 2077?C2099 in the future (A1B) period, use the CCSM3-generated SSTs as prescribed boundary conditions. Bias correction is applied to the monthly time-series of SSTs so that the simulated changes in SST mean and variability are preserved. Our comparison of CAM3 simulations with and without SST correction shows that the SST biases affect the precipitation distribution in CAM3 over many regions by introducing errors in atmospheric moisture content and upper-level (lower-level) divergence (convergence). Also, bias correction leads to significantly different precipitation and surface temperature changes over many oceanic and terrestrial regions (predominantly in the tropics) in response to the future anthropogenic increases in greenhouse forcing. The differences in the precipitation response from SST bias correction occur both in the mean and the percent change, and are independent of the ocean?Catmosphere coupling. Many of these differences are comparable to or larger than the spread of future precipitation changes across the CMIP3 ensemble. Such biases can affect the simulated terrestrial feedbacks and thermohaline circulations in coupled climate model integrations through changes in the hydrological cycle and ocean salinity. Moreover, biases in CCSM3-generated SSTs are generally similar to the biases in CMIP3 ensemble mean SSTs, suggesting that other GCMs may display a similar sensitivity of projected climate change to SST errors. These results help to quantify the influence of climate model biases on the simulated climate change, and therefore should inform the effort to further develop approaches for reliable climate change projection.     

Multiyear hindcast simulations of summer monsoon over South Asia using a nested regional climate model��BCC_RegCM1.0

Multiyear (1983?C2006) hindcast simulation of summer monsoon over South Asia has been carried out using the regional climate model of the Beijing Climate Centre (BCC_RegCM1.0). The regional climate model (hereafter BCC RCM) is nested into the global climate model of the Beijing Climate Centre BCC_CGCM1.0 (here after CGCM). The regional climate model is initialized on 01 May and integrated up to the end of the September for 24?years. Compared to the driving CGCM the BCC RCM reproduces reasonably well the intensity and magnitude of the large-scale features associated with the South Asia summer monsoon such as the upper level anticyclone at 200?hPa, the mid-tropospheric warming over the Tibetan plateau, the surface heat low and the 850?hPa moisture transport from ocean to the land. Both models, i.e., BCC RCM and the driving CGCM overestimates (underestimates) the 850?hPa southwesterly flow over the northern (southern) Arabian Sea. Moreover, both models overestimate the seasonal mean precipitation over much of the South Asia region compared to the observations. However, the precipitation biases are significantly reduced in the BCC RCM simulations. Furthermore, both models simulate reasonably the interannual variability of the summer monsoon over India. The precipitation index simulated by BCC RCM shows significant correlation (0.62) with the observed one. The BCC RCM simulates reasonably well the spatial and temporal variation of the precipitation and surface air temperature compared to the driving CGCM. Further, the temperature biases are significantly reduced (1?C4°C) in the BCC RCM simulations. The simulated vertical structure of the atmosphere show biases above the four sub-regions, however, these biases are significantly reduced in the BCC RCM simulations compared to the driving CGCM. Compared to the driving CGCM, the evolution processes of the onset of summer monsoon, e.g., the meridional temperature gradient and the vertical wind shear are well simulated by the BCC RCM. The 24-year simulations also show that with a little exception the BCC RCM is capable to reproduce the monsoon active and break phases and the intraseasonal precipitation variation over the Indian subcontinent.     

Evaluation of 2-m temperature and precipitation products of the Climate Forecast System version 2 over Iran

Using a continuous multi-decadal simulations over the period 1981–2010, subseasonal to seasonal simulations of the Climate Forecast System version 2 (CFSv2) over Iran against the Climatic Research Unit (CRU) dataset are evaluated. CFSv2 shows cold biases over northern hillsides of the Alborz Mountains with the Mediterranean climate and warm biases over northern regions of the Persian Gulf and the Oman Sea with a dry climate. Magnitude of the model bias for 2-m temperature over different regions of Iran varies by season, with the least bias in temperate seasons of spring and autumn, and the largest bias in summer. The model bias decreases as temporal averaging period increases from seasonal to annual. The forecast generally produces dry and wet biases over dry and wet regions of Iran, respectively. In general, 2-m temperature over Iran is better captured than precipitation, but the prediction skill of precipitation is generally high over western Iran. Averaged over Iran, observations indicated that 2-m temperature has been gradually increasing during the studied period, with a rate of approximately 0.5 °C per decade, and the upward trend is well simulated by CFSv2. Averaged over Iran, both observations and simulation results indicated that precipitation has been decreasing in spring, with averaged decreasing trends of 0.8 mm (observed) and 1.7 mm (simulated) per season each year during the period 1981–2010. Observations indicated that the maximum increasing trend of 2-m temperature has occurred over western Iran (nearly 0.7 °C per decade), while the maximum decreasing trend of annual precipitation has occurred over western and parts of southern Iran (nearly 45 to 50 mm per decade).     

A regional climate model downscaling projection of China future climate change

Climate changes over China from the present (1990–1999) to future (2046–2055) under the A1FI (fossil fuel intensive) and A1B (balanced) emission scenarios are projected using the Regional Climate Model version 3 (RegCM3) nests with the National Center for Atmospheric Research (NCAR) Community Climate System Model (CCSM). For the present climate, RegCM3 downscaling corrects several major deficiencies in the driving CCSM, especially the wet and cold biases over the Sichuan Basin. As compared with CCSM, RegCM3 produces systematic higher spatial pattern correlation coefficients with observations for precipitation and surface air temperature except during winter. The projected future precipitation changes differ largely between CCSM and RegCM3, with strong regional and seasonal dependence. The RegCM3 downscaling produces larger regional precipitation trends (both decreases and increases) than the driving CCSM. Contrast to substantial trend differences projected by CCSM, RegCM3 produces similar precipitation spatial patterns under different scenarios except autumn. Surface air temperature is projected to consistently increase by both CCSM and RegCM3, with greater warming under A1FI than A1B. The result demonstrates that different scenarios can induce large uncertainties even with the same RCM-GCM nesting system. Largest temperature increases are projected in the Tibetan Plateau during winter and high-latitude areas in the northern China during summer under both scenarios. This indicates that high elevation and northern regions are more vulnerable to climate change. Notable discrepancies for precipitation and surface air temperature simulated by RegCM3 with the driving conditions of CCSM versus the model for interdisciplinary research on climate under the same A1B scenario further complicated the uncertainty issue. The geographic distributions for precipitation difference among various simulations are very similar between the present and future climate with very high spatial pattern correlation coefficients. The result suggests that the model present climate biases are systematically propagate into the future climate projections. The impacts of the model present biases on projected future trends are, however, highly nonlinear and regional specific, and thus cannot be simply removed by a linear method. A model with more realistic present climate simulations is anticipated to yield future climate projections with higher credibility.     

Simulations of precipitation using NRCM and comparisons with satellite observations and CAM: annual cycle

The accurate representation of rainfall in models of global climate has been a challenging task for climate modelers owing to its small space and time scales. Quantifying this variability is important for comparing simulations of atmospheric behavior with real time observations. In this regard, this paper compares both the statistical and dynamically forced aspects of precipitation variability simulated by the high-resolution (36?km) Nested Regional Climate Model (NRCM), with satellite observations from the Tropical Rainfall Measuring Mission (TRMM) 3B42 dataset and simulations from the Community Atmosphere Model (CAM) at T85 spatial resolution. Six years of rainfall rate data (2000?C2005) from within the Tropics (30°S?C30°N) have been used in the analysis and results are presented in terms of long-term mean rain rates, amplitude and phase of the annual cycle and seasonal mean maps of precipitation. Our primary focus is on characterizing the annual cycle of rainfall over four land regions of the Tropics namely, the Indian Monsoon, the Amazon, Tropical Africa and the North American monsoon. The lower tropospheric circulation patterns are analyzed in both the observations and the models to identify possible causes for biases in the simulated precipitation. The 6-year mean precipitation simulated by both models show substantial biases throughout the global Tropics with NRCM/CAM systematically underestimating/overestimating rainfall almost everywhere. The seasonal march of rainfall across the equator, following the motion of the sun, is clearly seen in the harmonic vector maps. The timing of peak rainfall (phase) produced by NRCM is in closer agreement with the observations compared to CAM. However like the long-time mean, the magnitude of seasonal mean rainfall is greatly underestimated by NRCM throughout the Tropical land mass. Some of these regional biases can be attributed to erroneous circulation and moisture surpluses/deficits in the lower troposphere in both models. Overall, the results seem to indicate that employing a higher spatial resolution (36?km) does not significantly improve simulation of precipitation. We speculate that a combination of several physics parameterizations and lack of model tuning gives rise to the observed differences between NRCM and the observations.     

High resolution regional climate model simulations for Germany: part I—validation

A five-member ensemble of regional climate model (RCM) simulations for Europe, with a high resolution nest over Germany, is analysed in a two-part paper: Part I (the current paper) presents the performance of the models for the control period, and Part II presents results for near future climate changes. Two different RCMs, CLM and WRF, were used to dynamically downscale simulations with the ECHAM5 and CCCma3 global climate models (GCMs), as well as the ERA40-reanalysis for validation purposes. Three realisations of ECHAM5 and one with CCCma3 were downscaled with CLM, and additionally one realisation of ECHAM5 with WRF. An approach of double nesting was used, first to an approximately 50 km resolution for entire Europe and then to a domain of approximately 7 km covering Germany and its near surroundings. Comparisons of the fine nest simulations are made to earlier high resolution simulations for the region with the RCM REMO for two ECHAM5 realisations. Biases from the GCMs are generally carried over to the RCMs, which can then reduce or worsen the biases. The bias of the coarse nest is carried over to the fine nest but does not change in amplitude, i.e. the fine nest does not add additional mean bias to the simulations. The spatial pattern of the wet bias over central Europe is similar for all CLM simulations, and leads to a stronger bias in the fine nest simulations compared to that of WRF and REMO. The wet bias in the CLM model is found to be due to a too frequent drizzle, but for higher intensities the distributions are well simulated with both CLM and WRF at the 50 and 7 km resolutions. Also the spatial distributions are close to high resolution gridded observations. The REMO model has low biases in the domain averages over Germany and no drizzle problem, but has a shift in the mean precipitation patterns and a strong overestimation of higher intensities. The GCMs perform well in simulating the intensity distribution of precipitation at their own resolution, but the RCMs add value to the distributions when compared to observations at the fine nest resolution.     

High-resolution simulations of West African climate using regional climate model (RegCM3) with different lateral boundary conditions

To downscale climate change scenarios, long-term regional climatologies employing global model forcing are needed for West Africa. As a first step, this work examines present-day integrations (1981–2000) with a regional climate model (RCM) over West Africa nested in both reanalysis data and output from a coupled atmospheric–ocean general circulation model (AOGCM). Precipitation and temperature from both simulations are compared to the Climate Research Unit observations. Their spatial distributions are shown to be realistic. Annual cycles are considerably correlated. Simulations are also evaluated with respect to the driving large-scale fields. RCM offers some improvements compared to the AOGCM driving field. Evaluation of seasonal precipitation biases reveals that RCM dry biases are highest on June–August around mountains. They are associated to cold biases in temperature which, in turn, are connected to wet biases in precipitation outside orographic zones. Biases brought through AOGCM forcing are relatively low. Despite these errors, the simulations produce encouraging results and show the ability of the AOGCM to drive the RCM for future projections.     

Searching for Added Value in Simulating Climate Extremes with a High-Resolution Regional Climate Model over Western Canada

We evaluate the capacity of a regional climate model to represent observed extreme temperature and precipitation events and also examine the impact of increased resolution, in an effort to identify added value in this respect. Two climate simulations of western Canada (WCan) were conducted with the Canadian Regional Climate Model (version 4) at 15 (CRCM15) and 45?km (CRCM45) horizontal resolution driven at the lateral boundaries by data from the European Centre for Medium-range Weather Forecasts (ECMWF) 40-year Reanalysis (ERA-40) for the period 1973–1995. The simulations were evaluated using the spline-interpolated dataset ANUSPLIN, a daily observational gridded surface temperature and precipitation product with a nominal resolution of approximately 10?km. We examine a range of climate extremes, comprising the 10th and 90th percentiles of daily maximum (TX) and minimum (TN) temperatures, the 90th percentile of daily precipitation (PR90), and the 27 core Climate Daily Extremes (CLIMDEX) indices.

Both simulations exhibit cold biases compared with observations over WCan, with the bias exacerbated at higher resolution, suggesting little added value for temperature overall. There are instances, however, of regional improvement in the spatial pattern of temperature extremes at the higher resolution of CRCM15 (e.g., the CLIMDEX index for the annual number of days when TX?>?25°C). The high-resolution simulations also reveal similarly localized features in precipitation (e.g., rain shadows) that are not resolved at the 45?km resolution. With regard to precipitation extremes, although both simulations generally display wet biases, CRCM15 features a reduced bias in PR90 in all seasons except winter. This improvement occurs despite the fact that spatial and interannual variability of PR90 in CRCM15 is significantly overestimated relative to both CRCM45 and ANUSPLIN. We posit that these characteristics are the result of demonstrable differences between corresponding topographical datasets used in the gridded observations and CRCM, the resulting errors propagated to physical variables tied to elevation and the beneficial effect of subsequent spatial averaging. Because topographical input is often discordant between simulations and gridded observations, it is argued that a limited form of spatial averaging may contribute added value beyond that which has already been noted in previous studies with respect to small-scale climate variability.