Downscaling of precipitation for climate change scenarios: A support vector machine approach

The Climate impact studies in hydrology often rely on climate change information at fine spatial resolution. However, general circulation models (GCMs), which are among the most advanced tools for estimating future climate change scenarios, operate on a coarse scale. Therefore the output from a GCM has to be downscaled to obtain the information relevant to hydrologic studies. In this paper, a support vector machine (SVM) approach is proposed for statistical downscaling of precipitation at monthly time scale. The effectiveness of this approach is illustrated through its application to meteorological sub-divisions (MSDs) in India. First, climate variables affecting spatio-temporal variation of precipitation at each MSD in India are identified. Following this, the data pertaining to the identified climate variables (predictors) at each MSD are classified using cluster analysis to form two groups, representing wet and dry seasons. For each MSD, SVM- based downscaling model (DM) is developed for season(s) with significant rainfall using principal components extracted from the predictors as input and the contemporaneous precipitation observed at the MSD as an output. The proposed DM is shown to be superior to conventional downscaling using multi-layer back-propagation artificial neural networks. Subsequently, the SVM-based DM is applied to future climate predictions from the second generation Coupled Global Climate Model (CGCM2) to obtain future projections of precipitation for the MSDs. The results are then analyzed to assess the impact of climate change on precipitation over India. It is shown that SVMs provide a promising alternative to conventional artificial neural networks for statistical downscaling, and are suitable for conducting climate impact studies.     

Namaqualand's climate: Recent historical changes and future scenarios

A brief outline of some issues concerning global climate change research is presented before discussing local-scale changes in Namaqaland's rainfall. Using a gridded data set derived through interpolation of station records, trends in observed rainfall for the period 1950–1999 are discussed. To assess what changes may occur during the 21st century, the downscaled results of six different General Circulation Model projections are presented. The historical trends show some clear spatial patterns, which depict regions of wetting in the central coastal belt and the north-eastern part of the domain, and extensive drying along the escarpment. Reasonably good agreement is shown by the different downscaled projections. These suggest increased late summer convective precipitation in the north-east, but extensive drying along the coast in early and mid winter consistent with the poleward retreat of rain-bearing mid-latitude cyclones.     

降尺度方法在安徽省月降水量预测中的试用

基于NCEP/NCAR 500 hPa高度场、T63月动力延伸预报500 hPa高度场和安徽省降水资料,依据动力预报产品释用方法中所建立的月降水距平百分率预报方程,从月和旬两种不同时间尺度以及固定资料和选择资料来反演方程系数共4种降尺度方法来预报安徽省20个代表站月降水。1995—2005年11 a的回报检验表明了4种方法都具有较好的预报能力,从旬时间尺度较月尺度来预报月降水具有优势,在汛期和汛期降水偏多年更为明显。     

Study on Precipitation Image Downscaling Based on the Method of Ill-posed Problems Solving

Downscaling of remote sensing precipitation products and the forecasting of circulation model are always the intense interests in hydrology and meteorology. The essence of downscaling is primarily to enhance resolution of observation or simulated rainfall field, and to appropriately increase its details or high frequency characteristics. Precipitation, as the main driving factors of the earth’s hydrologic cycle, not only affects the moisture and heat condition of a certain river basin, but also affects the global water and heat circulation. Based on the properties of rainfall self-similarity structure, the mathematically ill-posed precipitation problem solving method was used in low resolution downscaling precipitation for high resolution reconstruction. When solving the downscaling ill-posed problem, the greedy method of orthogonal matching pursuit was introduced so as to get the best high-resolution estimation in an optimal sense. It is hard to imagine that we might be able to find very similar (in mathematical norms) precipitation patterns over relatively large storm-scales. However, finding similar features over sufficiently small sub-storm scales seems more feasible. Based on the characteristics that small scale organized precipitation features tend to recur across different storm environments, the precipitation of both high and low resolution was obtained by training, which could be used to reconstruct the desired high-resolution precipitation field. Multi-source merged precipitation products were used in this experiment. Given the consideration of incompleteness of merged precipitation dataset, it was firstly interpolated based on the method of Fields of Experts (FoEs), which could solve the problem that common interpolation method could hardly work on the interpolation for dataset where consecutive missing data exists. Secondly, ideal experiments of precipitation products downscaling were carried out, where smooth coupling sampling and resampling operator were adopted respectively. Assessment based on the metrics of fidelity and spatial structural similarity demonstrates that the method used in this paper is feasible.     

Sub-pixel Mapping of Coarse Satellite Remote Sensing Images with Stochastic Simulations from Training Images

Super-resolution or sub-pixel mapping is the process of providing fine scale land cover maps from coarse-scale satellite sensor information. Such a procedure calls for a prior model depicting the spatial structures of the land cover types. When available, an analog of the underlying scene (a training image) may be used for such a model. The single normal equation simulation algorithm (SNESIM) allows extracting the relevant pattern information from the training image and uses that information to downscale the coarse fraction data into a simulated fine scale land cover scene. Two non-exclusive approaches are considered to use training images for super-resolution mapping. The first one downscales the coarse fractions into fine-scale pre-posterior probabilities which is then merged with a probability lifted from the training image. The second approach pre-classifies the fine scale patterns of the training image into a few partition classes based on their coarse fractions. All patterns within a partition class are recorded by a search tree; there is one tree per partition class. At each fine scale pixel along the simulation path, the coarse fraction data is retrieved first and used to select the appropriate search tree. That search tree contains the patterns relevant to that coarse fraction data. To ensure exact reproduction of the coarse fractions, a servo-system keeps track of the number of simulated classes inside each coarse fraction. Being an under-determined stochastic inverse problem, one can generate several super resolution maps and explore the space of uncertainty for the fine scale land cover. The proposed SNESIM sub-pixel resolution mapping algorithms allow to: (i) exactly reproduce the coarse fraction, (ii) inject the structural model carried by the training image, and (iii) condition to any available fine scale ground observations. Two case studies are provided to illustrate the proposed methodology using Landsat TM data from southeast China.     

Downscaling法在贵州冬季气温和降水预测中的应用

基于CGCM模式输出500 hPa位势高度场、NCEP/NCAR再分析500 hPa高度资料、贵州冬季降水和气温历史资料,利用降尺度法,对贵州冬季降水和气温预报的技巧和预测效果进行了预测试验和改进。结果表明,该方法从动力与统计相结合的角度,给出季尺度大气环流与局地降水、气温之间的关系,有明确的动力学背景和天气学意义。20年回算及两年回报试验证明了该关系的合理性;对贵州冬季降水的预测率约70%,而对气温的预测率为65%左右。另外,通过对气温反演方程订正后,其预测率达67%左右;在极端异常年,该方法对降水的预测率变幅不大,而对气温的预测效果影响极大。最后利用该方法对2005年贵州冬季降水和气温趋势进行了展望。     

多模式集合预报及其降尺度技术在东亚夏季降水预测中的应用

利用动力季节模式输出的匹配域投影技术和多模式集合预报技术对多个国家和城市的站点月平均降水进行预报。预报变量是北京1个站、韩国60个站和曼谷地区8个站点的月平均降水,预报因子是从多个业务动力季节预报模式输出的多个大尺度变量。模式回报数据和站点观测降水数据时段是1983—2003年。降尺度预报降水的技巧是在交叉验证的框架下进行的。匹配域投影方法是设定一个可以活动的窗口在全球范围内大尺度场上进行扫描,寻求与目标站点降水最优化的因子和最相关的区域,目标站点的降水变率就是由该匹配域上大尺度环流场信息决定的。最终预报是用多个降尺度模式预报结果的集合预报(DMME)。多个降尺度模式预报结果的集合预报能显著地提高站点降水的预报技巧。北京站,多个降尺度模式预报结果的集合预报的预报和观测降水的相关系数可以提高到0.71;韩国地区,多个降尺度模式预报结果的集合预报平均技巧提高到0.75;泰国,多个降尺度模式预报结果的集合预报技巧是0.61。     

Downscaling biogeochemistry in the Benguela eastern boundary current

Dynamical downscaling is developed to better predict the regional impact of global changes in the framework of scenarios. As an intermediary step towards this objective we used the Regional Ocean Modeling System (ROMS) to downscale a low resolution coupled atmosphere–ocean global circulation model (AOGCM; IPSL-CM4) for simulating the recent-past dynamics and biogeochemistry of the Benguela eastern boundary current. Both physical and biogeochemical improvements are discussed over the present climate scenario (1980–1999) under the light of downscaling.Despite biases introduced through boundary conditions (atmospheric and oceanic), the physical and biogeochemical processes in the Benguela Upwelling System (BUS) have been improved by the ROMS model, relative to the IPSL-CM4 simulation. Nevertheless, using coarse-resolution AOGCM daily atmospheric forcing interpolated on ROMS grids resulted in a shifted SST seasonality in the southern BUS, a deterioration of the northern Benguela region and a very shallow mixed layer depth over the whole regional domain. We then investigated the effect of wind downscaling on ROMS solution. Together with a finer resolution of dynamical processes and of bathymetric features (continental shelf and Walvis Ridge), wind downscaling allowed correction of the seasonality, the mixed layer depth, and provided a better circulation over the domain and substantial modifications of subsurface biogeochemical properties. It has also changed the structure of the lower trophic levels by shifting large offshore areas from autotrophic to heterotrophic regimes with potential important consequences on ecosystem functioning. The regional downscaling also improved the phytoplankton distribution and the southward extension of low oxygen waters in the Northern Benguela. It allowed simulating low oxygen events in the northern BUS and highlighted a potential upscaling effect related to the nitrogen irrigation from the productive BUS towards the tropical/subtropical South Atlantic basin. This study shows that forcing a downscaled ocean model with higher resolution winds than those issued from an AOGCM, results in improved representation of physical and biogeochemical processes.     

The NAFE’06 data set: Towards soil moisture retrieval at intermediate resolution

The National Airborne Field Experiment 2006 (NAFE’06) was conducted during a three week period of November 2006 in the Murrumbidgee River catchment, located in southeastern Australia. One objective of NAFE’06 was to explore the suitability of the area for SMOS (Soil Moisture and Ocean Salinity) calibration/validation and develop downscaling and assimilation techniques for when SMOS does come on line. Airborne L-band brightness temperature was mapped at 1 km resolution 11 times (every 1–3 days) over a 40 by 55 km area in the Yanco region and 3 times over a 40 by 50 km area that includes Kyeamba Creek catchment. Moreover, multi-resolution, multi-angle and multi-spectral airborne data including surface temperature, surface reflectance (green, read and near infrared), lidar data and aerial photos were acquired over selected areas to develop downscaling algorithms and test multi-angle and multi-spectral retrieval approaches. The near-surface soil moisture was measured extensively on the ground in eight sampling areas concurrently with aircraft flights, and the soil moisture profile was continuously monitored at 41 sites. Preliminary analyses indicate that (i) the uncertainty of a single ground measurement was typically less than 5% vol. (ii) the spatial variability of ground measurements at 1 km resolution was up to 10% vol. and (iii) the validation of 1 km resolution L-band data is facilitated by selecting pixels with a spatial soil moisture variability lower than the point-scale uncertainty. The sensitivity of passive microwave and thermal data is also compared at 1 km resolution to illustrate the multi-spectral synergy for soil moisture monitoring at improved accuracy and resolution. The data described in this paper are available at www.nafe.unimelb.edu.au.     

基于GWR模型的陕西秦巴山区TRMM降水数据降尺度研究

应用TRMM降水数据,进行国内典型区域降尺度相关研究,可弥补应用气象站点数据研究带来的局限。以陕西秦巴山区为研究区,基于TRMM降水数据和NDVI数据,应用GWR模型和比例指数,获得GWR年、月降尺度数据并进行检验,最后分析地形对降尺度结果的影响。结果表明：获得的1 km分辨率的GWR降尺度降水数据,具有较强的细节表现能力;降尺度数据与实测降水数据年尺度上相关系数为0.88,月为0.93,相关性较好;与TRMM原始数据对比,降尺度结果降水值略小,整体低估降水;区内秦岭山地GWR降尺度结果精度变化幅度最小,相似地形条件下,海拔越高,GWR降尺度结果表现越好;采用GWR模型进行秦巴山区TRMM降水数据的降尺度研究,具有较强的适用性。