Logistic判别模型在强降水预报中的应用

利用Logistiv判别模型进行强降水预报,并设计3种方案进行对比分析。方案1直接使用14个影响因子进行判别预报,受因子共线性作用及噪音信号影响,虽然拟合效果较好,但预报效果明显下降。方案2对14个影响因子进行主成分分析,利用前6个主成分建模,虽然拟合效果较方案1降低,但由于消除了因子共线性作用以及噪音信号影响,预报效果较方案1提高。方案3运用Bootstrap抽样技术得到符干样本并建模计算模型参数,打乱了原有时间序列中的波动,仪保留平稳信息,拟合自由度进一步降低,导致拟合效果较方案案2下降,但预报效果却是3种方案中最好且最稳定的。在上述研究基础上,利用欧洲中心数值预报模式的预报场资料,建立基于Logistic判别模型的强降水客观预报系统,并在中央气象台业务运行。2013和2014年连续两年汛期预报检验结果表明,概模型对强降水预报的TS评分高于数值模式本身,具有一定的业务参考价值。     

基于作物模型的河南省旱稻干旱风险评估

采用作物模型与数理统计相结合的方法,利用长期历史气象资料,以作物模型和地理信息系统技术为工具,系统分析了河南地区旱稻生育期水分盈亏情况。以模型模拟的雨养条件下实际蒸散量相对于潜在条件下的蒸散量(即需水量)的亏缺率,即水分亏缺指数,以雨养条件下产量相对于潜在产量的损失率(即灾损指数)作为产量灾损强度评价指标,从受旱程度和产量损失两个角度构建干旱风险评估模型,进行干旱风险评估。结果表明:河南省旱稻生育期集中在6—9月,水分亏缺最多的阶段为出苗—穗分化阶段,水分亏缺指数变化在0.50~0.60,其次是开花—成熟阶段和穗分化—开花阶段,水分亏缺指数变化在0.11~0.43;全生育期水分亏缺指数在0.36~0.50。出苗—穗分化阶段干旱发生的风险最大,其次是开花—成熟阶段,穗分化—开花阶段的最小。河南旱稻生育期干旱风险呈现为由东南向西北逐渐升高的分布,其中三门峡、济源西部一带风险最高,洛阳南部和南阳西北部一带最低,黄河以北大部地区和豫东、豫南地区风险居中。     

地基探测与NCEP GFS模式预报云量在ARM SGP站点对比

使用大气辐射测量实验(Atmospheric Radiation Measurements:ARM)在美国南部大平原站点(Southern Great Plains:SGP)长时间序列(2001 2010年)的地基主动遥感云(Active Remote Sensing of Clouds:ARSCL)和美国国家环境预报中心(National Centers for Environmental Prediction:NCEP)全球预报系统(Global Forecast System:GFS)模式预报资料,对比分析了两者云量在不同时间尺度内(年际、月份和季节)的差异。结果表明,GFS模式预报总云量为83.8%,略高于地基观测结果(78.1%);两者总云量差异在秋季最大(8.8%),春季最小(2.2%)。在低垂直高度分辨率(≥3 km)时,地基探测低云、中云和高云的云量分别为46.1%、43.5%和61.2%;模式预报三类云的云量均要高于地基探测的云量,差异分别为9.6%、17.2%和9.1%。但是,在高垂直分辨率(250 m)时,地基探测云量在大多数高度层上要高于模式预报结果。这应该是两种资料廓线中有云出现的高度层数目存在差异引起的。地基观测和GFS模式预报同时表明,SGP站点上空云量垂直廓线呈现双峰结构,在边界层附近(1 km)和上对流层区域(8-12 km)云量较大,2-3 km高度范围内云量较小。在春夏秋冬四个季节内,两种资料在低层边界层附近的最大云量偏差分别为9.5%、8.8%、7.8%和11.2%。     

Study of high temperature targets identification and temperature retrieval experimental model in SWIR remote sensing based Landsat8

For surface features in short-wave infrared (SWIR, 1.3–3.0 μm) in remote sensing imagery, pixel values depict the total energy including reflection and emission. For surface features at normal temperature in SWIR band, emission energy can be ignored. While for surface features at high temperature in SWIR band, emission energy is equal to or even higher than the reflection energy. So remote sensing imagery of SWIR band can be used to separate emission and reflection energy as well as to realize temperature retrieval of high temperature targets. In this study, the seventh band (SWIR band) of Landsat8 OLI remote sensing imagery is used to perform the theoretical model research for temperature retrieval of high temperature targets. In the meantime, it is also used with the corresponding observation experiment of synchronization satellite to check the theoretical model. The result shows that the radiant flux density for mixed pixels with high temperature targets is higher than adjacent pixels without high temperature targets. Thus, the high temperature pixels can be identified in SWIR band. The retrieval results of temperature and fractional area for high temperature targets are consistent with reality. In the study, the result illustrates that it is effective to identify high temperature targets in remote sensing imagery of SWIR band and the model is appropriate for temperature retrieval use.     

Classification of forest land attributes using multi-source remotely sensed data

The aim of the study was to (1) examine the classification of forest land using airborne laser scanning (ALS) data, satellite images and sample plots of the Finnish National Forest Inventory (NFI) as training data and to (2) identify best performing metrics for classifying forest land attributes. Six different schemes of forest land classification were studied: land use/land cover (LU/LC) classification using both national classes and FAO (Food and Agricultural Organization of the United Nations) classes, main type, site type, peat land type and drainage status. Special interest was to test different ALS-based surface metrics in classification of forest land attributes. Field data consisted of 828 NFI plots collected in 2008–2012 in southern Finland and remotely sensed data was from summer 2010. Multinomial logistic regression was used as the classification method. Classification of LU/LC classes were highly accurate (kappa-values 0.90 and 0.91) but also the classification of site type, peat land type and drainage status succeeded moderately well (kappa-values 0.51, 0.69 and 0.52). ALS-based surface metrics were found to be the most important predictor variables in classification of LU/LC class, main type and drainage status. In best classification models of forest site types both spectral metrics from satellite data and point cloud metrics from ALS were used. In turn, in the classification of peat land types ALS point cloud metrics played the most important role. Results indicated that the prediction of site type and forest land category could be incorporated into stand level forest management inventory system in Finland.     

Estimating winter wheat biomass by assimilating leaf area index derived from fusion of Landsat-8 and MODIS data

A sufficient number of satellite acquisitions in a growing season are essential for deriving agronomic indicators, such as green leaf area index (GLAI), to be assimilated into crop models for crop productivity estimation. However, for most high resolution orbital optical satellites, it is often difficult to obtain images frequently due to their long revisit cycles and unfavorable weather conditions. Data fusion algorithms, such as the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) and the Enhanced STARFM (ESTARFM), have been developed to generate synthetic data with high spatial and temporal resolution to address this issue. In this study, we evaluated the approach of assimilating GLAI into the Simple Algorithm for Yield Estimation model (SAFY) for winter wheat biomass estimation. GLAI was estimated using the two-band Enhanced Vegetation Index (EVI2) derived from data acquired by the Operational Land Imager (OLI) onboard the Landsat-8 and a fusion dataset generated by blending the Moderate-Resolution Imaging Spectroradiometer (MODIS) data and the OLI data using the STARFM and ESTARFM models. The fusion dataset had the temporal resolution of the MODIS data and the spatial resolution of the OLI data. Key parameters of the SAFY model were optimised through assimilation of the estimated GLAI into the crop model using the Shuffled Complex Evolution-University of Arizona (SCE-UA) algorithm. A good agreement was achieved between the estimated and field measured biomass by assimilating the GLAI derived from the OLI data (GLAI_L) alone (R² = 0.77 and RMSE = 231 g m⁻²). Assimilation of GLAI derived from the fusion dataset (GLAI_F) resulted in a R² of 0.71 and RMSE of 193 g m⁻² while assimilating the combination of GLAI_L and GLAI_F led to further improvements (R² = 0.76 and RMSE = 176 g m⁻²). Our results demonstrated the potential of using the fusion algorithms to improve crop growth monitoring and crop productivity estimation when the number of high resolution remote sensing data acquisitions is limited.     

Re-parameterization of a quasi-analytical algorithm for colored dissolved organic matter dominant inland waters

Quasi-Analytical Algorithms (QAAs) are based on radiative transfer equations and have been used to derive inherent optical properties (IOPs) from the above surface remote sensing reflectance (R_rs) in aquatic systems in which phytoplankton is the dominant optically active constituents (OACs). However, Colored Dissolved Organic Matter (CDOM) and Non Algal Particles (NAP) can also be dominant OACs in water bodies and till now a QAA has not been parametrized for these aquatic systems. In this study, we compared the performance of three widely used QAAs in two CDOM dominated aquatic systems which were unsuccessful in retrieving the spectral shape of IOPS and produced minimum errors of 350% for the total absorption coefficient (a), 39% for colored dissolved matter absorption coefficient (a_CDM) and 7566.33% for phytoplankton absorption coefficient (a_phy). We re-parameterized a QAA for CDOM dominated (hereafter QAA_CDOM) waters which was able to not only achieve the spectral shape of the OACs absorption coefficients but also brought the error magnitude to a reasonable level. The average errors found for the 400–750 nm range were 30.71 and 14.51 for a, 14.89 and 8.95 for a_CDM and 25.90 and 29.76 for a_phy in Funil and Itumbiara Reservoirs, Brazil respectively. Although QAA_CDOM showed significant promise for retrieving IOPs in CDOM dominated waters, results indicated further tuning is needed in the estimation of a(λ) and a_phy(λ). Successful retrieval of the absorption coefficients by QAA_CDOM would be very useful in monitoring the spatio-temporal variability of IOPS in CDOM dominated waters.     

线图形简化与移位算子的协同方法

在地图综合过程中,线的图形简化和移位算子通常是分别执行的,图形简化和移位有时都会产生新的空间冲突。本文试图把这两种算子进行协同,避免在地图综合过程中进行空间冲突的多次探测,提高地图综合数据处理的效率。本文通过把线图形简化转换为线上的点移位,并构建邻近地图目标之间的移位传播路径,在考虑空间上下文关系和地图感受规则的前提下,使移位过程能考虑到线图形简化,并尽量保持有关地图目标的空间特征。最后,以道路和其周边的建筑物群为例,验证了该算法的有效性和可行性。     

部分变量误差模型的整体抗差最小二乘估计

部分变量误差模型(partial EIV model)的加权整体最小二乘(weighted total least-squares,WTLS)估计不具备抵御粗差的能力。鉴于粗差可能同时出现在观测值和系数矩阵中,本文在提出部分变量误差模型WTLS估计的两步迭代解法的基础上,运用抗差M估计的等价权方法,发展了一种整体抗差最小二乘(TRLS)估计方法,并采用一致最大功效统计量确定降权因子。针对WTLS估计两步迭代解法的特点,设计了两个不同的降权方案:第1个方案是在估计系数矩阵元素时,不对观测值降权,仅对系数矩阵降权;第2个方案是在估计系数矩阵元素时,既对系数矩阵降权,同时也对观测值降权。通过对模拟2D仿射变换和线性拟合实例进行计算和分析,结果表明第1方案优于第2方案,并且优于基于残差和验后单位权方差的抗差估计和现有的变量误差模型抗差估计。     

SAR图像海岸线检测的区域距离正则化几何主动轮廓模型

合成孔径雷达(SAR)卫星遥感图像可以极大地提高全国海岸线覆盖频率,然而受到海洋波浪所引起的随机海水表面粗糙度的影响,海岸目标与海水背景边界易混淆不清,因此本文提出了基于区域距离正则化几何主动轮廓模型(RDRGAC),引入距离正则项,解决重复初始化水平集函数为符号距离函数的问题,提高了算法收敛速度。此外,将区域面积项系数与SAR图像等效视数(ENL)建立非线性拟合关系,实现RDRGAC模型根据不同SAR遥感图像的自适应调整,改善海岸线自动提取精度。通过河北省北戴河和大连市金州湾SAR数据海岸线提取对比试验,验证了所提方法的有效性。