[{"data":1,"prerenderedAt":538},["ShallowReactive",2],{"blog-index":3},[4,12,17,22,27,32,37,42,47,52,57,62,67,72,77,82,87,92,97,103,108,113,118,124,130,136,142,148,154,160,166,172,178,184,190,196,202,208,214,220,226,232,238,244,250,256,262,268,274,280,286,292,298,304,310,316,322,328,334,340,346,352,358,364,370,376,382,388,394,400,406,412,418,424,430,436,442,448,454,460,466,472,478,484,490,496,502,508,514,520,526,532],{"_path":5,"title":6,"description":7,"date":8,"author":9,"category":10,"cover":11},"\u002Fblog\u002Fai-super-resolution-downscaling","AI 气象降尺度：超分辨率方法与传统降尺度对比","解读 AI 气象降尺度的超分辨率路线（SRCNN \u002F GAN \u002F 扩散模型）：讲清它如何把粗网格再分析与预报细化到公里级，与传统动力（WRF）、统计降尺度在物理一致性、计算成本、极端值上的取舍，并给出新能源风光功率预测的取数与偏差订正落地路径。","2026-06-17","南京运梦科技算法团队","weather-data","\u002Fog\u002Fblog\u002Fai-super-resolution-downscaling.png",{"_path":13,"title":14,"description":15,"date":8,"author":9,"category":10,"cover":16},"\u002Fblog\u002Fconvective-downburst-wind-risk","雷暴大风与下击暴流：对风电的冲击、识别与预警","面向风电算法与运营团队：讲清雷暴大风与下击暴流的强对流物理机理、对机组载荷与切出停机的冲击，如何用再分析数据复盘对流环境参数、用雷达外推与德国气象局预报做临近预警，并给出在运梦气象 API 上的可落地工作流与代码示例。","\u002Fog\u002Fblog\u002Fconvective-downburst-wind-risk.png",{"_path":18,"title":19,"description":20,"date":8,"author":9,"category":10,"cover":21},"\u002Fblog\u002Fdistributed-pv-aggregate-forecast","分布式光伏的聚合功率预测：从单站到区域的尺度跨越","分布式光伏海量、无遥测、台账残缺，逐站预测既不现实也无必要。本文讲清聚合预测的空间平滑效应、自下而上与自上而下两条技术路线、代表站点选取与容量缩放、辐照空间相关性建模，并给出运梦气象 API 多点 ERA5 取数实战。","\u002Fog\u002Fblog\u002Fdistributed-pv-aggregate-forecast.png",{"_path":23,"title":24,"description":25,"date":8,"author":9,"category":10,"cover":26},"\u002Fblog\u002Fforecast-economic-value-cost-loss","预报的经济价值：cost-loss 决策模型怎么把准确率换算成钱","用经典 cost-loss 决策模型把天气预报的准确率翻译成真金白银：讲清成本损失比、潜在经济价值 PEV、相对作业特征曲线，以及概率预报的最优决策阈值，给新能源运营把 RMSE 提升换算成收益的方法论。","\u002Fog\u002Fblog\u002Fforecast-economic-value-cost-loss.png",{"_path":28,"title":29,"description":30,"date":8,"author":9,"category":10,"cover":31},"\u002Fblog\u002Fgridded-vs-point-forecast","格点预报 vs 单点预报：取数口径与建模的取舍","格点预报与单点预报不是优劣之争，而是取数口径的取舍。本文讲清网格平均与点真值的代表性误差、最近邻\u002F双线性\u002F降尺度三种取点方式、面积代表性与建模影响，并给出运梦气象 API 单点取数示例，帮新能源团队选对口径。","\u002Fog\u002Fblog\u002Fgridded-vs-point-forecast.png",{"_path":33,"title":34,"description":35,"date":8,"author":9,"category":10,"cover":36},"\u002Fblog\u002Finverter-clipping-pv-forecast","逆变器限电（clipping）对光伏功率预测的影响与建模","高直交比电站正午削顶是预测系统性偏差的隐蔽来源。本文讲清 DC\u002FAC ratio、clipping 的物理成因、它如何让晴空高辐照时段预测偏高，并给出含限幅环节的预测建模方法与运梦气象 API 取数实战，帮新能源团队把午间出力曲线做对。","\u002Fog\u002Fblog\u002Finverter-clipping-pv-forecast.png",{"_path":38,"title":39,"description":40,"date":8,"author":9,"category":10,"cover":41},"\u002Fblog\u002Fmissing-data-imputation-weather","气象数据缺口插补：从线性插值到再分析回填与矩阵补全","气象数据缺口插补怎么做才不会毁掉功率预测？本文从线性插值的适用边界讲到再分析回填与矩阵补全，逐一拆解时间、空间、多变量三类缺口的方法选型与典型陷阱，并给出基于运梦气象 API（ERA5 历史回填）的可落地工程范式。","\u002Fog\u002Fblog\u002Fmissing-data-imputation-weather.png",{"_path":43,"title":44,"description":45,"date":8,"author":9,"category":10,"cover":46},"\u002Fblog\u002Fmulti-model-ensemble-power-forecast","多模式集成预报：模式融合如何提升功率预测的稳健性","从单模式的「单点风险」讲到多模式集成（multi-model ensemble）：为什么把多套数值预报与 AI 模型融合能系统性提升风电光伏功率预测的稳健性，对比简单平均、加权、分位回归与 BMA 等融合策略，并给出在运梦气象 API 多源数据上落地的工程路径。","\u002Fog\u002Fblog\u002Fmulti-model-ensemble-power-forecast.png",{"_path":48,"title":49,"description":50,"date":8,"author":9,"category":10,"cover":51},"\u002Fblog\u002Fnwp-resolution-impact-power-forecast","NWP 模式分辨率对功率预测精度的影响：从全球模式到中尺度降尺度","数值预报（NWP）的水平分辨率怎样一步步决定风电光伏功率预测的精度？本文从全球模式的网格尺度讲到中尺度降尺度，拆解地形、辐射、爬坡事件如何被分辨率约束，并给出运梦气象 API 的取数与回测落地路径，帮新能源团队按场景把分辨率花在刀刃上。","\u002Fog\u002Fblog\u002Fnwp-resolution-impact-power-forecast.png",{"_path":53,"title":54,"description":55,"date":8,"author":9,"category":10,"cover":56},"\u002Fblog\u002Fpower-curve-air-density-correction","风电功率曲线的空气密度订正：被忽视的发电量偏差","厂家功率曲线默认 1.225 kg\u002Fm³ 标准空气密度，但高原、高温、季节变化会让实际密度偏离 5% 甚至 10% 以上，直接错估发电量。本文讲清空气密度对风功率的物理影响、IEC 61400-12 系列的两种订正方法，以及在运梦气象 API 上用气温气压逐时订正功率曲线的实战路径。","\u002Fog\u002Fblog\u002Fpower-curve-air-density-correction.png",{"_path":58,"title":59,"description":60,"date":8,"author":9,"category":10,"cover":61},"\u002Fblog\u002Fpower-forecast-evaluation-metrics","新能源功率预测评价指标体系：准确率、合格率与 RMSE\u002FMAE 怎么算","系统梳理新能源功率预测评价指标：从 RMSE\u002FMAE\u002FMAPE 等统计量到两个细则的均方根准确率与单点合格率口径，讲清归一化分母、样本筛选、Skill Score 与误差归因，并给出 ERA5 回测加德国气象局预报的落地评估方法。","\u002Fog\u002Fblog\u002Fpower-forecast-evaluation-metrics.png",{"_path":63,"title":64,"description":65,"date":8,"author":9,"category":10,"cover":66},"\u002Fblog\u002Fprecipitation-nowcasting-radar","降水临近预报：雷达外推与对户外作业、运维的意义","雷达回波外推、光流与深度学习（DGMR、PySTEPS 思想）是降水临近预报（0–2 小时）的主力，数值预报在该尺度存在 spin-up 短板。本文讲清外推机制、可预报性边界与误差评估，并落到户外作业窗口、电站运维巡检与风光功率超短期订正等新能源价值，结尾给出运梦气象 API 取数示例。","\u002Fog\u002Fblog\u002Fprecipitation-nowcasting-radar.png",{"_path":68,"title":69,"description":70,"date":8,"author":9,"category":10,"cover":71},"\u002Fblog\u002Freanalysis-products-jra-cra-comparison","再分析产品横评：ERA5 \u002F JRA-3Q \u002F 中国区域再分析怎么选","面向新能源资源评估与功率预测团队，横向对比 ERA5、JRA-3Q 与中国区域再分析的同化体系、时空规格、东亚与复杂地形适配性，给出按风光场景与地理位置的选型决策树和交叉验证方法，并附运梦气象 API 多源切换示例。","\u002Fog\u002Fblog\u002Freanalysis-products-jra-cra-comparison.png",{"_path":73,"title":74,"description":75,"date":8,"author":9,"category":10,"cover":76},"\u002Fblog\u002Fweather-data-api-engineering","气象数据 API 接入工程：限流、缓存、重试与批量取数","气象数据 API 接入常被限流、超时、重复取数拖垮。本文系统讲解令牌桶限流、分层缓存与缓存键设计、指数退避加抖动的重试、并发分片批量取数四套工程手段，并结合运梦气象 API 的同步取数接口给出可复用的 Python 骨架，帮新能源团队把数据接入层做稳。","\u002Fog\u002Fblog\u002Fweather-data-api-engineering.png",{"_path":78,"title":79,"description":80,"date":8,"author":9,"category":10,"cover":81},"\u002Fblog\u002Fweather-data-versioning-reproducibility","气象数据的版本管理与可复现性：再分析重处理带来的坑","再分析数据并非一成不变：ERA5T 准实时流与最终流会改写、ERA5.1 重处理修正了 2000–2006 平流层冷偏差、CDS 迁移又换了版本与 DOI。本文讲清这些「同名不同值」如何悄悄破坏功率预测与电量回测的可复现性，并给出版本钉死、内容寻址、数据血缘的工程纪律与运梦气象 API 的对齐口径。","\u002Fog\u002Fblog\u002Fweather-data-versioning-reproducibility.png",{"_path":83,"title":84,"description":85,"date":8,"author":9,"category":10,"cover":86},"\u002Fblog\u002Fwind-measurement-mast-lidar-sodar","测风设备对比：测风塔、激光雷达与声雷达怎么选","面向风资源评估的测风设备选型指南：从测量原理、不确定度、IEC 与 MEASNET 标定，到测风塔、激光雷达（lidar）、声雷达（sodar）各自的边界与适用场景，并给出如何用运梦气象 API 的再分析长序列做长期订正的可运行示例。","\u002Fog\u002Fblog\u002Fwind-measurement-mast-lidar-sodar.png",{"_path":88,"title":89,"description":90,"date":8,"author":9,"category":10,"cover":91},"\u002Fblog\u002Fwind-turbine-load-fatigue-meteorology","风机载荷与疲劳的气象驱动：湍流、切变与极端阵风","面向风电工程与载荷团队：拆解湍流强度、风切变\u002F风向切变与极端阵风如何驱动风机疲劳与极端载荷，讲清 IEC 61400-1 的 NTM\u002FETM 与 EOG 设计工况，并给出在运梦气象 API 上用 ERA5 与德国气象局数据做载荷气象画像的可运行示例。","\u002Fog\u002Fblog\u002Fwind-turbine-load-fatigue-meteorology.png",{"_path":93,"title":94,"description":95,"date":8,"author":9,"category":10,"cover":96},"\u002Fblog\u002Fwind-veer-large-rotor-impact","风向切变（veer）对大叶轮风机出力与载荷的影响","风向切变（veer）指风向随高度顺转，在大叶轮风机上让叶尖与叶根\"吃\"的是不同方向的风。本文讲清 veer 的成因、对扫风面有效来流与功率曲线的削弱、对叶片疲劳载荷的放大，以及如何用运梦气象 API 的多高度风分量量化 veer，并把它纳入资源评估与功率预测。","\u002Fog\u002Fblog\u002Fwind-veer-large-rotor-impact.png",{"_path":98,"title":99,"description":100,"date":101,"author":9,"category":10,"cover":102},"\u002Fblog\u002Fdata-assimilation-explained","数据同化入门：从 OI、3D\u002F4D-Var 到 EnKF","用一篇讲清数据同化：最优插值 OI、3D-Var、4D-Var 与 EnKF 如何把观测和模式背景场最优融合，重建出 ERA5 这类再分析。面向新能源从业者，理清再分析数据底座怎么算出来、为什么能当训练金标准。","2026-06-16","\u002Fog\u002Fblog\u002Fdata-assimilation-explained.png",{"_path":104,"title":105,"description":106,"date":101,"author":9,"category":10,"cover":107},"\u002Fblog\u002Fensemble-postprocessing-emos-quantile","集合预报后处理：EMOS 与分位数回归生成概率功率预测","系统讲清集合预报后处理的两条主线：EMOS（非齐次高斯回归）做参数化校准、分位数回归做非参数概率建模。面向风光功率预测，剖析欠离散订正、CRPS 优化、单调性约束与落地工程，帮新能源团队把原始集合升级为可信的概率出力曲线。","\u002Fog\u002Fblog\u002Fensemble-postprocessing-emos-quantile.png",{"_path":109,"title":110,"description":111,"date":101,"author":9,"category":10,"cover":112},"\u002Fblog\u002Foffshore-wind-metocean-data","海上风电的 metocean 数据：风、浪、流的耦合、获取与工程对齐","面向海上风电的 metocean 数据指南：讲清风、浪、流为何必须耦合看待，设计波高与极值如何用 GEV\u002FGumbel 估算，再分析、后报、浮标与卫星各自的边界，以及如何用运梦气象 API 取轮毂高度风场作为耦合分析的大气驱动，配可运行示例。","\u002Fog\u002Fblog\u002Foffshore-wind-metocean-data.png",{"_path":114,"title":115,"description":116,"date":101,"author":9,"category":10,"cover":117},"\u002Fblog\u002Fweather-data-quality-control","气象数据质量控制：缺测、异常值与物理一致性检验","做新能源功率预测前，气象数据先要过质量控制这一关。本文系统拆解缺测识别、物理与气候学异常值检验、要素间物理一致性核查与时空连续性检查，给出标缺、插补与留痕的可审计流程，帮你在数据进模型前挡住坏点，避免功率预测被脏数据悄悄带偏。","\u002Fog\u002Fblog\u002Fweather-data-quality-control.png",{"_path":119,"title":120,"description":121,"date":122,"author":9,"category":10,"cover":123},"\u002Fblog\u002Fresource-interannual-variability","新能源资源年际变率与长期代表性评估","解读 Pryor 等 2018 年 Wind Energy Science 论文：风资源年际变率到底有多大，行业惯用的 6% 标准差是否高估，配合 Lee 等 27 种变率指标对比与长期代表性方法，并给出运梦气象 API 上 ERA5 多年回测的可运行示例。","2026-06-05","\u002Fog\u002Fblog\u002Fresource-interannual-variability.jpg",{"_path":125,"title":126,"description":127,"date":128,"author":9,"category":10,"cover":129},"\u002Fblog\u002Fcf-conventions-toolchain","再分析数据的 CF 规范与 Python 工具链","一篇讲清 CF（Climate and Forecast）元数据规范如何用 standard_name、canonical units、坐标轴与时间编码给再分析数据立\"语义合同\"，并结合 xarray、cfgrib、netCDF4 工具链，帮风电光伏团队把 NetCDF \u002F GRIB 字段稳稳落到功率预测管线。","2026-05-13","\u002Fog\u002Fblog\u002Fcf-conventions-toolchain1.jpg",{"_path":131,"title":132,"description":133,"date":134,"author":9,"category":10,"cover":135},"\u002Fblog\u002Fwind-curtailment-forecast-grid","风电弃风与气象预报：国际经验综述与中国场景","解读 Bird 等在 Renewable and Sustainable Energy Reviews 2016 发表的弃风国际综述：中美德丹等主要市场的弃风率成因，以及如何用德国气象局预报提升日前出力计划精度、结合 ERA5 历史数据优化储能调度降低弃风损失。","2026-04-21","\u002Fog\u002Fblog\u002Fwind-curtailment-forecast-grid.png",{"_path":137,"title":138,"description":139,"date":140,"author":9,"category":10,"cover":141},"\u002Fblog\u002Fenergy-forecasting-competition-gefcom","全球能源预测竞赛 GEFCom2012 — Hong 2014 与预测基准","解读 Hong、Pinson、Fan 在 International Journal of Forecasting 2014 发表的 GEFCom2012 综述：负荷预测与风电预测两大赛道、分层负荷与温度驱动、点预测评分（RMSE），以及竞赛对新能源功率预测方法与基准数据集的推动。","2026-03-29","\u002Fog\u002Fblog\u002Fenergy-forecasting-competition-gefcom.png",{"_path":143,"title":144,"description":145,"date":146,"author":9,"category":10,"cover":147},"\u002Fblog\u002Fwind-solar-resource-assessment-p50-p90","风光资源评估实战：ERA5 多年逐时数据做代表年与 P50\u002FP90 发电量估计","用 ERA5 多年逐时数据做风光资源评估：长序列按年切片下载、风速廓线从 10m 外推到 100m 与轮毂高度、辐照换算发电量，以及代表年、年际变率与 P50\u002FP90 的完整工程流程与可运行的运梦气象 API 下载示例。","2026-03-18","\u002Fog\u002Fblog\u002Fwind-solar-resource-assessment-p50-p90.png",{"_path":149,"title":150,"description":151,"date":152,"author":9,"category":10,"cover":153},"\u002Fblog\u002Foffshore-wind-reanalysis","海上风资源的再分析评估方法与实战：从 ERA5 到 NEWA","解读 NEWA 新欧洲风图谱两篇 GMD 论文与 ERA5 海上适用性：ERA5 在高桅杆评测下年均风速偏低约 1.5 m\u002Fs、强风段近岸系统性低估，WRF 中尺度降尺度把偏差压到接近零，并给出运梦气象 API 的可运行下载示例。","2026-03-07","\u002Fog\u002Fblog\u002Foffshore-wind-reanalysis.png",{"_path":155,"title":156,"description":157,"date":158,"author":9,"category":10,"cover":159},"\u002Fblog\u002Fera5-drought-indices","ERA5-Drought 全球干旱指数数据集解读 — ECMWF 官方综述","解读 ECMWF 团队发表于 Nature 旗下 Scientific Data（2025）的 ERA5-Drought：基于 ERA5 构建的 1940 年至今、0.25° 全球干旱指数数据集，含 SPI、SPEI 与确定性 \u002F 集合概率版本，并说明其对风电光伏资源评估、功率预测与历史气象数据使用的意义。","2026-02-12","\u002Fog\u002Fblog\u002Fera5-drought-indices.jpg",{"_path":161,"title":162,"description":163,"date":164,"author":9,"category":10,"cover":165},"\u002Fblog\u002Fera5-download-guide","ERA5 历史气象数据下载完全指南 — 运梦气象 API 与 cdsapi 官方通道实战对比","从 ERA5 数据集结构、变量字段、时空分辨率讲起，给出运梦气象 API 与 ECMWF CDS cdsapi 两条下载通道的 Python 代码示例与排队\u002F速率\u002F合规对比，帮助新能源算法团队选择更高效的方案。","2026-02-10","\u002Fog\u002Fblog\u002Fera5-download-guide.png",{"_path":167,"title":168,"description":169,"date":170,"author":9,"category":10,"cover":171},"\u002Fblog\u002Fboundary-layer-pblh-wind-energy","大气边界层高度（PBLH）与风能：ERA5 字段到工程应用","解读 Seibert 等在 Atmospheric Environment 2000 发表的混合层高度综述：探空、声雷达与风廓线雷达等业务方法的比较与不确定性，以及 ERA5 的 PBLH 字段如何指导风资源评估、尾流建模参数化与风机轮毂高度选择。","2026-01-21","\u002Fog\u002Fblog\u002Fboundary-layer-pblh-wind-energy.png",{"_path":173,"title":174,"description":175,"date":176,"author":9,"category":10,"cover":177},"\u002Fblog\u002Fweather-data-engineering-pitfalls","气象数据工程化最容易踩的五个坑：命名、单位、时区、缺测与数组对齐","做新能源功率预测时，气象数据最容易在变量命名、单位换算、时区对齐、缺测处理和 JSON 数组对齐这五处悄悄出错。本文逐一拆解 CF 名 vs GRIB 名、K↔℃、J·m⁻²↔W·m⁻²、timezone 必填等坑，并给出运梦气象 API 可运行示例。","2026-01-20","\u002Fog\u002Fblog\u002Fweather-data-engineering-pitfalls.png",{"_path":179,"title":180,"description":181,"author":9,"cover":182,"date":183,"category":10},"\u002Fblog\u002Fensemble-reanalysis-uncertainty","集合再分析与不确定性：把误差量化进资源评估","解读 Hersbach 等 QJRMS 2020 ERA5 论文中的不确定性集合（EDA）：10 成员、约 62km、3 小时一档、普遍欠离散。讲清如何把随机误差量化进风电与光伏的资源评估与 P50\u002FP90 决策。","\u002Fog\u002Fblog\u002Fensemble-reanalysis-uncertainty.png","2025-12-29",{"_path":185,"title":186,"description":187,"date":188,"author":9,"category":10,"cover":189},"\u002Fblog\u002Fwind-turbine-power-curve-ml","风机功率曲线机器学习建模与性能监控","解读 Kusiak 与 Verma 在 IEEE Transactions on Sustainable Energy 2013 的论文：基于 SCADA 数据以风速为输入建立功率、转速、桨距角三条参考曲线，用 k-means 聚类与马氏距离清洗数据，并以 Hotelling T² 控制图监控风机性能退化。","2025-12-07","\u002Fog\u002Fblog\u002Fwind-turbine-power-curve-ml.png",{"_path":191,"title":192,"description":193,"date":194,"author":9,"category":10,"cover":195},"\u002Fblog\u002Fcmip6-era5-future-scenarios","CMIP6 × ERA5：从历史校准到未来气候情景","解读 Eyring 等 GMD 2016 CMIP6 实验设计：DECK 四张入场券、1850–2014 历史模拟、SSP 情景与约 50 个建模组的协作框架，并讲清如何用 ERA5 做历史校准、把气候情景接到风电光伏长期评估。","2025-11-14","\u002Fog\u002Fblog\u002Fcmip6-era5-future-scenarios.png",{"_path":197,"title":198,"description":199,"date":200,"author":9,"category":10,"cover":201},"\u002Fblog\u002Frenewable-complementarity-wind-solar","风光互补性 — Jurasz 2020 综述与时空匹配评估","解读 Jurasz 等在 Solar Energy 2020 发表的可再生能源互补性综述：风光在时间与空间上的此消彼长如何降低出力波动、常用相关性与互补性度量，以及如何用运梦气象 API 的风光长序列量化站址互补性、优化容量配比。","2025-10-23","\u002Fog\u002Fblog\u002Frenewable-complementarity-wind-solar.png",{"_path":203,"title":204,"description":205,"date":206,"author":9,"category":10,"cover":207},"\u002Fblog\u002Fecmwf-ensemble-eps-renewable","ECMWF 集合预报系统（ENS）在可再生能源中的应用","解读 Buizza 等在 Monthly Weather Review 2005 发表的全球集合预报系统比较论文：ECMWF ENS、加拿大 MSC 与美国 NCEP 的初始扰动方案、集合配置与预报技巧对比，及集合预报在风光功率不确定性量化中的工程价值。","2025-09-30","\u002Fog\u002Fblog\u002Fecmwf-ensemble-eps-renewable.png",{"_path":209,"title":210,"description":211,"date":212,"author":9,"category":10,"cover":213},"\u002Fblog\u002Fera5-station-bias-validation","ERA5 与站点观测的偏差特征：用前必看的验证","汇总 Olauson 2018 风、Urraca 2018 辐照、Zhao 与 He 2022 气温三项核心验证，讲清 ERA5 相对站点观测的系统偏差量级、复杂地形与近海退化规律，以及在风电光伏功率预测中先订正再用的工程纪律。","2025-09-08","\u002Fog\u002Fblog\u002Fera5-station-bias-validation.png",{"_path":215,"title":216,"description":217,"date":218,"author":9,"category":10,"cover":219},"\u002Fblog\u002Fwind-power-probabilistic-forecast-pinson","风电功率概率预测：Pinson 2013 运营挑战综述","解读 Pinson 在 Statistical Science 2013 发表的风电功率预测综述：从点预测到概率预测（分位数回归、预测区间、密度预测）的演进，评分规则 Pinball Loss 与 CRPS，以及集成预报在新能源调度决策中的框架。","2025-08-16","\u002Fog\u002Fblog\u002Fwind-power-probabilistic-forecast-pinson.png",{"_path":221,"title":222,"description":223,"date":224,"author":9,"category":10,"cover":225},"\u002Fblog\u002Fmcp-mast-reanalysis-fusion","测风塔与再分析数据融合：MCP 长期订正实战","测风塔短期实测如何外推到代表性长期风况？本文讲解 MCP（测量-相关-预测）方法：以 ERA5 长序列再分析为参考源，把场站几个月的测风订正为多年代表水平，覆盖线性回归、方差比与矩阵法，并附运梦气象 API 取数示例。","2025-07-25","\u002Fog\u002Fblog\u002Fmcp-mast-reanalysis-fusion.png",{"_path":227,"title":228,"description":229,"date":230,"author":9,"category":10,"cover":231},"\u002Fblog\u002Fsolar-dimming-brightening-pv-yield","太阳辐射的变暗与变亮：Wild 2005 与光伏长期产量预估","解读 Wild 等在 Science 2005 发表的全球太阳辐射变暗与变亮论文：气溶胶与云变化驱动地表辐射十年趋势，以及它对光伏典型气象年选取、ERA5 历史辐射序列代表性和长期 AEP 预估系统偏差的工程意义。","2025-07-03","\u002Fog\u002Fblog\u002Fsolar-dimming-brightening-pv-yield.png",{"_path":233,"title":234,"description":235,"date":236,"author":9,"category":10,"cover":237},"\u002Fblog\u002Fload-forecast-temperature-sensitivity","电力负荷预测里的气象输入与温度敏感度","解读 Hong & Fan 概率负荷预测综述：温度与负荷的非线性关系、HDD\u002FCDD 度日定义、recency 滞后温度效应，并给出在运梦气象 API 用 era5 回测、ger 预报取气温的可运行示例。","2025-06-10","\u002Fog\u002Fblog\u002Fload-forecast-temperature-sensitivity.png",{"_path":239,"title":240,"description":241,"date":242,"author":9,"category":10,"cover":243},"\u002Fblog\u002Fwind-power-forecasting-review","风电功率预测方法全景 — Foley 2012 综述精读","解读 Foley 等在 Renewable Energy 2012 发表的风电功率预测综述：物理方法、统计方法与混合方法的分野，NWP 作为预测核心驱动、持续法基准与时间尺度划分，并结合运梦气象 API 说明工程落地路径。","2025-05-19","\u002Fog\u002Fblog\u002Fwind-power-forecasting-review.png",{"_path":245,"title":246,"description":247,"date":248,"author":9,"category":10,"cover":249},"\u002Fblog\u002Firradiance-decomposition-erbs-model","GHI 分解为 DNI 与 DHI：Erbs 1982 扩散比模型与光伏应用","解读 Erbs、Klein、Duffie 在 Solar Energy 1982 发表的辐射分解经典论文：如何从全球水平辐照 GHI 用晴空指数估算扩散分量 DHI 与直接分量 DNI，以及对 CSP、双面组件与固定倾斜光伏发电量建模的工程意义。","2025-04-26","\u002Fog\u002Fblog\u002Firradiance-decomposition-erbs-model.png",{"_path":251,"title":252,"description":253,"date":254,"author":9,"category":10,"cover":255},"\u002Fblog\u002Fextreme-weather-renewable-energy","极端天气下的风光出力冲击与运行预案：ERA5 复盘 + 德国气象局预警实战","面向新能源算法与运营团队：用运梦气象 API 的 ERA5 历史数据复盘寒潮、台风、连续阴雨、无风小风四类极端事件对风光出力的冲击，用德国气象局预报提前预警，量化发电量与电网偏差并给出可落地的运行预案与 Python 示例。","2025-04-04","\u002Fog\u002Fblog\u002Fextreme-weather-renewable-energy.png",{"_path":257,"title":258,"description":259,"date":260,"author":9,"category":10,"cover":261},"\u002Fblog\u002Fsolar-nowcasting-cloud-motion","光伏超短期预测：0–4 小时云团外推临近预报","云团外推（云移动矢量）是光伏 0–4 小时临近预报的主力方法。本文核实 Lorenz、Hammer\u002FKühnert、Logothetis、Straub 等关键文献，解读卫星与全天空成像云外推机制、晴空指数与误差量级，并给出运梦气象 API 上手示例。","2025-03-12","\u002Fog\u002Fblog\u002Fsolar-nowcasting-cloud-motion.png",{"_path":263,"title":264,"description":265,"date":266,"author":9,"category":10,"cover":267},"\u002Fblog\u002Foffshore-wind-wake-losses","海上风电场尾流损失：Barthelmie 2009 测量与建模","解读 Barthelmie 等在 Wind Energy 2009 发表的海上风电场尾流论文：Horns Rev 与 Nysted 大型海上风电场实测显示尾流损失 10–20%，并评估 Jensen\u002FPark 等工程尾流模型的精度与大气稳定度对尾流延伸的影响。","2025-02-18","\u002Fog\u002Fblog\u002Foffshore-wind-wake-losses.png",{"_path":269,"title":270,"description":271,"date":272,"author":9,"category":10,"cover":273},"\u002Fblog\u002Fpv-temperature-irradiance-model","光伏组件温度损失与辐照-功率建模实战","从辐照到交流功率的物理链路里，组件温度是最容易被低估的一环。本文核实 Faiman、Sandia SAPM 与 pvlib 三篇权威来源，讲清温度系数、NOCT 与组件温度估算，并给出运梦气象 API 取数实战。","2025-01-26","\u002Fog\u002Fblog\u002Fpv-temperature-irradiance-model.png",{"_path":275,"title":276,"description":277,"date":278,"author":9,"category":10,"cover":279},"\u002Fblog\u002Fwind-shear-weibull-hub-height","风切变与威布尔分布：轮毂高度风速外推实战","从 IEC 61400-12-1 标准、幂律\u002F对数律风廓线到 Justus 与 Weibull 的经典模型，讲清楚怎样把 10m\u002F100m 风速外推到真实轮毂高度，并给出在运梦气象 API 上用 ERA5 与德国气象局数据落地外推的可运行示例。","2025-01-04","\u002Fog\u002Fblog\u002Fwind-shear-weibull-hub-height.png",{"_path":281,"title":282,"description":283,"date":284,"author":9,"category":10,"cover":285},"\u002Fblog\u002Ftmy-typical-meteorological-year","典型气象年（TMY）方法 — 从多年序列到一年代表","解读 NREL 的 TMY3 用户手册与 Finkelstein-Schafer 选月方法：典型气象年如何从多年观测中拼出 12 个最具代表性的月份、为何代表典型而非极端，以及如何用运梦气象 API 的 ERA5 长序列自行构建站址 TMY。","2024-12-12","\u002Fog\u002Fblog\u002Ftmy-typical-meteorological-year.png",{"_path":287,"title":288,"description":289,"date":290,"author":9,"category":10,"cover":291},"\u002Fblog\u002Fpv-degradation-long-term-yield","光伏组件衰减率与长期发电量 — Jordan & Kurtz 2013 分析综述","解读 Jordan 与 Kurtz 在 Progress in Photovoltaics 2013 发表的光伏衰减率综述：同一约 2000 组数据中位约 0.5%\u002F年、均值约 0.8%\u002F年，分布右偏，温度辐照湿热驱动衰减，以及如何用 ERA5 历史序列订正长期发电量预估的系统性偏差。","2024-11-20","\u002Fog\u002Fblog\u002Fpv-degradation-long-term-yield.png",{"_path":293,"title":294,"description":295,"date":296,"author":9,"category":10,"cover":297},"\u002Fblog\u002Fera5-land-evapotranspiration","ERA5-Land 估算作物参考蒸散精度评估：西西里 39 站验证论文解读","解读 Ippolito 等 2024 年发表于《Agricultural Water Management》的论文：用西西里 39 个地面站 2006–2015 年实测数据验证 ERA5-Land 参考蒸散 ETo，日尺度 RMSE 约 0.4–1.3 mm\u002Fd，并谈再分析数据在风光资源评估中的借鉴价值。","2024-10-28","\u002Fog\u002Fblog\u002Fera5-land-evapotranspiration.png",{"_path":299,"title":300,"description":301,"date":302,"author":9,"category":10,"cover":303},"\u002Fblog\u002Fquantile-mapping-bias-correction","分位数映射偏差订正：再分析数据落地必修课","解读 Cannon、Sobie、Murdock 在 Journal of Climate 2015 的分位数映射偏差订正论文，并引 Maraun 2016 综述：讲清 QM、DQM、QDM 的机制与边界，映射到风电光伏功率预测如何用再分析做偏差订正。","2024-10-06","\u002Fog\u002Fblog\u002Fquantile-mapping-bias-correction.png",{"_path":305,"title":306,"description":307,"date":308,"author":9,"category":10,"cover":309},"\u002Fblog\u002Fwrf-dynamical-downscaling","WRF 动力降尺度：复杂地形风电的气象增强","解读 WRF 模式权威文献（Powers 等 2017 BAMS 与 Skamarock 等 2021 NCAR 技术报告），讲清动力降尺度如何把粗网格再分析细化到复杂地形百米级风场、它的灰区局限，以及如何映射到运梦气象 API 的风电选址与功率预测上手路径。","2024-09-14","\u002Fog\u002Fblog\u002Fwrf-dynamical-downscaling.png",{"_path":311,"title":312,"description":313,"date":314,"author":9,"category":10,"cover":315},"\u002Fblog\u002Fmerra2-reanalysis-overview","MERRA-2 NASA 再分析全解读 — 运梦 `nasa` 数据源的原始论文","解读 Gelaro 等在 Journal of Climate 2017 发表的 MERRA-2 论文：NASA 第二代全球再分析的同化体系、时空规格与 ERA5 互补关系，以及在风电光伏资源评估中如何选用 nasa 数据源。","2024-08-22","\u002Fog\u002Fblog\u002Fmerra2-reanalysis-overview.png",{"_path":317,"title":318,"description":319,"date":320,"author":9,"category":10,"cover":321},"\u002Fblog\u002Fai-weather-models-power-forecast","AI 气象大模型如何落地新能源功率预测","梳理 GraphCast、Pangu-Weather、FourCastNet、FengWu、AIFS 五大 AI 气象模型的核实事实与关键数字，讲清它们如何作为气象输入接入风电光伏功率预测，并给出运梦气象 API 的上手示例与边界限定。","2024-07-31","\u002Fog\u002Fblog\u002Fai-weather-models-power-forecast.png",{"_path":323,"title":324,"description":325,"date":326,"author":9,"category":10,"cover":327},"\u002Fblog\u002Feuropean-wind-atlas-wasp","欧洲风图集与 WAsP 方法 — 风资源评估的奠基之作","解读 Troen 与 Petersen 1989 年的《欧洲风图集》：奠定了 WAsP 风资源评估方法学——地转拖曳定律、粗糙度变化、地形与障碍物订正、广义风气候，并说明如何用运梦气象 API 的 ERA5 长序列做现代风资源评估的长期参考。","2024-07-08","\u002Fog\u002Fblog\u002Feuropean-wind-atlas-wasp.png",{"_path":329,"title":330,"description":331,"date":332,"author":9,"category":10,"cover":333},"\u002Fblog\u002Ffengwu-beyond-10-days","FengWu 风乌解读：把中期预报技巧推过 10 天","解读上海人工智能实验室 FengWu（风乌，arXiv 2023）：多模态多任务架构、0.25° 网格、37 层、39 年 ERA5 训练，首次把 z500 有效预报技巧推到 10.75 天，并映射到运梦气象 API 的 ERA5 与德国气象局双数据源落地。","2024-06-16","\u002Fog\u002Fblog\u002Ffengwu-beyond-10-days.png",{"_path":335,"title":336,"description":337,"date":338,"author":9,"category":10,"cover":339},"\u002Fblog\u002Fwind-short-term-forecast","风电短期功率预测链路 — 从 100m 风速到出力曲线的工程化拆解","风电短期 (0-72h) 功率预测的完整工程链路：NWP 选型、100m 风速到塔筒高度的高度修正、功率曲线建模、复杂地形动力降尺度、实测同化与机器学习偏差订正。给出常见踩坑与 ROI 排序的优化路径。","2024-05-24","\u002Fog\u002Fblog\u002Fwind-short-term-forecast.png",{"_path":341,"title":342,"description":343,"date":344,"author":9,"category":10,"cover":345},"\u002Fblog\u002Fecmwf-aifs-forecast","ECMWF AIFS 解读：业务级数据驱动预报系统","AIFS 是 ECMWF 的数据驱动预报系统（arXiv:2406.01465），确定性版 2025 年 2 月、集合版 7 月相继业务化。本文解读其 GNN+Transformer 架构、ERA5 训练与多项核实数字，及对风电光伏功率预测的意义。","2024-05-02","\u002Fog\u002Fblog\u002Fecmwf-aifs-forecast.png",{"_path":347,"title":348,"description":349,"date":350,"author":9,"category":10,"cover":351},"\u002Fblog\u002Ffuxi-cascade-15day","FuXi 解读：级联机器学习把全球预报推到 15 天","解读复旦团队 FuXi（npj Climate and Atmospheric Science 2023）：三段级联（短\u002F中\u002F长）U-Transformer、0.25° 网格、39 年 ERA5 训练，15 天预报技巧比肩 ECMWF 集合均值，并映射到运梦气象 API 的 ERA5 与德国气象局双数据源落地。","2024-04-09","\u002Fog\u002Fblog\u002Ffuxi-cascade-15day.png",{"_path":353,"title":354,"description":355,"date":356,"author":9,"category":10,"cover":357},"\u002Fblog\u002Faurora-atmosphere-foundation-model","Aurora 解读：大气基础模型与新能源气象应用","Aurora（Nature 2025，Microsoft）是 13 亿参数的地球系统基础模型，预训练超百万小时数据，覆盖天气、空气质量、海浪、台风路径多任务。本文解读其预训练-微调范式与对风电光伏功率预测的意义。","2024-03-18","\u002Fog\u002Fblog\u002Faurora-atmosphere-foundation-model.png",{"_path":359,"title":360,"description":361,"date":362,"author":9,"category":10,"cover":363},"\u002Fblog\u002Fsolar-forecast-accuracy","光伏功率预测准确率怎么算 — RMSE\u002FMAE\u002FMAPE 与工程样本","从光伏功率预测的物理链路讲起，定义 RMSE\u002FMAE\u002FMAPE\u002FSkillScore 等准确率指标，给出实际项目工程中短期（24h）\u002F 超短期（4h）\u002F 中期（72h）三种预测时长的实测基线分布，附 Python 评估代码与自绘的准确率分布图。","2024-02-24","\u002Fog\u002Fblog\u002Fsolar-forecast-accuracy.jpg",{"_path":365,"title":366,"description":367,"date":368,"author":9,"category":10,"cover":369},"\u002Fblog\u002Fpv-soiling-dust-losses","光伏积灰（Soiling）损失 — Sarver 2013 综述与气象关联","解读 Sarver、Al-Qaraghuli、Kazmerski 在 RSER 2013 发表的灰尘影响综述：积灰如何削减光伏出力、干旱与少雨地区损失更重、降水的自清洁作用，以及如何用运梦气象 API 的降水与气象序列量化积灰风险与清洗排期。","2024-02-02","\u002Fog\u002Fblog\u002Fpv-soiling-dust-losses.png",{"_path":371,"title":372,"description":373,"date":374,"author":9,"category":10,"cover":375},"\u002Fblog\u002Fneuralgcm-hybrid-model","NeuralGCM 解读：物理与机器学习混合的气候模式","NeuralGCM（Nature 2024，Google Research）把可微分动力核与神经网络参数化结合，1–15 天预报对标 ECMWF-ENS，气候积分可稳定数十年。本文解读其混合架构、评测口径，及对新能源功率预测与资源评估的意义。","2024-01-10","\u002Fog\u002Fblog\u002Fneuralgcm-hybrid-model.png",{"_path":377,"title":378,"description":379,"date":380,"author":9,"category":10,"cover":381},"\u002Fblog\u002Ffourcastnet-fourier-operator","FourCastNet 解读：傅里叶神经算子的全球天气预报","FourCastNet（Pathak 等，arXiv 2022，NVIDIA）用自适应傅里叶神经算子在 0.25° 全球网格上做数据驱动预报，一周预报不到 2 秒。本文解读 AFNO 机制、ERA5 训练与对标 IFS 口径，及对风电光伏功率预测的意义。","2023-12-19","\u002Fog\u002Fblog\u002Ffourcastnet-fourier-operator.png",{"_path":383,"title":384,"description":385,"date":386,"author":9,"category":10,"cover":387},"\u002Fblog\u002Fera5-backtest-forecast-model","用 ERA5 历史数据回测新能源功率预测模型：walk-forward 实战","把 ERA5 当作训练\u002F检验底座，用 walk-forward 回测新能源功率预测模型，讲清 RMSE\u002FMAE\u002FMAPE\u002F合格率口径、怎样避免数据泄漏，并衔接 ger 预报做在线评估，附运梦气象 API 可运行示例。","2023-11-26","\u002Fog\u002Fblog\u002Fera5-backtest-forecast-model.png",{"_path":389,"title":390,"description":391,"date":392,"author":9,"category":10,"cover":393},"\u002Fblog\u002Fstorage-dispatch-weather-forecast","储能调度遇见气象：用风光功率预测优化充放电与现货套利","面向新能源算法与运营读者，讲清如何用德国气象局短期预报驱动风光功率预测，进而优化储能充放电、削峰填谷与现货套利，并系统拆解预测不确定度如何反过来重塑 SOC 策略与最终收益，文末附运梦气象 API 实战代码。","2023-11-04","\u002Fog\u002Fblog\u002Fstorage-dispatch-weather-forecast.png",{"_path":395,"title":396,"description":397,"date":398,"author":9,"category":10,"cover":399},"\u002Fblog\u002Fecmwf-vs-gfs","ECMWF-HRES vs GFS — 新能源场景下两个全球预报模式怎么选","ECMWF-HRES 与 NOAA GFS 是两个最常被新能源行业拿来对比的全球数值预报模式。本文从模式分辨率、物理过程、预报技巧、字段完整度、数据访问政策与中国区表现五个维度做技术普及对比，结论是「运梦气象 API 现阶段以德国气象局为主预报源」。","2023-10-13","\u002Fog\u002Fblog\u002Fecmwf-vs-gfs.png",{"_path":401,"title":402,"description":403,"date":404,"author":9,"category":10,"cover":405},"\u002Fblog\u002Fpower-market-weather-bidding","电力现货市场的气象账本：用功率预测支撑报量报价与偏差考核","面向新能源现货交易者：拆解系统负荷、节点电价与场站出力背后的天气驱动，讲清报量报价与偏差考核如何与功率预测分布咬合，并用运梦气象 API 的 ERA5 历史规律加德国气象局短期预报，支撑报价决策与天气风险管理闭环。","2023-09-20","\u002Fog\u002Fblog\u002Fpower-market-weather-bidding.png",{"_path":407,"title":408,"description":409,"date":410,"author":9,"category":10,"cover":411},"\u002Fblog\u002Fwrf-solar-forecast","WRF-Solar：面向光伏功率的太阳能专用数值预报","解读 Jimenez 等在 BAMS 2016 发表的 WRF-Solar：首个面向太阳能预测增强的数值天气预报模式，强化气溶胶-辐射-云的相互作用与快速辐射更新，直接输出 GHI\u002FDNI\u002FDHI，并说明它与运梦气象 API 辐射数据的关系。","2023-08-29","\u002Fog\u002Fblog\u002Fwrf-solar-forecast.png",{"_path":413,"title":414,"description":415,"date":416,"author":9,"category":10,"cover":417},"\u002Fblog\u002Fgencast-probabilistic","GenCast 论文解读：扩散模型如何把 AI 气象推向概率预报前沿","Nature 2024 的 GenCast 由 Google DeepMind 团队提出，把扩散模型适配到地球球面几何，生成 15 天集合概率预报，并在多数指标上优于 ECMWF ENS。本文解读其方法、数据与结论，并梳理对风电光伏资源评估、功率预测与 ERA5 历史气象数据使用的意义。","2023-08-06","\u002Fog\u002Fblog\u002Fgencast-probabilistic.png",{"_path":419,"title":420,"description":421,"date":422,"author":9,"category":10,"cover":423},"\u002Fblog\u002Freanalysis-solar-bias-storage","ERA5 vs MERRA-2 光伏出力系统偏差与长时储能误差传播解读","解读 Applied Energy 2023 年 Mathews 等人的研究：MERRA-2 全球高估光伏出力，会在低往返效率的长时储能链路中放大误差，扭曲充放电与总能量需求；ERA5 更好复现计量发电情景。文末给出运梦气象 API 的 ERA5 字段与 downloadSync 上手思路，帮助选对再分析数据源。","2023-07-15","\u002Fog\u002Fblog\u002Freanalysis-solar-bias-storage.png",{"_path":425,"title":426,"description":427,"date":428,"author":9,"category":10,"cover":429},"\u002Fblog\u002Fweatherbench2-benchmark","WeatherBench 2 解读：AI 气象大模型的统一基准与公平裁判","为 WeatherBench 2（arXiv 2023，JAMES 2024）撰写的中文解读：它如何用 ERA5 真值、统一指标与持续排行榜，为 GraphCast、Pangu-Weather 等 AI 气象大模型与传统 NWP 建立公平基准，并落地风电光伏取数。","2023-06-22","\u002Fog\u002Fblog\u002Fweatherbench2-benchmark.png",{"_path":431,"title":432,"description":433,"date":434,"author":9,"category":10,"cover":435},"\u002Fblog\u002Fpower-forecast-two-rules-compliance","两个细则准确率考核怎么过：ERA5回测+德国气象局预报降考核电量实战","面向新能源场站，逐条拆解国家两个细则功率预测准确率考核口径、短期与超短期合格率及分段罚款机制，并给出用 ERA5 历史回测做系统偏差订正、用德国气象局预报压低考核电量的可落地方法与运梦气象 API 调用示例。","2023-05-31","\u002Fog\u002Fblog\u002Fpower-forecast-two-rules-compliance.png",{"_path":437,"title":438,"description":439,"date":440,"author":9,"category":10,"cover":441},"\u002Fblog\u002Fpangu-weather-3d","Pangu-Weather 解读：3D 神经网络如何重塑中期全球天气预报","解读华为云 Pangu-Weather（Nature 2023）：3D Earth-Specific Transformer 与分层时间聚合、39 年逐小时 ERA5 训练、约 2.56 亿参数，推理较传统 NWP 提速约一万倍，并映射到运梦气象 API 的 ERA5 与德国气象局双数据源落地。","2023-05-08","\u002Fog\u002Fblog\u002Fpangu-weather-3d.png",{"_path":443,"title":444,"description":445,"date":446,"author":9,"category":10,"cover":447},"\u002Fblog\u002Fpower-forecast-horizon-selection","功率预测时效怎么选：超短期、短期、中期的数据源、误差与考核场景","按时效拆解新能源功率预测的方法选型：超短期(≤4h)靠实测外推与爬坡识别，短期(~3天)靠德国气象局(ger)数值预报加实测在线订正，中期(~10天)输出趋势与可信区间，并附运梦气象 API 取数示例。","2023-04-16","\u002Fog\u002Fblog\u002Fpower-forecast-horizon-selection.png",{"_path":449,"title":450,"description":451,"date":452,"author":9,"category":10,"cover":453},"\u002Fblog\u002Firradiance-transposition-perez","Perez 各向异性转置模型 — 从水平面辐照到组件倾斜面","解读 Perez 等在 Solar Energy 1990 发表的各向异性散射转置模型：如何把水平面辐照（GHI\u002FDHI\u002FDNI）转换为光伏组件倾斜面（POA）辐照，区分环日与地平亮带，精度优于各向同性模型，并说明在运梦气象 API 上的工程落地。","2023-03-24","\u002Fog\u002Fblog\u002Firradiance-transposition-perez.png",{"_path":455,"title":456,"description":457,"date":458,"author":9,"category":10,"cover":459},"\u002Fblog\u002Fgraphcast-ai-weather","GraphCast 解读：图神经网络改写中期天气预报，对新能源的意义","GraphCast（Science 2023，Google DeepMind）以图神经网络在 1 分钟内输出未来 10 天全球预报，多数指标超越 ECMWF-HRES。本文解读其方法、ERA5 训练与评测口径，及对新能源资源评估与功率预测的意义。","2023-03-02","\u002Fog\u002Fblog\u002Fgraphcast-ai-weather.png",{"_path":461,"title":462,"description":463,"date":464,"author":9,"category":10,"cover":465},"\u002Fblog\u002Freanalysis-wind-uncertainty","再分析数据用于风资源评估的不确定性：RSER 2022 权威综述解读","基于 Gualtieri 2022 发表于 RSER 的综述撰写的中文解读：系统验证全球 322 个异质站点，界定 ERA5 在海上与平坦内陆可靠、复杂地形受限，并给出风资源评估与运梦气象 API 上手口径。","2023-02-07","\u002Fog\u002Fblog\u002Freanalysis-wind-uncertainty.png",{"_path":467,"title":468,"description":469,"date":470,"author":9,"category":10,"cover":471},"\u002Fblog\u002Fera5-land-reanalysis","ERA5-Land 论文解读 — 陆面再分析金标准与新能源数据底座综述","解读 ECMWF 在 ESSD（2021）发表的 ERA5-Land 论文：它对 ERA5 做无耦合陆面重跑，分辨率提升至约 9km、覆盖 1950 年至今逐小时，土壤湿度、积雪、径流、蒸散等要素与站点观测吻合更好。本文梳理论文方法、关键结论及其对风电光伏资源评估与历史气象数据使用的意义。","2023-01-16","\u002Fog\u002Fblog\u002Fera5-land-reanalysis.png",{"_path":473,"title":474,"description":475,"date":476,"author":9,"category":10,"cover":477},"\u002Fblog\u002Fera5-land-pv-timeseries","ERA5-Land 长时序光伏模拟是否下一跃迁？2020 论文解读","本文基于已核实事实解读 Ramírez Camargo 与 Schmidt 2020 年发表于 Elsevier 的论文：ERA5-Land 结合 PV_LIB 物理模型生成逐时光伏时序，用智利 57 座电站实测验证并与 MERRA-2 对比，剖析其对新能源建模的意义。","2022-12-25","\u002Fog\u002Fblog\u002Fera5-land-pv-timeseries.png",{"_path":479,"title":480,"description":481,"date":482,"author":9,"category":10,"cover":483},"\u002Fblog\u002Fera5-offshore-wind","全球海上风能资源新基准：ERA5 应用经典论文解读","解读里斯本大学 Soares 等人 2020 年发表于 Environmental Research Letters 的经典论文：如何用 ERA5 再分析在全球各国专属经济区刻画海上风功率密度的年度与季节分布，为海上风电选址与资源量化提供基准，并映射到运梦气象 API 的字段与上手路径。","2022-12-02","\u002Fog\u002Fblog\u002Fera5-offshore-wind.png",{"_path":485,"title":486,"description":487,"date":488,"author":9,"category":10,"cover":489},"\u002Fblog\u002Fpv-forecast-weather-inputs","光伏功率预测的气象输入详解：GHI\u002FDNI\u002FDHI 三分量与组件温度修正","从 GHI\u002FDNI\u002FDHI 三分量物理意义讲到 ERA5 的 rsds\u002Fdni\u002Fdhi 获取、用 tas 做组件温度修正、倾角转置与污渍衰减对出力的影响，并给出运梦气象 API 上可运行的 downloadSync 取数与特征组织实战。","2022-11-10","\u002Fog\u002Fblog\u002Fpv-forecast-weather-inputs.png",{"_path":491,"title":492,"description":493,"date":494,"author":9,"category":10,"cover":495},"\u002Fblog\u002Fsolar-position-algorithm-spa","太阳位置算法 SPA — Reda & Andreas 2004 与光伏建模","解读 Reda 与 Andreas 在 Solar Energy 2004 发表的太阳位置算法（SPA）：以约 ±0.0003° 精度计算太阳天顶角与方位角，是光伏跟踪、辐射转置与发电量建模不可或缺的几何基础，并说明如何配合运梦气象 API 的辐射字段使用。","2022-10-18","\u002Fog\u002Fblog\u002Fsolar-position-algorithm-spa.png",{"_path":497,"title":498,"description":499,"date":500,"author":9,"category":10,"cover":501},"\u002Fblog\u002Fera5-hersbach-2020","ERA5 全球再分析奠基论文解读：新能源数据的引用金标准","解读 Hersbach 等 QJRMS 2020 ERA5 奠基论文：31km、137 层、1979 年以来生产系统，并说明当前 CDS 已扩展到 1940 年至今。梳理 4D-Var、规范引用和风电光伏历史气象建模用法。","2022-09-26","\u002Fog\u002Fblog\u002Fera5-hersbach-2020.png",{"_path":503,"title":504,"description":505,"date":506,"author":9,"category":10,"cover":507},"\u002Fblog\u002Fera5-explained","ERA5 完全解读 — 地球 80 年高清气象日记与新能源金标准数据集","一篇读懂 ERA5 再分析数据集：它如何用数据同化把 1940 年至今的全球天气重建成\"无缝地图\"，0.25° 网格、逐小时、137 层垂直要素，以及在风电光伏资源评估、功率预测与长期气候分析中的实战字段选取。","2022-08-12","\u002Fog\u002Fblog\u002Fera5-explained.png",{"_path":509,"title":510,"description":511,"date":512,"author":9,"category":10,"cover":513},"\u002Fblog\u002Fera5-hydrology-north-america","ERA5 能否替代观测驱动模型？北美 3138 流域 HESS 论文解读","解读 2020 年 HESS 顶刊高引论文（被引 535 次）：在北美 3,138 个流域上，将 ERA5 的降水与气温同地面观测及 ERA-Interim 对比并驱动水文模型，看 ERA5 能否替代观测，以及对风电光伏资源评估与功率预测的意义。","2022-07-20","\u002Fog\u002Fblog\u002Fera5-hydrology-north-america.png",{"_path":515,"title":516,"description":517,"date":518,"author":9,"category":10,"cover":519},"\u002Fblog\u002Fweather-data-source-selection","四源气象数据交叉验证与场景选型：ERA5、DWD、卫星、地面实测实战","面向新能源算法与工程团队，系统对比 ERA5 再分析、德国气象局 DWD 数值预报、卫星反演与地面实测四类气象数据源的时空分辨率、时效与偏差，给出交叉验证方法与按风光场景组合选型策略，并附运梦气象 API 可运行示例。","2022-06-28","\u002Fog\u002Fblog\u002Fweather-data-source-selection.png",{"_path":521,"title":522,"description":523,"date":524,"author":9,"category":10,"cover":525},"\u002Fblog\u002Fera5-solar-irradiance-urraca","ERA5 太阳辐照精度评估解读 — Urraca 2018 Solar Energy 奠基综述","解读 Urraca 等 2018 年发表于 Solar Energy 的 ERA5 全球水平辐照（GHI）评估论文：用 BSRN 高质量站、欧洲地面站与卫星产品逐日验证 ERA5，证实其偏差较前代再分析明显降低、内陆可媲美卫星，并映射到风电光伏资源评估与功率预测的实战字段选取。","2022-06-05","\u002Fog\u002Fblog\u002Fera5-solar-irradiance-urraca.png",{"_path":527,"title":528,"description":529,"date":530,"author":9,"category":10,"cover":531},"\u002Fblog\u002Fnwp-quiet-revolution-bauer","数值天气预报的安静革命 — Bauer 2015 Nature 综述","解读 Bauer、Thorpe、Brunet 在 Nature 2015 发表的数值天气预报综述：预报技巧每十年提升约一天、NWP 作为初值问题、资料同化与超算的协同，以及它对新能源功率预测数据底座的意义。","2022-05-14","\u002Fog\u002Fblog\u002Fnwp-quiet-revolution-bauer.png",{"_path":533,"title":534,"description":535,"date":536,"author":9,"category":10,"cover":537},"\u002Fblog\u002Fera5-wind-power-olauson","ERA5 是风电建模新标杆吗？Olauson 2018 顶刊基准论文解读","结构化解读 Olauson 2018 年发表于 Renewable Energy 的 ERA5 风电建模基准论文：ERA5 对照 MERRA-2 全面胜出，MAE\u002FRMSE 平均约低 20%，对海上与平坦地形尤为可靠，并附运梦气象 API 风资源字段上手指南。","2022-04-21","\u002Fog\u002Fblog\u002Fera5-wind-power-olauson.png",1783840669917]