In this study, a comprehensive assessment has been provided for the performance of three newly-developed high resolution reanalysis (IMDAA, NGFS, ERA5) datasets w.r.t. IMD rainfall observation over six homogeneous monsoon regions of India during 1999–2018. It is necessary to understand how these reanalyses rainfall datasets are performing spatio-temporally, specifically during monsoon in order to apply them in further research and developmental purposes. Focus has been given to assess their relative performances during monsoonal rainfall extremes against IMD observations. Results reveal that all these three reanalysis datasets captured the spatial patterns of observed rainfall climatology reasonably well in all the seasons, except for hilly regions where there is over-estimation against the observed patterns. Moreover, wet biases over the entire foothills of Himalayas extending up to north-east are more dominant during monsoon for IMDAA reanalysis, whereas bias is mixed type and region-specific in all seasons for NGFS and ERA5 reanalysis. Both IMDAA and NGFS reanalysis captured the probability distribution of observed daily rainfall intensity diligently especially in the range of extremes while ERA5 reanalysis underestimated the same. However, spatial distribution for rainfall extremes (based on intensity and 95th percentile) reveals that all the reanalyses have failed to capture patterns in the hilly topography as well as over peninsular Indian region. In comparison to IMDAA and ERA5 reanalysis, NGFS captured reasonably well for heavy and very heavy rainfall over Western Ghats. ERA5 reanalysis always underestimated all the extreme rainfall categories in comparison to IMDAA and NGFS reanalysis which is also confirmed from probability distribution curve. Moreover, bi-decadal (pre- and post-2008) spatial trend distribution for 95th percentile rainfall extremes indicate a reversal in trend (wetter to drier and vice versa) over India. Both IMDAA and NGFS have captured the moist and desiccated trend region-specifically along with ERA5 reanalysis against observational trends. Thus, the result provides useful insights for the climate modeling to establish appropriate standards for performing climate model evaluation over Indian region.

在这项研究中，已对1999-2018年印度六个均质季风区的IMD降雨观测到的三个最新开发的高分辨率再分析数据集（IMDAA，NGFS，ERA5）的性能进行了全面评估。有必要了解这些重新分析的降雨数据集在时空上的表现如何，特别是在季风期间，以便将其应用于进一步的研究和开发目的。重点是根据IMD观测评估其在季风降雨极端期间的相对性能。结果表明，这三个再分析数据集在所有季节中都很好地捕获了观测到的降雨气候的空间格局，除了丘陵地区，那里对观测到的格局有过高估计。而且，在季风期间，喜马拉雅山整个山麓直至东北的湿偏向在IMDAA重新分析中更为明显，而在NGFS和ERA5重新分析的所有季节中，偏见是混合类型和区域特定的。IMDAA和NGFS重新分析都努力地捕获了观测到的每日降雨强度的概率分布，尤其是在极端范围内，而ERA5重新分析低估了该概率分布。但是，针对极端降雨的空间分布（基于强度和第95个百分位数）表明，所有重新分析均未能捕获丘陵地形以及印度半岛上空的格局。与IMDAA和ERA5的再分析相比，NGFS在西高止山脉上的大雨和大雨中都捕获得相当好。与IMDAA和NGFS重新分析相比，ERA5重新分析总是低估了所有极端降雨类别，这也从概率分布曲线中得到了证实。此外，印度第95个百分位降雨极端事件的双年代际（2008年之前和之后）空间趋势分布表明印度的趋势发生了逆转（从干燥到干燥，反之亦然）。IMDAA和NGFS都已经捕获了潮湿和干燥的趋势区域，同时针对观察趋势对ERA5进行了重新分析。因此，结果为气候建模提供了有用的见识，以建立适当的标准以在印度地区进行气候模型评估。95％百分率的极端降雨的双年代际（2008年之前和之后）空间趋势分布表明印度的趋势发生了逆转（从干燥到干燥，反之亦然）。IMDAA和NGFS都已经捕获了潮湿和干燥的趋势区域，同时针对观察趋势对ERA5进行了重新分析。因此，结果为气候建模提供了有用的见识，以建立适当的标准来执行印度地区的气候模型评估。95％百分率的极端降雨的双年代际（2008年之前和之后）空间趋势分布表明印度的趋势发生了逆转（从干燥到干燥，反之亦然）。IMDAA和NGFS都已经捕获了潮湿和干燥的趋势区域，同时针对观察趋势对ERA5进行了重新分析。因此，结果为气候建模提供了有用的见识，以建立适当的标准以在印度地区进行气候模型评估。