Abstract Precipitation regionalization deals with an investigation of the seasonality and its temporal variability and is useful for a wide variety of applications in hydro-meteorology. The d homogeneous regions can be used as a basis for transforming the information from gauged to ungauged sites and can reduce the uncertainty in estimating the seasonal characteristics of precipitation across India. Despite several studies stressing the importance of seasonality and temporal variability to the environment, there is a lack of studies on accounting for temporal variability in regionalization. Precipitation regionalization must account for both the precipitation magnitude and its temporal variability at multiple time-scales to extract the seasonality of a region representing coherent local and inter-annual variability. Therefore, in this study, we propose a framework for precipitation regionalization, considering both precipitation magnitude and its temporal variability. High resolution (0.25° × 0.25°) gridded daily precipitation time series over the period 1901–2013 from Indian Meteorological Department (IMD) was used for the evaluation of the framework. First, the historical daily time series was transformed into multiple time scales, i.e., annual, seasonal, and monthly time scales. Entropy-based standardized variability index was used to measure the inter-annual variability of precipitation at each time scale. Regionalization of grid points was performed using self-organizing maps, an artificial neural network. Ten distinct regions were identified that can be tied back to two general categories, such as climate characteristics and physical characteristics. Coupling of the self-organizing map with standardized variability index reveals unique seasonal distribution of precipitation for each region. The temporal evolution of clusters unravels a new emerging pattern across Central India. Consideration of temporal variability plays an insignificant role in the shape, size and stability of south-central India, south-eastern coastlines, and Konkan Coast. Intriguingly, separate Rain-belt and Rain-shadow Western Himalayas are formed due to the difference in topography and seasonal characteristics of precipitation. The temporal evolution of clusters unravels a significant change in the occurrence of the 50th percentile monsoon after the 1940s across the north-western region; a significant increase in the 50th percentile monsoon after the 1940s across western India, and decrease in the 50th percentile monsoon after the 1980s in the north-central Region.

中文翻译：

---

基于自组织图结合信息论的改进降水区划的时间变异性解释

摘要 降水区域化涉及季节性及其时间变化的调查，可用于水文气象学的广泛应用。d 均质区域可用作将信息从测量地点转换为未测量地点的基础，并可以减少估计印度各地降水季节性特征的不确定性。尽管有几项研究强调了季节性和时间变异对环境的重要性，但缺乏对区域化时间变异性的研究。降水区域化必须考虑到降水量及其在多个时间尺度上的时间变化，以提取代表连贯的局部和年际变化的区域的季节性。因此，在本研究中，我们提出了一个降水区域化框架，同时考虑了降水量及其时间变异性。来自印度气象部门（IMD）的高分辨率（0.25°×0.25°）网格日降水时间序列在1901-2013年期间用于评估该框架。首先，将历史日时间序列转化为多个时间尺度，即年、季、月时间尺度。基于熵的标准化变率指数用于衡量各时间尺度降水的年际变率。网格点的区域化是使用自组织地图（一种人工神经网络）进行的。确定了 10 个不同的区域，它们可以与两个一般类别联系起来，例如气候特征和物理特征。自组织地图与标准化变异指数的耦合揭示了每个地区降水的独特季节性分布。集群的时间演变揭示了印度中部新出现的模式。考虑时间变化对印度中南部、东南海岸线和康坎海岸的形状、大小和稳定性起着微不足道的作用。有趣的是，由于降水的地形和季节特征的差异，形成了独立的雨带和雨影西部喜马拉雅山脉。星团的时间演化揭示了 1940 年代后西北地区 50% 季风发生的显着变化；1940 年代后印度西部第 50 个百分点的季风显着增加，