Dry spell studies are of vital importance for agricultural planning and water management. This study characterized dry spells in the state of Rio de Janeiro (SRJ) - southeastern Brazil - based on statistical tests, multivariate analysis and spatial distribution. Daily rainfall data from 1995 to 2017 were obtained from 86 rainfall stations located in the SRJ and neighbouring states. The data were submitted to quality analysis and gap filling of data using the simple linear regression method. The start of a dry spell was considered after three consecutive days with rainfall < 1 mm during the rainy season (November to March). A dry spell was considered the period with at least three consecutive dry days (CDD) and is divided in four classes of dry spells - Class A (3 - 6 days), B (7-10 days), C (11-14 days) and D (15 days or more) – were established for the SRJ. The Shapiro-Wilk (SW), Anderson-Darling (AD), Kolmogorov-Smirnov (KS), Jarque-Bera (JB) and Bartlett (B) tests were also applied to the time series to validate data. The SW (83.72%), AD (74.42%), KS (55.81%) and JB (80.23%) tests indicated non-normality of the data. The classes of dry spells registered different frequencies of occurrence, with Class A (70.03%), B (17.98%), C (6.08%) and D (5.91%). Spatially, there was a high variability of dry spells in the south of the state with the shortest prolonged dry spells, while in the north dry spells are usually longer, with emphasis on February and March. Principal Component Analysis (PCA) was applied to eight variables for Class A (most frequent), and identified latitude, longitude and, particularly elevation, as variables that influence the spatial distribution of dry spells, with highlights for the summer (December and January) season. The high spatial-temporal variability of dry spells in Rio de Janeiro is influenced by multi-scale meteorological systems, with an emphasis on frontal systems and physiographic factors. The applied methodology and presented results can be used to improve public policies regarding water management and mitigate the effects of droughts assuring the quantity and quality of water resources in the development of the SRJ.

干式拼写研究对于农业计划和水管理至关重要。这项研究基于统计测试，多元分析和空间分布，对巴西东南部里约热内卢州（SRJ）的干旱季节进行了特征分析。1995年至2017年的每日降雨量数据是从SRJ和邻近州的86个降雨站获得的。使用简单的线性回归方法将数据提交给质量分析和数据缺口。在雨季（11月至3月）连续三天降雨小于1毫米之后，才考虑开始干旱。干旱时期被认为是至少连续三个干旱日（CDD）的时期，分为四个干旱时期类别-A类（3-6天），B类（7-10天），为SRJ建立了C（11-14天）和D（15天或更长时间）。Shapiro-Wilk（SW），Anderson-Darling（AD），Kolmogorov-Smirnov（KS），Jarque-Bera（JB）和Bartlett（B）测试也应用于时间序列以验证数据。SW（83.72％），AD（74.42％），KS（55.81％）和JB（80.23％）测试表明数据非正常。干旱时期的类别记录了不同的发生频率，其中A类（70.03％），B类（17.98％），C类（6.08％）和D类（5.91％）。在空间上，该州南部的干旱时期变化很大，而干旱时期则最短，而在北部，干旱时期通常较长，主要集中在2月和3月。将主成分分析（PCA）应用于A类（最常见）的八个变量，并确定了纬度，经度，尤其是海拔高度，作为影响旱季的空间分布的变量，夏季（十二月和一月）季节的亮点。里约热内卢干旱时期的高时空变异性受到多尺度气象系统的影响，重点是额叶系统和地理因素。所应用的方法和提出的结果可用于改善有关水资源管理的公共政策，并减轻干旱的影响，从而确保SRJ开发过程中水资源的数量和质量。