Identifying Prolonged Zero Value Periods as Part of Quality Control on Daily Rainfall Records in East Java, Indonesia

Ensuring the quality of surface rainfall records is crucial for obtaining highly representative data and facilitating further comprehensive analysis. Given that surface rainfall observations are predominantly conducted using conventional gauges, they are still susceptible to human errors that can significantly impact data quality. Among various types of errors that may arise, the issue of zero rainfall records is relatively overlooked. Prolonged zero rainfall periods may introduce uncertainty, as mistyped missing data can be erroneously replaced with zero values. The challenge in handling this issue is complicated by the absence of sufficient evidence to conclusively determine the validity or suspicion of consecutive zero rainfall periods. Therefore, we implemented the Affinity Index, altitude difference, and maximum distance approaches to detect and evaluate (validate or reject) any potential invalid sequences of prolonged zero values in the rainfall dataset. The Affinity Index quantifies the agreement of rain and non-rain events between two meteorological stations, functioning as a metric to evaluate the similarity of their rainfall patterns. Utilizing daily data from 682 rain gauge stations in East Java, Indonesia, spanning from January 2010 to December 2019, we identified two major concerns: zero rainfall accumulation during the peak of the rainy season (December/January/February) and extended dry spells lasting more than 180 days. To address the first issue, we flagged the corresponding station and excluded it from the dataset. For the second issue, we established reference stations for each target station to enable meaningful comparisons. The study found that 8.8% of stations detected zero rainfall accumulation during the peak of the rainy season. Regarding prolonged dry spells, we successfully assessed 98% of extended dry spell events in East Java. The majority of these events were considered valid, while around 3% were deemed dubious.