Contents lists available at ScienceDirect Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat Operational soil moisture modeling using a multi-stage approach based on the generalized complementary principle Nastiti Andini a , Daeha Kim b, *, Jong Ahn Chun b, * a Climatological Station of West Seram, Indonesian Agency for Meteorology Climatology and Geophysics (BMKG), Jl. Hunitetu Km.1 Kairatu, Seram Bagian Barat, Maluku, 97566, Indonesia b APEC Climate Center, 12 Centum 7-ro, Haeundae-gu, Busan, 48058, Republic of Korea ARTICLE INFO Keywords: Soil moisture Evapotranspiration Generalized complementary relationship Single bucket model El Niño ABSTRACT A higher drought risk in Java Island is generally known than the other regions in Indonesia. Tracking soil moisture can be an alternative way to monitor drought rather than precipitation-based drought indices. The objective of this study was to assess root-zone water storage (defined by root-zone soil moisture contents) based on a linked approach between the generalized complementary relationship (GCR) and a single bucket model in Java Island. Since it does not require precipitation for estimating actual evapotranspiration (ET a ), the GCR allowed implementation of a simple single bucket model. The ET a and root-zone soil moisture estimated in this study were compared against the Global Land Evaporation Amsterdam Model (GLEAM) and the root-zone water storage additionally compared with the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 reanalysis data products. Overall, the GCR ET a estimates were higher than those from GLEAM, and similar patterns of the root-zone water storage were found in the comparisons of both GLEAM and ERA5. The com- parative evaluation suggests a further study on the adjustment of Priestley-Taylor coefficient value in Java for better application of the GCR. The soil moisture estimated by the single bucket model and the root-zone soil moisture products of GLEAM were highly correlated (0.8 or greater Pearson correlation coefficients). Low root- zone water storage and high ET a rates were found in eastern Java relative to the other areas, indicating high water shortage risks in dry season. This study found that El Niño clearly contributed to the variability of the root- zone water storage in Java especially in wet seasons (December to February). It is also suggested that the proposed approach can be useful to operationally provide soil water availability in Java from readily available meteorological observations. 1. Introduction Indonesia is a hazard-prone country where more than 80% of nat- ural disasters are driven by hydrometeorological extremes (e.g., heavy rainfall and persisting dryness). Indonesian National Board for Disaster Management (BNPB, 2016) assessed that drought occurrences have increased and threatened agricultural productivity in rural areas in Indonesia. Given the high dependency of Indonesian economy on agriculture (Kishore et al., 2000), operational monitoring of soil moisture deficiency is essential for timely management of agricultural droughts and societal resilience. Highly accurate monitoring of soil moisture status could be achieved from point-scale observations. However, those datasets often have limitations due to poor spatial coverage (Miralles et al., 2012; Peled et al., 2010). To make up for this drawback, remote-sensing techniques have been used, e.g., European Space Agency’s Climate Change Initiative (Liu et al., 2012) and from ERA-Interim/Land (Balsamo et al., 2015). The remote-sensing data products could be further processed for more detailed analysis of land-surface energy fluxes and root-zone soil moisture (e.g., the Global Land Evaporation Amsterdam Model (GLEAM); Martens et al., 2016). However, required time costs for data processing and modeling could hinder proactive management of agricultural drought risks. To resolve the challenges, soil moisture can be synthesized by a water balance model. As a typical example, the Tank model (Sugawara, 1995) could be used for generating water storage in upper and lower soil zones. While this conceptual model was formulated with multiple layers of linear reservoirs to generate catchment runoff responses, changes in terrestrial water storage could be synthesized together. The simplest variation of the tank model is to employ a single linear tank https://doi.org/10.1016/j.agwat.2020.106026 Received 28 March 2019; Received in revised form 26 December 2019; Accepted 9 January 2020 ⁎ Corresponding authors. E-mail addresses: d.kim@apcc21.org (D. Kim), jachun@apcc21.org (J.A. Chun). Agricultural Water Management 231 (2020) 106026 Available online 14 January 2020 0378-3774/ © 2020 Elsevier B.V. All rights reserved. T