Freshwater is an essential resource for human survival and underpins a wide range of economic activities. Among the various sectors that are dependent on freshwater, agriculture is the largest consumer, accounting for approximately 85 % of all human freshwater consumption (D'Odorico et al., 2020). Projections for the year 2050 indicate a global population increase to more than 9.0 billion, which will further exacerbate the pressure on existing freshwater resources for food production, particularly groundwater (Boretti and Rosa, 2019). Recent studies have highlighted the significant depletion of groundwater resources in major food-producing regions and other areas worldwide (Feng et al., 2013; Jasechko et al., 2024; Kuang et al., 2024; Liu et al., 2022). In response to this issue, there is an urgent need for the conjunctive and optimal use of groundwater and surface water resources. Nations such as India, Pakistan, and Bangladesh have increasingly promoted the use of surface water resources to alleviate pressure on groundwater reserves (Paul and Hasan, 2021). To enhance the effectiveness of these efforts, systematic and frequent water accounting is essential. This process should comprehensively account for direct human water use, including agricultural, municipal, commercial, and industrial consumption, as well as indirect losses due to reservoir evaporation and evapotranspiration (Richter et al., 2024). Such analyses are critical for quantifying stress on different water sources and informing strategies for their sustainable and efficient management to meet growing water demands.

Assuming a scenario where both surface water and groundwater resources are sufficient to meet food production needs, and we have a robust infrastructure for supplying from both sources, surface water is the more sustainable source due to its multifaceted benefits. Firstly, surface water often contains higher nutrient levels than groundwater, which can lead to increased crop yields. Secondly, by reducing the reliance on groundwater pumping, surface water allows aquifers to recover naturally by acting as a “storage battery” during dry seasons and ensuring a long-term water supply for agricultural and other uses. Thirdly, and perhaps most importantly, surface water irrigation systems are mostly gravity-driven and require less energy than groundwater pumping systems. This reduction in energy consumption translates to lower greenhouse gas emissions, thereby reducing the carbon footprint of agricultural practices (Qin et al., 2024). Accounting for government policies, over-exploitation of existing irrigation water resources, and the benefits of surface water irrigation, there is a need to manage surface water resources sustainably and efficiently.

To manage surface water irrigation more efficiently, adapting traditional irrigation practices to address the challenges posed by climate variability and increasing demand is crucial. Rising temperatures and changing rainfall patterns significantly impact crop growth and surface water availability. This in turn, demands that current surface water irrigation practices become more optimized. Studies led by Kovenock and Swann (2018) and Farhat et al. (2021) highlight the complex relationship between rising temperatures and biomass production. Kovenock and Swann (2018) have reported that increased carbon dioxide (CO2) concentrations can negatively affect biomass productivity and alter the leaf area, which reduces the resilience against climate change. Farhat et al. (2021) showed that the elevated temperature leads to a higher uptake of toxic substances in groundwater such as arsenic. This bioaccumulation eventually results in plant failure at elevated temperatures, affecting both food quality and quantity. In many places, such as in Bangladesh where groundwater can often be contaminated by arsenic (Podgorski and Berg, 2020), the surface water usually has a better quality for irrigation applications. Overall, surface water can present a better alternative, in many places, in the light of climate change, for mitigating these adverse effects on food production in terms of quality and sustainability.

To fully leverage the benefits of surface water and adapt to climate variability, it is essential to overcome the limitations of traditional irrigation practices. Traditional-irrigation-practices specifically refer to those employed by the water-providers, encompassing temporal, intensity-based, and spatial components of irrigation decision-making. These components are broken down into the following:

When to irrigate (time – T) – determining whether water should be distributed in the current cycle based on the combined effects of the previous irrigation cycle, precipitation response, and crop water need;

In reality, traditional irrigation practices are not limited to water-providers but extend to water-consumers (farmers), who rely heavily on traditional irrigation calendars. These calendars, which are static and developed based on the historical climatology of the previous century assume a fixed cropping pattern. These calendars inform farmers of the irrigation schedule to follow but are becoming increasingly inadequate in the face of rising food production demands. The reliance on these static crop calendars leads to inefficiencies in water management and reduced agricultural productivity, as evidenced in many countries such as India, Bangladesh, Pakistan, and Australia (Bose et al., 2021; Bretreger et al., 2020; Franch et al., 2022; Hossain et al., 2017). Regardless of the region, farmers typically are not adaptive to rapid climate variability and cannot accurately predict precipitation events and their amounts at sub-seasonal irrigation timescales. This uncertainty often leads to overwatering of the fields to avoid the risk of crop loss due to rising temperatures. Overwatering, however, can inadvertently wash away essential soil nutrients, while underwatering can stress crops, both resulting in reduced yields. These practices lead to inefficiencies in water use and diminish soil fertility and agricultural productivity (Hossain et al., 2017).

Considering the effects on agricultural yield due to altered rainfall patterns and rising temperatures, there is an urgent need for a scientific-data-driven approach to optimize existing irrigation practices to achieve maximum crop water productivity. The potential of using satellite data for sustainable agriculture and efficient irrigation has been widely acknowledged in recent years (Deines et al., 2019; Sishodia et al., 2020; Zhang et al., 2022). Several studies have employed crop coefficients (Kc) based evapotranspiration techniques to enhance the understanding of irrigation schedules and irrigate fields based on actual water needs (Bretreger et al., 2020; Gabr and Fattouh, 2021). Tools like the Food and Agriculture Organization's (FAO) CROPWAT, which utilizes the Penman–Monteith equation (Allen et al., 1998), have provided valuable insights into regional water requirements (Dong, 2018; Khaydar et al., 2021; Solangi et al., 2022; Surendran et al., 2015).

Despite significant advancements, substantial gaps remain in the current body of research that need to be addressed. These gaps arise from limitations in the temporal and spatial resolution of past studies. The trade-off between temporal and spatial resolution of satellite data makes it challenging to estimate evapotranspiration (ET) at high temporal and spatial resolutions simultaneously. Typically, temporal resolution ranges from daily to monthly time steps. When the temporal resolution is one day, the spatial resolution is usually at the kilometer scale (Chen et al., 2021). Conversely, if a medium spatial resolution sensor having 30 m pixels is selected, the temporal resolution is typically around 16 d from polar-orbiting optical satellites (Tasumi, 2019). To achieve an optimal balance between spatial and temporal resolution, a combination of satellites needs to be used. Additionally, challenges exist in operationalizing findings or tools due to the latency of the available latest meteorological data. For instance, the Global Land Data Assimilation System (GLDAS), which provides global meteorological forcing data, can be used in estimating evapotranspiration (Pareeth and Karimi, 2023), has an early product stream with a latency of about 1.5 months. This latency hinders the ability to use these datasets for near real-time operational studies.

Tools available for surface water irrigation also suffer from an inability to accurately distinguish between the actual water provided to plants and the water needed by plants. This arises from the difficulty in estimating ET under ambient water stress conditions for each crop and lack of metering and monitoring. The incomplete information makes it difficult to estimate net water requirements and project potential water savings. Additionally, there is a lack of integration of precipitation forecasts at the operational level to optimize irrigation even further. It is important to acknowledge that, to the best of our knowledge, there is no freely available open-source tool that addresses these gaps in optimizing surface water irrigation schemes based on satellite data.

In this study, we present Satellite Data Rendered Irrigation using Penman–Monteith and SEBAL (sDRIPS, pronounced as “drips”), a cloud-based optimized surface water irrigation advisory system that leverages earth observations and weather models. The founding concept of sDRIPS originated from the Irrigation Advisory System (IAS), first prototyped by Hossain et al. (2017). Later, Bose et al. (2021) integrated Gravity Recovery and Climate Experiment (GRACE) satellite-based water storage data with Landsat data to quantify the impact of IAS. Bose et al. (2021) also used satellite and model weather data to estimate ET using the Penman–Monteith method (Allen et al., 1998) and the Surface Energy Balance Algorithm for Land (SEBAL; Bastiaanssen et al., 1998) as proxies for crop water demand and actual water consumed, respectively (Bose et al., 2021).

The IAS framework has since been operationalized across South Asia under region-specific implementations, including IAS in Pakistan (Hossain et al., 2017), Provision for Advisory on Necessary Irrigation (PANI) in India (Hossain et al., 2020), and Integrated Rice Advisory System (IRAS) in Bangladesh (Hossain et al., 2022). IRAS generates irrigation advisories for farmers by comparing Penman–Monteith based estimates of crop water demand with SEBAL-derived actual evapotranspiration. This comparison enables systematic identification of potential over- and under-irrigation at the field scale using Landsat thermal infrared observations, forming the basis for actionable, crop-specific irrigation guidance (Bose et al., 2021).

Due to these capabilities, IRAS has been adopted by the Department of Agricultural Extension (DAE), of the Ministry of Agriculture of Bangladesh (https://iras.bamis.gov.bd/, last access: 9 June 2026, or visit https://bamis.gov.bd, last access: 9 June 2026, and click “Satellite Products”). Currently, IRAS operates on DAE servers, providing irrigation advisories to farmers across approximately 2000 agricultural districts located in the Northwestern and Northeastern regions of Bangladesh (McNamee and Mersmann, 2023).

Given the recent adoption of the farmer/user-centric advisory system of IRAS that optimizes water use at the consumer side, we believe there is now an opportunity for an operator-centric or water-provider-centric irrigation advisory system for surface water irrigation at the supply side. This proposed operator-centric (hereafter alternated with “provider-centric”) advisory can potentially guide irrigation water managers to optimize water delivery at the supply side, where water is being centrally withdrawn from a river for dispatch to the irrigation command areas. Water suppliers and canal operators can be informed as to how much water should be allocated or delivered to each region or canal based on the water needs of all the farmers of that region. By assimilating nowcast and forecast precipitation data that are now available from weather forecast systems with the previous week's water allotment to the region and the current crop water demand, an accurate estimate of net water requirement for each region can be estimated. This net water requirement can be translated to the flux of surface water (L3/T) that would need to be maintained in water-supplying canals at the supply side for water managers. Consequently, such a system can support dispatch decisions for surface water irrigation based on data-driven science, crop water needs, and ambient environmental conditions.

In this study, our central research question is: How can we dynamically optimize a multi-branched canal irrigation system using earth observation satellites? We believe that addressing the sustainable use of surface water requires tackling it from both the consumer side (by farmers/users) and supply side (by water managers/canal operators) with an operator-centric system as a complement to the user-centric system. The key goal of this study is to guide canal operators and help them make informed decisions on how much surface water to dispatch centrally from the source at the river for the command areas to maximize agricultural water productivity. We developed and implemented sDRIPS for the Teesta River Irrigation system located in Northern Bangladesh where there is currently an urgent need to optimize the use of surface water resources.

The sDRIPS study focuses on the Teesta Barrage Project (TBP) in Bangladesh, the country's largest surface water irrigation project. Established in 1990, the TBP supplies irrigation water from January to April and spans the Teesta River at Dalia-Doani Point in the Lalmonirhat district (Fig. 1 Middle Panel). This project features a 615 m-long concrete structure equipped with 44 radial gates, providing a discharge capacity of 361 m3s-1 (12 750 cfs) and supporting a command area of 1.5425×109 m2 (154 250 ha) through a 4500 km network of canals (River Research Institute, 2023). For more details on TBP, readers are referred to (River Research Institute, 2023).

The main TBP canal divides into three primary canals: Dinajpur Canal, Rangpur Canal, and Bogra Canal (Fig. 1 Middle Panel), which further branches into an intricate network of secondary and tertiary canals. Selecting the TBP as a study site offers two primary advantages: the utilization of in-situ water supply data at the head canal and the opportunity to study a complex canal network system. A successful implementation of our study to implement sDRIPS for TBP here means it can be adapted and applied to surface water irrigation systems around the world. For this study, irrigation canal shapefiles, command areas, and in-situ water supply data were provided by the Bangladesh Water Development Board (BWDB).

The publicly available datasets used in sDRIPS allow for accurate assessments of the actual water needs of command areas while considering climatic factors. To facilitate implementation and reduce technical barriers, the cloud computing platform Google Earth Engine (GEE), developed by Gorelick et al. (2017), was utilized. The cloud computing platform helps sDRIPS to eliminate the need to download and process satellite and weather data on the local machine. Table 1 provides a comprehensive list of the datasets utilized by sDRIPS. Some of the key datasets, which are referenced by their initials in Table 1 but not previously mentioned in the text are – Shuttle Radar Topography Mission (SRTM), Global Land Cover Characterization (GLCC), National Oceanic and Atmospheric Administration (NOAA), Soil Moisture Active and Passive (SMAP), and International Soil Reference and Information Centre (ISRIC).

An overview of sDRIPS is first provided along with an illustration (Fig. 2), followed by a brief description of the steps of the processes. These steps are then covered in detail in the methodology section.

When a Landsat satellite overpasses a command area, sDRIPS collects crucial information required to estimate the actual water needs of the crop based on its growth stage using Penman–Monteith ET, and the amount of water applied to the crop using SEBAL ET. The impact of precipitation is also considered to refine the water demand estimation. This includes both nowcast precipitation data and forecasts of incoming precipitation. The system then compares these estimates to determine whether the water needs of all crops in the given command area are met. If the net water need is not satisfied, the area is classified as a deficit region, and the volume of water needed for dispatch for the coming week is quantified (Fig. 2). Conversely, if the water demand is met and there is an excess of provided water, the area is classified as a surplus region. The volume of surplus water that can be potentially stored locally and used later for nearby deficit regions is quantified (Fig. 2). This quantification allows water providers to monitor and adjust the water supply and dispatch decisions from the main canal efficiently to meet the specific needs of each command area.

Figure 3 illustrates the proposed methodology for the irrigation advisory system, sDRIPS. Before starting, users need to have a GEE account (https://signup.earthengine.google.com/, last access: 9 June 2026) to estimate Penman–Monteith and SEBAL evapotranspiration and a Precipitation Processing System (PPS) account (https://registration.pps.eosdis.nasa.gov/registration/, last access: 9 June 2026) to download the GPM IMERG precipitation data.

Step 1:

Create command areas based on the availability of command area data: This includes creating configuration files for the command areas and updating the script configuration files accordingly to run the system.

In this study, we co-developed and implemented a water-provider-centric irrigation advisory system, sDRIPS, to manage water resources sustainably and robustly while addressing existing gaps in the literature. sDRIPS integrates satellite and model data to estimate the actual water needs of command areas. This crucial information aids water providers in determining the appropriate water allocation for each command area. The actual water need depends on several factors, including the amount of water applied in the previous irrigation cycle, current precipitation, and forecasted precipitation. By considering these factors and the available water supply, sDRIPS advises canal operators on which command areas require water and the amount needed. During periods of excess water supply, sDRIPS advises canal operators on how much water can be stored for future irrigation needs or diverted to water-deficit regions outside the irrigation project. Conversely, during water shortages, sDRIPS generates water stress maps to identify regions experiencing water shortages and guides on the amount of water needed in deficit areas. sDRIPS also allows for the simulation of various scenarios using historical data on different crops. This capability enables stakeholders to evaluate the impact of future policies on water supply conditions if these policies are implemented today and to devise science-based responses. By offering a flexible and comprehensive tool for water management, sDRIPS has the potential to contribute significantly to sustainable irrigation practices globally, addressing the challenges posed by a changing climate and inefficiency in managing surface water and groundwater resources.