TL;DR
This paper introduces and validates a new global irrigation scheme within the ORCHIDEE land surface model. The scheme improves the representation of irrigation's impact on water, energy, and carbon cycles, reducing biases in evapotranspiration and leaf area index, though it still faces limitations in accurately capturing regional irrigation variations and their full hydrological consequences.
What Problem Does This Paper Address?
Irrigation is a critical human activity, accounting for a massive portion of global water withdrawals and significantly influencing land surface processes, hydrology, and climate. While land surface models (LSMs) are crucial for simulating these interactions, accurately representing irrigation globally remains a challenge. Existing LSMs often simplify irrigation, leading to potential biases in simulated evapotranspiration (ET), leaf area index (LAI), and water resource dynamics. This paper aims to address this by developing and validating a more sophisticated global irrigation scheme within the ORCHIDEE LSM, focusing on better capturing the constraints of water availability and local infrastructure.
Methods
The authors implemented a new global irrigation scheme in the ORCHIDEE v2.2 land surface model. This scheme is based on flood irrigation techniques and incorporates several key features:
- Root Zone Definition: A dynamic root zone depth is defined based on cumulative root density and a user-defined threshold (Rootlim).
- Soil Moisture Deficit Calculation: Irrigation is triggered when the soil moisture falls below a target value, influenced by a parameter $\beta$ that relates the target to field capacity.
- Irrigation Requirement: The irrigation requirement (Ireq) is calculated based on the soil moisture deficit, limited by a maximum hourly irrigation rate (Imax) to prevent excessive runoff.
- Water Availability: The available water (Aw) for irrigation is constrained by three routing reservoirs (stream, overland, and renewable groundwater), with environmental flow restrictions applied.
- Water Adduction: The scheme allows for water adduction from neighboring grid cells to supplement supply when local sources are insufficient.
- Parameter Tuning and Sensitivity Analysis: A one-at-a-time screening method was used to assess the sensitivity of global irrigation rates to various parameters. Parameter tuning was performed using observed irrigation data to find optimal values, particularly for $\beta$.
Datasets
The study utilized a variety of datasets for model input and validation:
- Meteorological Forcing: GSWP3 (Global Soil Wetness Project Phase 3) dataset.
- Irrigated Surfaces: Historical Irrigation Dataset (HID) from Siebert et al. (2015) and the Land Use Harmonization v2 (LUHv2) dataset.
- Validation Data:
- Irrigation Withdrawals: FAO AQUASTAT database (country-level) and a gridded dataset from Sacks et al. (2009) (0.5° resolution).
- Evapotranspiration (ET): GLEAM v3.3a and FLUXCOM (RS+METEO products).
- Leaf Area Index (LAI): LAI3g dataset.
- River Discharge: Monthly data from the Global Runoff Data Centre (GRDC).
- Total Water Storage Anomalies (TWSAs): GRACE satellite observations (CSR, TELLUS, and GSFC products).
- Landscape Descriptors: Fraction of irrigated rice (MIRCA2000) and dam capacity (GRanD database).
Key Insights and Ideas
- Irrigation's Dual Role: The new scheme highlights how irrigation can both reduce negative biases in ET and LAI (especially in heavily irrigated areas) but also exacerbate positive LAI biases in less intensively irrigated regions.
- Water Availability is Key: The model's ability to simulate irrigation is strongly constrained by the availability of freshwater, emphasizing the importance of realistic water allocation rules.
- Regional Heterogeneity Matters: Global uniform parameters, while simplifying the model, lead to significant regional underestimations and overestimations of irrigation, particularly in areas with distinct agricultural practices like paddy rice cultivation.
- Beyond Water Supply: The study points to the need to incorporate other human water management aspects, such as dam operations and fossil groundwater use, to better capture observed trends in water storage and discharge.
- ET Sensitivity: The FLUXCOM ET product appears more responsive to irrigation than GLEAM, suggesting potential limitations in GLEAM's representation of irrigated areas.
Key Results
- The new irrigation scheme matches global irrigation withdrawal amounts reasonably well, though it underestimates them in key regions like India, China, and the USA.
- In all irrigated areas, the scheme reduces the negative bias of ET.
- The scheme exacerbates the positive bias of LAI, except in very intensively irrigated areas where it reduces a negative LAI bias.
- Increased ET due to irrigation leads to decreased river discharge, but this doesn't always improve the representation of discharge dynamics.
- Irrigation has a limited impact on simulated TWSAs and their trends, partly due to the absence of non-renewable groundwater use.
- The parameter $\beta$ was found to have the strongest effect on global irrigation rates, with a value of 0.9 chosen as a compromise for global simulations.
Limitations and Future Work
The authors acknowledge several limitations:
- Simplistic Irrigation Scheme: The scheme uses a single irrigation technique, simplified rules for triggering and allocation, and a single soil column for both rainfed and irrigated crops. Conveyance losses are not represented.
- Oversimplified Parameter Tuning: The use of globally uniform parameters is a significant simplification. Spatially distributed parameters, especially for $\beta$, could better represent local irrigation systems and biases.
- Data Uncertainty: The use of a single meteorological forcing dataset and a single year for observed irrigation values introduces uncertainty in parameter adjustment.
- Model Uncertainties: ORCHIDEE itself has inherent uncertainties related to bare soil evaporation, infiltration, and other hydrological processes that can impact irrigation estimates and efficiency.
- Absence of Non-renewable Groundwater: The lack of representation for fossil groundwater abstraction likely contributes to the underestimation of irrigation in arid and semiarid regions.
- No Dam Management: The scheme does not explicitly represent dam operations, which are crucial for modulating water supply in many large river basins.
Future work suggested includes:
- Explicit Paddy Rice Representation: Incorporating specific irrigation techniques for paddy rice to reduce biases in South Asia.
- Dam Management and Fossil Groundwater: Including these processes to better represent water supply and potentially reduce negative biases in TWSAs and discharge.
- Spatially Distributed Parameters: Moving beyond uniform parameters to better capture regional variations in irrigation practices and their impacts.
- Coupled Simulations: Utilizing the improved irrigation scheme in coupled atmosphere-land surface models to explore the joint evolution of climate, water resources, and irrigation under future scenarios.
Summary updated