Estimation of Spatial Actual Evapotranspiration to Close Water

Estimation of Spatial Actual Evapotranspiration to Close Water Balance in Irrigation Systems 1 - Key Research Issues 2 - Evapotranspiration through Remote Sensing 3 - SEBAL Applications 4 - Data Requirements and Way Forward Mohsin Hafeez and Shahbaz Khan CSIRO Land Water, Wagga CSIRO LAND and WATER

Water losses and gains are part of the water cycle ET is important at all scales loss gain CSIRO LAND and WATER

Murrumbidgee System Water Account (1991) CSIRO LAND and WATER

Key Research Issues l ET is coupled mass/energy process, linking the energy and water cycles l Estimation of ET is critical for on-farm and regional models in irrigation systems l ET is the largest water balance component after rainfall and irrigation input l Water quantification (i. e. productive and nonproductive use) is important for irrigated agriculture. CSIRO LAND and WATER

Why determine spatial ET? l Classical methods will measure ET at the field scale. – Penman - Monteith (PM) method l Need to have accurate estimates of spatially distributed ET at multi-scales. • Remote sensing provide spatially distributed actual evapotranspiration – Accurate and cheap for large landscape systems – Many RS algorithms developed in last decades CSIRO LAND and WATER

Current state-of-the-art Approaches for measuring ET In-situ measurement (Bowen ratio tower, Lysimeters, etc. ) l l Air-borne measurement (fluxes) l Satellite measurement – High Spatial Resolution (ASTER and Landsat) – High Temporal Resolution (MODIS and NOAA-AVHRR) l Modelling Approaches (plant to catchment) CSIRO LAND and WATER

Methods for Quantification of ET through RS Empirical direct methods 1. - Residual methods of the energy budget 2. – Combination of empirical relationship and physical modules (SEBAL, & SEBS) Deterministic methods 3. – 4. Characterizing crop water status through the cumulative temperature difference (Ts-Ta) Soil-Vegetation-Atmosphere Transfer models (SVAT) Vegetation Index methods CSIRO LAND and WATER

Methods Advantage local Disadvantage Simplified Relationship Operational from regional scale to Spatial variation of coefficients Inference models (Kc f(NDVI)) Operational if combined with ground measurements Requires calibration for each crop type; Kc varies according to water stress Empiricalphysical (SEBAL, …. ) Operational, low cost, need no additional climatic data. Requires presence of wet and dry pixels. Some empirical relationship Deterministic Estimation of intermediate (SVAT, …. ) variables (LAI), links with climate, hydrological models, assimilation to find some parameters Requires more parameters ± easy to estimate. Requires accurate remote sensing data models (PBL, …) Complex and high cost for CPU, only short simulation for high spatial resolution Estimation of climatic data, lateral exchange accounted, possible to stimulate landuse modification CSIRO LAND and WATER

SEBAL Surface Energy Balance Algorithm for Land (SEBAL); thermodynamically based model, which partitions between sensible heat flux and latent heat of vaporization flux. The core of SEBAL is based on the assumption that at hot/dry pixels, all energy flux into the atmosphere is sensible heat and at cool/wet pixels all is latent heat. SEBAL robustly interpolates values at intermediate pixels but is very sensitive to the right choice and flux values at the extremes. CSIRO LAND and WATER

Surface Energy Balance ET is calculated as a “residual” of the energy balance Rn (radiation from sun and sky) ET H (heat to air) ET = R n - G - H Basic Truth: Evaporation consumes Energy Adapted from IDAHAO CSIRO LAND and WATER The energy balance includes all major sources (Rn) and consumers (ET, G, H) of energy G (heat to ground)

SEBAL Derived Actual Evapotranspiration Energy Balance Equation Rn = Go + H+ λE Evaporative Fraction Daily ETa Seasonal ETa CSIRO LAND and WATER The energy balance components

Pre-processing of satellite image Emissivity Surface Temperature Albedo ALBEDO CSIRO LAND and WATER NDVI

Study Area (Lower Murrumbidgee) CSIRO LAND and WATER

Land use Classification for Lower Murrumbidgee 24 October 1990 CSIRO LAND and WATER

Actual ET Distribution for Lower Murrumbidgee 24 October 1990 CSIRO LAND and WATER

ET from different land use classes using Landsat 5 TM sensor 24 October 1990 CSIRO LAND and WATER

Data Requirements l Ground based – temporal variation: – Micro-meteorology and fluxes – Calibration data (soil temperature, LAI, NDVI, LST, albedo, and net radiation) – Vegetation description and surface roughness l Airborne based – spatial variation: – Surface conditions - soil moisture, LST, NDVI, LAI, albedo, – Surface fluxes – Low flying over irrigation supply channels l Satellite based - model requirements: – NDVI, LAI, LST, albedo, emissivity, net radiation, surface roughness – Other data (rainfall, soil moisture, fluxes…. ) CSIRO LAND and WATER

Way Forward l Uncertainty analysis of different input parameters for remote sensing based ET models l Validation of remote sensing derived ET by ground airborne fluxes. l Customization of remote sensing based algorithms for ET estimation for Australian landscape. l Integration of spatial estimation of seasonal ET for water balance studies using system level approach l Flexible for any irrigation system CSIRO LAND and WATER

Working across different scales with universities and other partners as one CSIRO LAND and WATER

Seasonal Evapotranspiration (ETseasonal) l Step 1: Decide the length of the season Step 2: Determine period represented by each satellite image Step 3: Compute the cumulative ETr for period represented by image. Step 4: Compute the cumulative ET for each period l (n = length of period in days) Step 5: Compute the seasonal ET. l l l ETseasonal = ETperiod CSIRO LAND and WATER