top of page
  • Writer's pictureArpit Shah

Estimating Evotranspiration using Remote Sensing

Updated: Jul 3


When Jessey Dickson, a bright Ghanaian pursuing his MSc. in Environmental Quality Sciences on scholarship from the Hebrew University of Jerusalem reached out via LinkedIn late-January this year with a particular request - 'Can you make a tutorial for estimating Evotranspiration using SNAP software?', my initial feeling of surprise was overcome by a sense of satisfaction.

Israel Agriculture
Figure 1: Jessey at a remote agri-research site in Israel

I felt a bit bemused as I had never even heard of 'Evotranspiration' before, let alone know how to derive it. But to realize that my work on this Geo-blog was increasingly being seen and appreciated by young students globally signalled to me that while I grapple with the the private sector in India in my quest to offer a variety of mapping solutions that aid in operations improvement, moments such as these would bring necessary joy and encouragement along my journey.

Nonetheless, as lazy as I tend to be, a few superficial responses wouldn't satisfy the ever-so-persistent Jessey who had come in with a singular objective - to find a way to learn and complete the challenging assignment on 'estimating Evotranspiration using Remote Sensing' as given by his professor. I eventually acceded and became more involved with him.

While Jessey would focus his study on the farmlands around Gadot in Israel, I gave myself another area of interest: a similar 1400 sq. km zone in Punjab, India, to do my Evotranspiration study in parallel. After hours of exchanges, video meetings & SNAP software workings, we were finally able to fulfill our objectives - mine being this elaborate, end-to-end video tutorial on estimating field-scale Daily (Actual) Evotranspiration using Remote Sensing, which also forms the cornerstone of this article.




Evotranspiration & factors affecting it

Measuring Evotranspiration

Utility of Evotranspiration

Using Remote Sensing to estimate field-scale Daily (Actual) Evotranspiration

My Evotranspiration Study over an Agri-Zone in Punjab (IN) as on 4th June 2022

Final Observations



Evotranspiration (ET), which represents the total loss of water (& energy) from the earth's surface into the atmosphere, is a combination of two terms - ‘Evaporation’, which is the direct movement of water - from soil, canopies, capillary fringe of the groundwater table and water bodies on land - into the atmosphere & ‘Transpiration’, which is the indirect transfer of water from the soil surface into the atmosphere via the leaves & roots of vegetation. This release of water vapour into the atmosphere forms a crucial component (& the largest after precipitation) of our planet's Hydrologic Cycle.

Water Cycle Diagram. Left half of it pertains to Evotranspiration.
Figure 2: Water Cycle Diagram. Left half of it pertains to Evotranspiration. Source: LangeLeslie, CC BY-SA 4.0 <>, via Wikimedia Commons

[What could be the possible benefits of computing water (& energy) transfer into the atmosphere? Perhaps you can brainstorm internally before I elaborate later in this article.]

The factors which impact Evaporation can be summarized under five categories as outlined below-

Factors affecting Evotranspiration
Figure 3: Factors affecting Evotranspiration


There are three ways in which Evotranspiration (ET) measurements are conveyed:

Actual ET, Potential ET & Reference ET.

Actual Evotranspiration, as the name suggests, involves the computation of ET in a real-world setting. To expand, it involves figuring how much actual water vapour (& energy) was released by the vegetation & soil, in a particular farm for example, into the atmosphere over a period of time. Actual ET has to factor in a) actual Precipitation & b) existing Soil moisture, among other variables.

Potential Evotranspiration (PE) is a model-based way of measuring ET and is typically used when measuring Actual ET is infeasible. PE is estimated assuming that the moisture availability, from precipitation and soil i.e., is ample / unlimited. Since moisture availability is a central 'variable' in the computation of Actual ET, computing PE instead is a way of saying - What is the maximum water (& energy) that can be transmitted from the soil and vegetation into the atmosphere over a period of time, given the prevalent non-moisture variables such as temperature & wind speed conditions?

When Potential Evotranspiration is measured on a reference surface - for example, at a weather station nearby or at a field with an even stretch of vegetation, it is called Reference Evotranspiration (ETref).

In a real-world setting, as you'd infer, there is no guarantee that moisture availability would be ample always. Hence, Actual ET is always less than or, at best, equal to Potential ET. However, PE (reference surface) is a practical way to estimate ET at a regional scale and beyond. Its measurements are published by earth observation agencies and is the go-to metric to determine the climatic condition of a region.

USA Evotranspiration
Figure 4: Large-scale ET readings (accumulation of weather station results) can be found online. Source of USA ET above:

Evotranspiration, in its three forms as explained above, can be measured at a Local, Field, Watershed, Regional & at a Continental scale.

At local & field level, equipment such as the Pan / Weighing Lysimeter is used to measure Actual ET.

Weighing Lysimeter for Evotranspiration
Figure 5: Above ground & Below ground view of a weighing lysimeter. Source:

As the measurement area increases - typically field-scale & beyond - Remote Sensing using satellite observations is preferred to estimate Evotranspiration (both potential & actual). Besides the obvious cost-benefits as the geographic extent increases, Remote Sensing medium is preferred as it offers a synoptic view over the Area of Interest at regular time intervals.

For my study, I have used satellite observations from ESA Copernicus' Sentinel-2 (optical) & Sentinel-3 (thermal) platforms along with meteorological data from Copernicus' Climate Data Store to estimate Daily (Actual) Evotranspiration at field-scale over an agri-zone in Punjab, India.

Energy Balance Model
Figure 6: Energy Balance Model used in estimating ET. Snapshot Source: Rohit Pradhan's Deck, SAC)

This presentation deck, prepared by Mr. Rohit Pradhan from Space Applications Centre (SAC) Ahmedabad, would make for a substantive reference material in case you wish to know more on Evotranspiration and some of the mathematical models and algorithms used to derive it.

Evotranspiration is typically expressed in millimeters/unit of time (in water vapour released terms) or in megajoules (in energy released terms).



Once the fundamentals of ET are understood, it is not difficult to gauge the value of measuring and monitoring Evotranspiration. Monitoring the ever-changing relationship between precipitation (water inflow) & evotranspiration (water outflow) is beneficial for several applications, be it social, commercial or environmental in nature.

One way in which the social and environmental impact of any global-scale phenomenon can be assessed is by seeing if, and how much it could help in achieving the '17 Sustainable Development Goals (SDGs) for peace and prosperity for people and the planet', as laid down by the United Nations in 2015.

Evotranspiration is already being used / aids in workflows pertaining to:

- Irrigation planning & scheduling - supplying water to agri-zones that face moisture scarcity

- Watershed & water rights management - who should get to use the water and how much?

- Crop yield forecast - both excess & lack of moisture impacts harvest & results in food shortage

- Drought monitoring - severity of which can be determined by the interplay between E & T of ET

- Climate change impact - Warming affects Hydrologic cycle with ET being its integral component

- Drainage studies - because if water doesn't evaporate, it transfers into the soil, water table & runoff

Thus, the applications of ET mapping are "directly relevant for reaching at least two of the UN's SDGs: Goal 2 - Zero hunger, and Goal 6 - Clean water and sanitation and could also prove useful in trying to achieve other SDGs (e.g. Goal 15 - life on land)" as established by SEN-ET (Sentinels for Evotranspiration).



For this field-scale, daily (Actual) Evotranspiration estimation study over an agri-zone in Punjab (India), I have utilized the the methodology developed by the the European Space Agency-funded Sentinels for Evapotranspiration (Sen-ET) project. You can refer to Chapter 1 & Chapter 2 in their user manual which outlines -

a) the literature pertaining to measuring ET using remote sensing,

b) how leveraging the synergies between Sentinel-2 & Sentinel-3 observations allows for field-scale measurement of ET (something that was not possible before, at least on a regular basis), and

c) expands on the remote sensing methodology they've developed to estimate Daily (Actual) ET.

I have used Version 9.0.0 of SNAP Software to perform this study. The process flow of the software steps involved can be diagrammatically represented as below-

(clicking on the graphic will open an enlarged view)

SNAP Process Flow for estimating Evotranspiration
Figure 7: Process Flow for Estimating Daily (Actual) ET using Sen-ET methodology on SNAP Software

a) Having prior working knowledge of the SNAP software, b) installing the Sen-ET related plugin and its revised scripts carefully, c) having access to the user manual + its technical review document, & d) referring to the Sen-ET community forum discussions are vital aspects to be taken into consideration if you decide to do an ET study on your own for the first time. Knowing the theory and fundamentals of Evotranspiration also helps in understanding the SNAP workflow and the generated outputs better.

Besides my interest to perform the ET study over my chosen Area of Interest (AoI) and write an article for my geo-blog, another objective was to develop a singular resource which could act as a ready reckoner - a detailed, step-by-step, video tutorial - for students & practitioners. This was because currently, the information and tweaks pertaining to performing this study seamlessly is primarily in a text format scattered at a few locations on the web. From my own experience, sometimes it took Jessey and me hours to figure what went wrong and to find a working solution when we were stuck.

While I will elaborate on the processing steps and outputs pertaining to my study subsequently in this article, the video tutorial below would be the definitive guide to understand the process of estimating Daily Actual ET right from scratch, in a visual and engaging manner.

Video 1: Tutorial demonstrating the process of estimating Daily (Actual) Evotranspiration at field-scale using Remote Sensing.

The Tutorial covers the following aspects (hyperlinked):

00:05 - Case Details

00:20 - P1: Background & Setting up

00:24 - P1.1: Understanding the Area of Interest (AoI)

01:12 - P1.2: Downloading the Geographic Extent of the AoI

03:25 - P1.3: Downloading Sentinel-2 Satellite Imagery Dataset

06:35 - P1.4: Downloading Sentinel-3 Satellite Imagery Dataset

11:25 - P1.5: Set-up Intricacies

14:00 - P1.6: SNAP Software Set-up

18:27 - P2: Sentinel-2 Processing Workflow

18:30 - P2.1: Pre-Processing Graph

26:10 - P2.2: Add Elevation Graph

28:04 - P2.3: Add Landcover Graph

33:24 - P2.4: Estimating Leaf Reflectance & Transmittance

35:47 - P2.5: Estimating Fraction of Green Vegetation

38:34 - P2.6: Producing Maps of Vegetation Structural Parameters

42.47 - P2.7: Estimating Aerodynamic Roughness

44.37 - P3: Sentinel-3 Processing Workflow

45.12 - P3.1: Loading Sentinel-3 Dataset

46.02 - P3.2: Pre-Processing Graph

53:15 - P3.3: Warp to Template

55:55 - P3.4: Sharpen LST

59.51 - P4: ERA-5 Pre-Processing Workflow

59.54 - P4.1: Downloading ECMWF ERA-5 Reanalysis Data

01:07:04 - P4.2: Preparing ECMWF ERA-5 Reanalysis Data

01:10:00 - P5: Land-surface Energy Fluxes Modelling Workflow

01:10:04 - P5.1: Estimating Longwave Irradiance

01:12:07 - P5.2: Estimating Net Shortwave Radiation

01:14:32 - P5.3: Estimating Land-Surface Energy Fluxes

01:17:29 - P5.4: Estimating Daily (Actual) Evotranspiration

01:19:39 - Video Summary Note



Geographic Extent of Area of Interest Punjab
Figure 8: Geographic Extent of Area of Interest (AoI) - between Jalandhar & Ludhiana in Punjab, India- as depicted by the Daily (Actual) ET estimate as on June 04, 2022

Even before I highlight the outputs generated as part of the processing steps, let me explain why I shortlisted Punjab as my Area of Interest (AoI) for this Evotranspiration study -

a) I wanted to shortlist the agri-zone within my home country - India.

b) Punjab routinely features as the top staples producer in the country, particularly for wheat & rice.

c) Average landholding is high at 3.62 hectares/farmer, suitable from an imagery analytics perspective

d) The rice–wheat (RW) system in North-West India is facing excessive Ground Water Decline.

e) The timeline of imagery (S-3 captured on 4th June'22) was when the still-ongoing export ban on wheat was initially implemented due to unseasonal rains damaging the harvest, thereby the need to protect India's food security and keep the food-inflation in check amidst the Ukraine war which was impacting the agri-supply chain.

Slider 1: Estimated Daily (Actual) ET results over the RW agri zone on Google Earth imagery base

As depicted in the process flow in Figure 7, there are three base datasets which need to be processed to estimate daily (Actual) Evotranspiration using Remote Sensing: a) Sentinel-2 Optical Imagery, b) Sentinel-3 Thermal Imagery & c) ECMWF ERA5 meteorological dataset, in that order respectively.

(To enhance your understanding, you can refer to Figure 3 which lists the factors affecting Evotranspiration and compare it with both - the process flow in Figure 7 as well as the video tutorial or the explainer content below)

The objective of Sentinel-2 imagery processing is 'to characterize the biophysical state of the land surface at 20 m resolution' and it entails the generation of the following outputs:

- Biophysical parameters such as Leaf Area Index, Fraction of Absorbed Photosynthetically Active Radiation, Fraction of vegetation cover, Chlorophyll content in the leaf & Canopy Water Content

- Sun Zenith Angle

- Sunlight Reflectance readings as captured by the satellite sensor, at nine different spectral resolutions

- S-2 Cloud Mask

- Digital Elevation Model

- Landcover layer using ESA's Climate Change Initiative dataset

- Leaf Spectra i.e. Leaf Reflectance & transmittance readings in Visual & Near-Infrared spectrum based on plant chlorophyll and water content

- Fraction of vegetation cover which is green

- Structural parameters of vegetation such as height, height-width ratio, leaf width, fractional cover

- Surface Aerodynamic Roughness

The timeline of capture of S-2 imagery was 27th May 2022 at 05:36 am. Sharing some visuals of the S-2 processing outputs for your reference:

Sentinel 2 Reflectance
Figure 9: S-2 Processing - Reflectance output visualized in RGB (subsetted to AoI)
Sentinel-2 Leaf Area Index
Figure 10: S-2 Processing - Leaf Area Index (Biophysical Parameter) - LAI is satio of one-sided leaf area per unit ground area
Sentinel-2 Landcover Map
Figure 11: S-2 Processing - Land Cover Map with a wide variety of LC classes subsetted to the AoI
Sentinel-2 Fraction of Vegetation which is Green
Figure 12: S-2 Processing - Fraction of Vegetation cover that is green. ~66% & beyond as can be inferred from the output visual
Sentinel-2 Vegetation Height
Figure 13: S-2 Processing - Vegetation height map in metres

The objective of the Sentinel-3 LST imagery processing chain is to 'establish the bottom boundary

condition of the Land Surface Energy Model'. In simpler words, we are estimating the Land Surface Temperature as this is an important input that goes into the estimation of the energy and water vapour release. S-3 imagery has a low resolution (~ 1 km spatial resolution), which is why the processing chain also entails enhancing the resolution to match S-2's resolution which is significantly higher at 20 m (hence, suitable for field-scale studies) using the Data Mining Sharpener (DMS) Machine-learning model. The following outputs are generated as part of the S-3 processing chain -

- S-3 Land Surface Temperature (LST) band

- S-3 Cloud Mask

- S-3 Observation Geometry

- High Resolution (20 m) Observation Geometry

- Sharpened Land Surface Temperature (LST) band

The timeline of capture of S-3 imagery was 04th June 2022 at 05:26 am Sharing some visuals of the S-3 processing outputs for your reference:

Sentinel-3 Cloud Mask
Figure 14: S-3 Processing - Cloud Mask (Black Pixels as Null Data Values)
Sentinel-3 Land Surface Temperature
Figure 15: S-3 Processing - Default LST (Above) & Sharpened LST (Below). Temperature Range is measured in Kelvin

The objective of the ECMWF ERA5 Meteorological Data processing chain is to 'establish conditions which drive (e.g. air temperature) and modulate (e.g. wind speed) the energy transfer between the surface and the atmosphere' and it entails the generation of the following outputs:

- ECMWF ERA5 Meteorological data conversion from 2 m above ground to 100 m above ground, interpolated to match the timeline of S-3 capture (4th June 2022) as well as the resolution of the S-2 capture (20 m). This enhanced meteorological dataset contains the following parameters: - Air Temperature, Vapour Pressure, Air Pressure, Wind Speed, Clear Sky Solar Radiation & Average Daily Solar Irradiance.

Subsequently, this newly prepared meteorological dataset is clubbed with the some of the outputs generated in the S-2 & S-3 processing chain previously, to estimate:

- Longwave Irradiance &

- Net Shortwave Radiation (of canopy and soil respectively)

Finally, several output layers generated previously are utilized to estimate:

- Land Surface Energy Fluxes using the Two Source Energy Balance Model. There are four instantaneous fluxes which are estimated as per the S-3 capture timeline: Sensible Heat Flux, Latent Heat Flux, Ground Heat Flux & Net Surface Irradiation. Remember that Evotranspiration is also measured in energy terms besides water vapour terms - and the Latent Heat Flux represents the energy used for Evotranspiration.

- Finally, the instantaneous Latent Heat Flux i.e. energy released is converted to Daily Evotranspiration i.e. water vapour released.

Sharing some visuals of the ECMWF ERA5 processing outputs for your reference:

ERA5 Meteorological Parameters
Figure 16: ERA5 Processing - Prepared ERA5 meteorological parameters (at the time of S3 overpass - 4th June 2022 at 05.26 am)
ERA5 - Net Shortwave Radiation
Figure 17: ERA5 Processing - Contrasting Net Shortwave Radiation outputs - Canopy (above) & Soil (below)

ERA5 Latent Heat Flux
Figure 18: ERA5 Processing - Latent Heat Flux (ET in energy released terms) output measured in Watts per sq. m.

Actual Daily Evotranspiration Punjab 2022 June 03
Figure 19: Daily (Actual) Evotranspiration in mm/day at Field-scale over an agri-zone in Punjab, north of the Sutlej as on 4th June 2022, 05:26 am

The Daily (Actual) ET measurement over the AoI ranges from a minimum of 0.2 mm / day to a maximum of 10.0 mm/day with a mean of 3.8 mm/day. (Bulk of the pixel values lie between 1.3 and 6.1 mm / day).



So how does one go about interpreting my Daily Evotranspiration output? I'm sure you'll be having questions such as - Is it high / low, good / bad, improving / deteriorating?

- While I am not a hydrology expert, interpreting this single ET output would not lead to any definitive conclusion nor do I think it would be right to even attempt doing so. A consistent time-series of ET observations would help in determining the trend of evotranspiration, certainly.

That being said, you can choose to browse online and see how this output compares with research studies on ET done in India or overseas.

- The rate of Evotranspiration (water outflow) is naturally, dependent on the rate of water inflow. So the interpretation of the time-series ET measurements should be done keeping in mind the linkages and correlation with precipitation and soil moisture measurements (obtainable from local weather stations).

- The other factors that impact Evotranspiration (as detailed in Figure 3) should also be taken into consideration. For example, the meteorological data which I prepared - temperature being a component - was at the time of S-3 overpass which was at dawn - a relatively cooler part of the day even during the peak summer season in June. So the next question would be - would the algorithm interpolate the daily ET in a different way were I to select S-3 imagery captured at noon? I do not have a definitive answer. Moreover, the satellite orbit stipulates that the overpass over a region is made at specified times of the day on a regular basis so access to imagery at a preferred time is also not possible, generally speaking.

- An interesting observation which I was able to spot was when I used the same imagery to estimate Daily ET but this time over a different Area of Interest (AoI) within. Please see the output below (this AoI was located north-west to the AoI which I used for my main study).

Daily (Actual) Evotranspiration in mm/day - Punjab - 04 June 2022
Figure 20: Daily (Actual) Evotranspiration in mm/day at field-scale over an agri-zone in Punjab, located north-west to the original AoI for the same timeline

Compare the Daily ET output in Figure 20 above with that in Figure 19. What would you infer?

As you'd observe, the south-east half of Figure 20 is red in shade signifying a comparatively higher rate of evotranspiration (between 7-9 mm / day). In comparison, Figure 19 is predominantly having a lower ET range (between 3-5 mm / day). All else constant, doesn't this strike you as surprising? What could be the reason(s) behind it?

In my opinion, this could be because of the differing type of crops being grown in the respective AoI's. Remember I had mentioned earlier in this article that Punjab lies in the RW (rice & wheat) belt. As mentioned in the Factors affecting Evotranspiration visual (Figure 3), the nature of crop, its root system, agricultural practises used, growth stage of the crop - are all contributory factors to evotranspiration. Hence, I surmise that this large difference in ET values could be particularly due to the moisture consumption and retention properties of the crop being grown in the respective AoI's.

Do you agree? I hope you found this article and the video tutorial to be interesting. Your feedback and suggestions would be welcome.



Intelloc Mapping Services | Mapmyops is engaged in providing mapping solutions to organizations which facilitate operations improvement, planning & monitoring workflows. These include but are not limited to Supply Chain Design Consulting, Drone Solutions, Location Analytics & GIS Applications, Site Characterization, Remote Sensing, Security & Intelligence Infrastructure, & Polluted Water Treatment. Projects can be conducted pan-India and overseas.

Several demonstrations for these workflows are documented on our website. For your business requirements, reach out to us via email - or book a paid consultation (video meet) from the hyperlink placed at the footer of the website's landing page.


Arpit Shah

Credits: Jessey Dickson, ESA Sen-ET, DHI Gras, Sandholt ApS, SNAP | STEP Forum

bottom of page