Lessons from a remote sensing stint in Afghanistan

13 Mar 2020

When my supervisor asked me if I wanted to work on a project that would require me to travel frequently to Afghanistan, my answer was a reluctant “yes”. References to the war in Afghanistan – in the papers and on television – raced through my mind. This was four years ago.

Since then, I have worked in close collaboration with colleagues at the International Centre for Integrated Mountain Development (ICIMOD) headquarters in Kathmandu, Nepal; researchers based in the NASA SERVIR Science Coordination Office (SCO) in Alabama, USA; and the Ministry of Agriculture and Irrigation and Livestock (MAIL) in Kabul, Afghanistan. The team is developing a methodology that uses satellite imagery to estimate wheat sown areas in Afghanistan.

Afghanistan: combatting preconceived notions

For many people around the world, Afghanistan is synonymous with instability, insecurity, and war. My experiences in the country have taught me otherwise. Afghanistan is a land of beautiful landscapes, warm hospitality, rich cultures, and very friendly people.

I experienced Kabul during three different seasons – spring, summer, and winter. In the spring, the mountains in Kabul are green – beautiful flowers and orchards punctuate the landscape. In the summer, the same mountains are dry and dusty, and snow engulfs the mountains in the winter.

I got to understand the culture, landscape, history, food, and people first-hand during my visits. I also learnt how Afghanistan shares a common heritage with other South Asian countries.

The Afghan cuisine deserves special mention here. It has a unique style with a delicious diversity. It is neither too spicy nor too hot. Generous servings of dried fruits and nuts fried in plentiful oil lend rich and mellow flavours to the meats and vegetables. A self-professed vegan, I ended up trying a variety of dishes like lamb kebab, kabuli naan, chopan kebab, kofta kebab, kabuli pulao, halwaua-e-aurd-e-sujee, among many others.

However, working in Afghanistan is not a cinch. There are enormous challenges. An adverse security situation, political instability, lack of human resources, inadequate IT infrastructure, internet connectivity, and inaccessibility to field sites are common challenges. There are other challenges that are not as upfront.

Accommodating different working cultures

Unlike a Monday through Friday, 9–5 work routine, working hours in Afghanistan are from 7:30 in the morning to 4:30 in the evening. Offices remain closed on Thursdays and Fridays.

Based out of the Kathmandu office, you only get a window of three overlapping working days to collaborate in real time with colleagues in Kabul. Add in colleagues based in Alabama, USA, and you make the dynamic all the more confounding and interesting.

You have to be cognizant of different cultures, and working habits – time zones, working days, and working hours. This means being part of multiple group conversations on Skype, teleconferences, and multitudes of emails sent around coordinating such calls.

Contingencies – Plan A, Plan B, and a Plan C

Detailed planning, daily communication, and rigorous coordination are essential for running research activities in Afghanistan. Contingency plans need to be drawn up and implemented on the fly as more often than not, the strategies will fail.

We encountered several such instances during my assignment. Some of the major challenges that we encountered during the inception of the wheat mapping exercise were as follows:

  • Baseline: Several organizations are working in the agriculture and food security sector in Afghanistan. However, a baseline for wheat sown area does not exist.
  • Bandwidth: To interpret wheat fields from satellite imagery for a country like Afghanistan, we had to download and process multiple time-series satellite imagery that are over 10 terabytes in size. For this, we needed access to high bandwidth – both internet and computational.
  • Resolution: The relatively small size of the wheat fields was difficult to capture using Landsat 8 satellite imagery, which has a spatial resolution of 30 metres. A single pixel in an image represents an area on the ground that is 30 metres across. Land holdings in Afghanistan often range from a few metres to hectares across.
  • Clouds: Cloud cover obstructs optical sensors in satellites especially in the winter season. To interpret wheat fields, you need cloud free images – you either source imagery without clouds or develop composites to remove clouds.
  • Field data: Sourcing accurate field data – ground control points to train the algorithm and validate findings – was a major concern as field sites were inaccessible due to security concerns.

To address the aforementioned problems, we developed a custom research framework and made use of the Google Earth Engine (GEE), a cloud computing platform. Unlike conventional workstations, GEE provides highly scalable computational power to process satellite imagery, and a suite of other geospatial and observation tools and datasets that can be ported on-the-fly.

With the new datasets, platform and field data in place, we generated the first set of maps that showed estimates for wheat sown maps for the year 2017. This exercise is possibly a first for the Hindu Kush Himalaya, and the results showed promise. More information on the application can be accessed here.

View application: Wheat Mapping Application for Afghanistan

Concurrently, we also developed a cloud computing application to semi-automate the process using computer vision and machine-learning techniques, and a user manual for our Afghan colleagues.

In recognition of the framework design that uses cutting edge technology and the rigorous challenges our team overcame, NASA and USAID awarded our team the SERVIR Impact Team Award 2017 at the SERVIR Annual Global Exchange (SAGE) in Bilbao, Spain in 2017.

Working in Afghanistan also has its perks. I am possibly the only non-Afghan staff at ICIMOD who has visited the Quargha dam near Kabul. The reservoir and its surroundings provide majestic views and recreational facilities.

The challenging environment also improved my personal competencies – patience, self- motivation, prompt decision-making, and conflict management. Overall, I had a great experience and enjoyed working in Afghanistan and would definitely love to visit again.

Update 24 June 2020

The team published an article titled “Wheat Area Mapping in Afghanistan Based on Optical and SAR Time-Series Images in Google Earth Engine Cloud Environment” in Frontiers in Environmental Science. Dataset generated from the exercise is availalbe for download from ICIMOD’s RDS portal.

  • Kabul, the City of Peace
    Kabul, the City of Peace (Photo: Varun Tiwari)
  • The Ministry of Agriculture, Irrigation and Livestock campus in winter
    The Ministry of Agriculture, Irrigation and Livestock campus in winter (Photo: Varun Tiwari)
  • The SERVIR Hindu Kush Himalaya team at the Ministry of Agriculture, Irrigation and Livestock
    The SERVIR Hindu Kush Himalaya team at the Ministry of Agriculture, Irrigation and Livestock
  • Kabul from above
    Kabul from above
  • Varun Tiwari presenting a framework for a wheat mapping exercise at the Ministry of Agriculture, Irrigation and Livestock
    Varun Tiwari presenting a framework for a wheat mapping exercise at the Ministry of Agriculture, Irrigation and Livestock


Varun Tiwari

Remote Sensing and Geo-Information Analyst at ICIMOD
based in Kathmandu, Nepal

Varun is part of a research team comprising members from ICIMOD, NASA SERVIR SCO, and MAIL formed after a country consultation meeting in September 2016, where officials from the Ministry of Agriculture, Irrigation and Livestock (MAIL) in Afghanistan expressed the need for timely forecasts/estimation of wheat production. The study was carried out under the SERVIR Hindu Kush Himalaya (SERVIR-HKH) Initiative, a joint development initiative of USAID and NASA. ICIMOD implements the SERVIR-HKH Initiative – one of five regional hubs of the SERVIR network – in its regional member countries, prioritizing activities in Afghanistan, Bangladesh, Myanmar, Nepal and Pakistan.