Manual Testing of the LandPKS App for Quality Assurance

Huong Tran is the Quality Assurance Coordinator for LandPKS. She is based in Las Cruces, NM and has been working with the project since Jan 2016. Huong’s primary roles with the project include creating test plans and test scenarios, working with software developers to ensure quality standards, and performing manual tests for various products of LandPKS. She enjoys catching LandPKS bugs before users do and is glad that she can contribute to provide confidence towards software quality.

Scenario testing is done to make sure that the end to end functioning of the LandPKS app and all the process flows are working well. In scenario testing, the tester puts herself in the end user’s shoes and figures out the real world scenarios or use cases which can be performed on the app by the end user. Scenario testing helps testers to explore how the app will work in the hands of an end user. This is very important for catching bugs in the LandPKS app and designing the app with users’ needs in mind.

Even though the main goal of testing is to be able to detect and catch many of the bugs, the automated tools cannot test for visual considerations like gestures, image color or font size. However, the manual testing that Huong does can judge these types of app features. Manual testing can also test the User Experience and User Interface. Any bugs in data connection and/or slope measurement, which are two critical functionalities of the LandPKS app, can also be caught with manual testing

In light of the fact that access to an Internet connection in Africa is still limited, Huong has prioritized her testing efforts and focused on the reliability of the apps in an unstable data connection environment. She has tried to replicate the scenarios that a common end user might face when s/he is in the field collecting data and using the app. Her tests are done to ensure that the GPS is still working well, the app is behaving as expected, and that entered data is saved when there is no network or data connection.


Introducing Thanh: A Longtime LandPKS Team Member

Thanh Nguyen is a PhD candidate in the Computer Science Department at New Mexico State University. Thanh has been working on the LandPKS project for several years now and has taken a critical role in the development of the LandPKS app.  During his time with LandPKS, he has worked on the following:

  1. Developed Data Analytics System and Prediction Model that applied Machine Learning Algorithms to get as much knowledge as possible from the soil profile, weather, and water data to build models for analyzing soil potential.
  2. Developed the LandPKS Database and LandPKS API System that allowed developers and users to interact with and access LandPKS data (LandInfo, LandCover, etc). LandPKS Database and API System are hosted in Google Cloud Platform and people can visit:
  3. Developed Web Data Portal to allow users to have the ability to access and download LandPKS data in Web Browser. In addition, the Data Portal provides tools that analyze and displays the user data.
  4. Developed LandPKS mobile application that allows users to collect and interact with LandPKS data in Android and iOS.
  5. Developed Big-Data processing module using Map-Reduce (Hadoop) to create accessible climate data and soil profiles for all locations in the world

thanh blog

Thanh graduated with his Masters of Computer Science at James Cook University in Australia. His thesis, entitled Data Mining in Internet Banking, is currently being used in a number of Asian banks.  His research interests include Data Mining and Knowledge Discovery (classification, clustering, association rules and prediction models), Artificial Intelligence (knowledge representation and reasoning, planning, logic programming, answer set programming and Web semantics – Services Composition), Machine Learning and Collective Intelligence (recommendation system, discovering groups, searching and ranking, collaborative filtering, document filtering, generative modelling, advanced classification, etc.) and Big-Data processing.  His primary research now focuses on Automation Web Services Composition in Web Semantics. He is developing a completed end-to-end AI system to collect requirements from users in Natural Language and explore workflows that can satisfy the users requirements automatically. After the workflow is achieved, our system is able to execute each Web Service component in workflow sequence in order to achieve the goal.


Introducing Brent and the LandPKS Application Programming Interface (API) System

Brent, an undergraduate student at New Mexico State University, has been working with LandPKS on the API (Application Programming Interface) and an automated build and testing suite. Whenever a change is made to any LandPKS app, the automation suite builds the newest version with the changes that were just made. After building these changes, the suite runs tests on the application that was changed to catch bugs or mistakes that may occur. The suite launch emulators for all the main operating systems and goes through input process. After inputting the data the suite then validates results from LandPKS models and creates tables and stores results in the tables. This system allows changes and bugs to be tested constantly and the results stored for tracking purposes.

The LandPKS API on the backend is a single insertion and retrieval point. This allows anyone to send and receive data from a single point. This API integrates all the insertions and updates the plot areas whenever data is received. The API also integrates our models into the data as it received. The API allows all this multidirectional data flow to be unnoticeable and easily manipulated to the end user. The API also allows anyone to use our data in other applications with simple http requests. When a request for a site is received, the API retrieves all the relevant data and returns the results via the same query. Integrated into the LandPKS API is “API Explorer” which allows anyone to become acquainted with the “API” by showing the types of requests that can be passed as well as the responses received.

Brent Barnett is a dual undergrad student in the Chemical Engineering Department and the Computer Science Department at New Mexico State University. He is a transfer student from Austin, Texas. He has worked with automation and database engineering for six years. Prior to joining Jornada he worked for Versasuite and IBM as a software and database engineer respectively. His primary roles are build/test automation and LPKS API development. His interests include database development, prediction algorithms, process design and automation.

query images_1

Brent Barnett – API Data Manager

Land-Potential and Soil Texture

The knowledge engine of LandPKS supports land use planning, land restoration, future agricultural scenarios, climate change adaption, and conservation programs. However, a critical first step in using our integrated suite of smartphone applications includes the evaluation of soil and vegetation properties. As soil serves as the media for growth for all kinds of plants, identification of soil properties is vital for land managers, policy makers, and researchers in order for them to assess land potential.


Soil texture is considered one of the soil’s most important properties, influencing nearly all soil processes and functions. Soil texture is defined by the relative fractions of sand, silt, and clay-sized particles in a soil sample. Soil is often divided into 12 soil texture classes, allowing for communication of soil type amongst land resource specialists. Each particle type has benefits to plant growth, yet sometimes can be unfavorable when only one soil class (all sand, for example) makes up the majority of the soil sample. For example, clay holds water well and is usually fertile. However, clays swell when they get wet (limiting the water available to plant roots), and harden when dry (becoming difficult to manage). Deep sands drain easily and do not hold water effectively. Silt-sized grains retain water and nutrients, but can easily become water-logged and prevent movement of water, air, and roots throughout the soil profile. While most soil types can be managed, often Loam is considered the most desirable for plant growth because Loam contains equal parts of sand, silt, and clay. Different particle sizes allow for air, water, and roots to easily move through the soil, with Loam having enough sand to drain well, yet also enough clay and silt to hold onto water and nutrients.


The LandInfo module of LandPKS walks users through estimation of texture by probing and working the soil. Users are asked to take a sample and test it for grittiness (sand), smoothness (silt), and stickiness (clay). Currently, the LandInfo module streamlines steps from a soil texture-by-feel flow chart which tests the relative fraction of sand, silt, and clay. However, through feedback from our trainings we have learned that often the user questions accuracy of their own texture-by-feel estimates. Our next steps to help address this issue include: (1) evaluating the accuracy of these texture-by-feel estimates, and (2) improving our decision support tools to allow users additional manipulative tests shown to differentiate between texture classes. Look for the results of these steps in the near future!


LandPKS for Soil Identification: Using Soil Texture and Color

Soil is one of the most important factors that control crop yields and land potentials. There are many possible soils with different properties within a given location. How to correctly identify soils and the subsequent soil properties is critically important for farmers, natural resource managers, policy makers, and scientists to make decisions and predictions regarding land suitability, productivity, profitability, and sustainability.

In our effort to help make soil identification in the field easier, LandPKS has two team members, Zhaosheng Fan and Samira Pakravan, who are working on improving our current LandInfo module, as well as developing a completely new module, SoilColor. Fan and Samira are currently focused on, 1) developing algorithms that can be used to identify correct soils with spatial location (latitude and longitude) and other easily-measured field observations input by LandPKS users (e.g., soil texture by depth), and 2) developing a SoilColor module that can be used to measure soil color in the field with a smartphone camera. Soil color is one of the most important attributes of the soil which provides useful information about many other significant soil properties, such as soil organic carbon content. The SoilColor module will use smartphone camera images of soil samples to measure the soil color. Once complete, LandPKS users will be able to identify soil color without the use of expensive soil color books.

Soil color app photos

Further, the two efforts mentioned above are tied together – the measured soil color with the SoilColor module can, in turn, be used by the soil-identification algorithms to further improve the accuracy of the identified soils. Once the correct soils are identified, the corresponding soil properties will be used to drive crop models to simulate crop yields, erosion, and crop-failure risks. Please do look for the release of SoilColor in the near future!

Zhaosheng Fan – Postdoctoral Researcher
Samira Pakravan – Soil Color Developer

The Role of LandPKS in Land Use Planning

In Tanzania, the Land Potential Knowledge System (LandPKS; has been working with the National Land Use Planning Commission as well as USAID’s Land Tenure Assistance (LTA) Project to assist in effective land use planning and land tenure efforts.  LandPKS is a mobile app that helps users identify their soil, and the potential productivity and long term agricultural sustainability of that soil.  By identifying areas with sustainable agricultural potential, land use planners can integrate biophysical assessments into their participatory land use planning process.  The LTA project is helping to revise village-level land use plans. Once land use planning is complete, LTA is using a mobile technology called MAST (Mobile Application to Secure Tenure) to assist village members attain a CCRO (Certificate of Customary Right of Occupancy) for their farms and properties. By integrating LandPKS soil information into their land-use plans, LTA and other land use planners can make better decisions about which areas within a village are suitable for agriculture, and which are not. Focusing agricultural growth on soils suitable for sustainable agriculture not only increases farmer revenues and productivity, but saves other village land for other, less intensive uses such as grazing areas or forest reserves.

At the national-level, Dr. Stephen Nindi, Director General of the National Land Use Planning Commission, is working with the LandPKS team to implement LandPKS tools in the future for the national land use planning process. Tanzania currently uses a 6-step process for participatory land use management, involving community members and stakeholders at every step. On the biophysical side of land use planning process, Tanzania draws from the seven land capabilities classes. Categorizing the land into these seven classes helps planners to determine which livelihood activities are sustainable in which areas. For example, land capability class one refers to land that is suitable for all land uses with normal land management practices, such as flat, well drained and fertile land. LandPKS could play a key role in helping land use planners classify land into these seven classes based on soil texture, soil water holding capacity, potential erosion risk, and potential productivity. While soil texture is not the only important soil characteristic, it can be a critical predictor of a lands potential and its degradation risk. Implementing LandPKS in the land use planning process will simply, and cheaply, help the National Land Use Planning Commission include more biophysical data into their land use planning process.

Therefore, LandPKS has a role not only in improving sustainable land management for farmers and pastoralists, but also on a larger scale through the land use planning process. Matching appropriate land uses to their proper soil types can increase productivity and decrease land degradation, important goals for long-term environmental sustainability. For more information, please write to

Nyamihuu, Tanzania – Comparing 3 LandPKS Plots

LandPKS Modeling for Agricultural Sustainability: Meet our Modelers

We are lucky at LandPKS to have a wonderful team of people who work on various aspects of LandPKS.  Behind the scenes, we have two dedicated modelers who are working to take LandPKS input and model future agricultural scenarios to help LandPKS users make more sustainable land management decisions.  While these types of outputs are not currently available on LandPKS, we hope that in the near future such modeled future scenarios will be delivered to LandPKS users on their phones.

1model outputs

Dr. Won Seok Jang is a hydrologist and modeler for LandPKS based at the University of Colorado Boulder, in the Sustainability Innovation Lab at Colorado (SILC).  Won Seok’s primary roles include developing a modeling framework and assessing the effect of soil degradation on crop productivity and local/regional/global scale hydrological modeling for climate change impact assessment.  He currently uses EPIC (Environmental Policy Integrated Climate) model to estimate potential crop yield and soil erosion and is developing an EPIC parallel computing framework for global modeling with big data.  The goal of his research is to explore and measure the impact of climate change on food production worldwide, to develop multi-objective optimization of crop yield and minimization of soil erosion initially in Eastern Africa with a plan to expand to the United States and other global locations in the future.

2tegenu and won

Dr. Tegenu Engda is a postdoctoral researcher, also located at the University of Colorado Boulder, in SILC. His primary role is to understand Soil-Plant-Water interactions under different environmental conditions. He mainly uses the EPIC model to investigate factors affecting crop yield trends including hydrologic processes, erosion and management practices in the case of East Africa. Currently, he is working on sensitivity analysis of EPIC input soil properties to better understand yield estimates across soil types.  Tegenu is also working on EPIC model calibration and nutrient dynamics exploration, as well as incorporation of local knowledge to improve EPIC yield calculations.


A Creative Collaboration for Developing LandPKS Trainings

Last week, Amy Quandt (LandPKS Global Coordinator) and Michaela Buenemann (Associate Professor of Geography, New Mexico State University) traveled to Nairobi, Kenya to help develop a comprehensive one-day LandPKS training.  Amy and Michaela worked in partnership with the Regional Centre for Mapping of Resources for Development (RCMRD), a LandPKS collaborator.  RCMRD is located in Nairobi, Kenya and works as a leading regional center for GIS and remote sensing technologies and trainings.  Lillian Ndungu at RCMRD has been a part of the LandPKS team and a major proponent of LandPKS in Kenya.  Antony Ndubi from RCMRD also joined the training development team.

1Core development team

The training development team spent Monday through Wednesday (September 4th-6th) developing a comprehensive one-day LandPKS training. The training is meant to be a general overview of LandPKS, explain how to use the apps, and help users understand how to analyze and interpret the information that they receive from LandPKS. Importantly, the training also focuses on the potential uses and applications of LandPKS for rangelands, agriculture, and biodiversity conservation. The training materials include a step-by-step outline of the course for the instructor, presentations, hand-outs, activities, worksheets, and assessments. Anyone can pick up the training materials and teach this one-day training. In order to test out the training materials, on Thursday (Sept 7th) the training development team conducted a training of the trainers or ToT. For the ToT, the training development team worked with instructors at RCMRD’s training college and did a ‘dry run’ of the one-day training (see photos). The instructors enjoyed learning about the LandPKS tools and also provided feedback on the training materials.



RCMRD plans to integrate this training into their own curriculum and courses, while the LandPKS team will also utilize from these training materials to conduct trainings with various partners and LandPKS users in the future. If you have questions or are interested in these training materials please e-mail

4analyzing output

5Practice training group

LandPKS in Ethiopia

Since 2015, the US Forest Service has been using the Land Potential Knowledge System (LandPKS) to collect monitoring data in Ethiopia to develop a baseline for rangeland condition, and to evaluate the efficacy of rangeland management treatments in an adaptive management approach. As part of the USAID Pastoral Resilient Improved Market Expansion (PRIME) project, US Forest Service (USFS) ecologists Tom DeMeo and Sabine Mellman-Brown have been working with CARE Ethiopia Rangeland Specialists Gudina and Beressa Edessa to track local vegetation changes over time, and to record grazing effects and soil erosion trends. USFS has been conducted monitoring work primarily in the Borana, Guji and Afar regions of Ethiopia, using both the LandPKS LandInfo and LandCover applications for electronic data collection and online cloud storage. To date this work has resulted in 149 monitoring plots.

Guest Post Authored By: Allison Holt – Forest Service

Local perceptions of land potential

How do local perceptions of land potential compare with soil types and textures measured with LandPKS?

That is what we set out to figure out in the rural village of Lyamgungwe, Tanzania. We asked the village government officials and elders to identify two locations within their village: one with a soil that is highly productive and does well growing maize, and one with a soil that is not very productive and where farmers have a hard time growing maize. The results were quite dramatic and the local perceptions were supported by LandPKS. The locally perceived productive soil was a Sandy Clay Loam until about 20cm depth where it turns into a Sandy Clay, and then a Silty Clay after 50cm. The locally perceived unproductive soil was a Clay for the first 20cm, then a Sandy Clay Loam, and a Loamy Sand after 50cm.


LandPKS results showed that the first soil can hold a lot more water (X vs Y cm in the top 70cm). Future LandPKS interpretations would also indicate that the first soil also has a higher potential infiltration, so it should be able to capture more water before it runs off. This type of information can be used to decide which land to – and not to – cultivate, which can help with land use planning for both agriculture and conservation.