GIS and Remote Sensing in Agriculture

I am interested in using GIS and remote sensing to build models for measuring agricultural land use change.  I wrote this paper for an agroecology class, during the spring of 2011.  For a pdf of the paper go here: GIS and Remote Sensing in Agricultural Land Use Planning

Remote Sensing and GIS in Land Use Monitoring and Planning:

Case Studies in Sustainable Agricultural Development

By Lauren Vanderlugt

May 30, 2011

FAIR 436N: Agroecology

Arable land is a limited resource.  As the global population rapidly increases, humans must manage their land to maximize its productivity, while sustaining its benefit for future generations.  With the use of current technology, agricultural land use planning can be done more efficiently and on a larger spatial scale than in the past.

Geographic Information Systems (GIS) and remote sensing are two technologies used to monitor regional land use change and to make decisions for future land use planning.  Although they are often used together, they are separate technologies.  GIS is a data management system designed to analyze and display spatial data (Ellis, Bentrup & Schoeneberger, 2004).  Remote Sensing is the use and manipulation of images captured by satellites and aircraft.  These technologies are used in sync when applied to land use change.

Images that are remotely sensed can be integrated into a GIS system for further analysis because the images have geospatial properties.  Each pixel in a remotely sensed image can be classified into a certain land cover type.  Land cover types can be as simple as sorting out developed land from undeveloped land, or as complicated as discerning between different types of crops.  After an image has been partitioned into the desired land cover types, it can be brought into a GIS system to be combined with other spatial data layers, such as road networks, soil maps and farm locations.

There are several benefits to using remote sensing and GIS for land cover change monitoring.  First, they can be used for monitoring areas at almost any spatial scale.  Satellite imagery is available at a pixel size as small as 1 meter, capturing great detail for a small area, or at a larger pixel size which can cover entire regions.  Although most satellite imagery is expensive, some is provided at no charge, usually from government agencies.  Widely available datasets like LANDSAT, provided by the U.S. government, have been a boon to scientists and community planners.  These technologies allow for efficient and inexpensive monitoring of land use change, at a wide range of spatial and temporal scales.  The following case studies demonstrate some of the applications for GIS and remote sensing in the field of agricultural land use/land cover (LULC) change.

A common application of GIS is to delineate areas based on their suitability for a specific crop.  A study by Wu, Liu, Dai, Li and Sun (2011) mapped out highly productive areas for citrus orchards near the city of Chongqing, China.  This study was spurred by the Three Gorges dam, which will displace at least 1.3 million people, including over 85% of Chongqing’s population.  This research assisted displaced people by locating areas for resettlement that would be suitable for citrus production, an important cash crop in the local economy.

The research consisted of two phases: (1) make an inventory of current citrus growing conditions including terrain, soil, and climate properties and (2) identify potential cultivation areas for citrus at a regional scale.

The soil properties included soil type, pH, and soil organic matter.  Terrain attributes included aspect, slope, altitude and water sources.  Climate data consisted of average annual precipitation, temperature and sunshine.  Like all plants, citrus plants are more productive when they are in the optimal conditions.  The researchers did field tests on 50 representative orchards to discern those optimal conditions.  For example, they found that suitable soil pH levels were between 5.5 and 7.5 and the optimal temperature range was between 13 and 37 degrees Celsius.

Next, the researchers collected spatial data for the resettlement area.  They collected a layer of data for each of the soil, terrain and climate variables.  Using GIS software, they mapped out areas that fit the requirements for optimal conditions of each variable.  The resulting map delineated areas where conditions were ideal for citrus growing.

The information produced by this study provides a tool for sustainable agricultural planning.  Productivity is increased by planting orchards on areas with optimal conditions.  This practice makes for a sustainable use of resources.

Bandyopadhyay, Jaiswal, Hegde, and Jayaraman (2009) approached land use planning in a similar way.  In this study, a watershed in India was divided into zones based on their suitability for crop production.  The objective of the researchers was to evaluate land parcels for their inherent capability to support and sustain agricultural production.  In the previous example, researchers were planning for primary development on resettlement land.  This study differs because the researchers were evaluating land that was already under human use.

The researchers developed a land suitability index, within the context of a pre-existing land use pattern.  This research was prompted by the environmental conditions and agricultural practices occurring in the watershed.  The study area is characterized as a rain-fed, semi-arid to arid dryland, where soil and water must be precisely managed for sustainability.

The researchers created land suitability maps from five parameters: LULC, soil type, organic matter, soil depth and slope.  They created a GIS model which weighted each of the five parameters according to their importance for optimized productivity and aggregated them into a single number in an index.  The result was a map of the area with parcels of land rated by the land suitability index.  The advantage of creating a model is that data for other watersheds can be entered into the model to create additional land suitability maps.  The researchers used the land suitability maps and water supply data to formulate recommendations for the development of community action plans.  The action plans would assist planners in their goal of using resources in such a manner that the productivity potential of the watershed was optimized.

Action plans included techniques for sustainable farming, based on knowledge acquired through the land suitability analysis and existing LULC in the study area. They suggested that areas where land potential was in the “good” to “moderate” range should be rotated between pure agricultural crops, such as ragi, maize and goundnut, and leguminous crops, like cowpea and greengram.  They also recommended that soil and water conservation measures be implemented, such as grass waterways, percolation ponds and check dams. They also proposed that land with an “average” rating should be brought under agro-pastoral cropping systems and that “poor” rated land might be used for silvi-pastoral or forest plantation systems.  Although the study area was already developed, this research can direct future changes to the land use of the watershed.   The next example will expand on the idea of soil and water conservation measures.

Many agricultural areas are situated near water sources for irrigation purposes.  However, agricultural activity is often a source of deteriorated stream health due to pesticide runoff, increased nutrient influx and overheating.  These problems can be alleviated by expanding native riparian vegetation zones around streams.  Native vegetation along streams can improve water quality by filtering contaminates from runoff, slowing the influx of nutrients, and providing shade.

GIS can be used to effectively improve buffer zones along streams.  Due to resource limitations, farmers and ecologists cannot restore all buffer zones, so they must choose which areas to restore.  GIS can be an aid in these decisions, allowing the user to map out areas where buffer zones will be most effective.  Research by Tomer et al. (2009) explains how a buffer area can be more effective in some areas over others.  One way to prioritize locations for buffer establishments is to use topographic and stream flow data layers to identify locations where buffers are most likely to intercept water moving towards streams.  Another technique uses soil data to rank soil map units by how effectively a buffer placed on those units would trap sediment.  Ecologists and farmers can then direct their resources towards developing riparian buffers in areas where the buffer will most effectively improve the health of the stream.  Research on fringe zones, the land where agricultural development meets natural areas, is a topic of concern to scientists across the globe.

A study by Asner, Townsend, Bustamantes, Nardoto, and Olander (2004) was conducted on the fringes of human development in the Amazon Basin.  The goal of the study was to better understand biogeochemical dynamics in forest-to-pasture conversion zones.  This research was motivated by reports of pasture degradation, occurring after about 5-15 years of pasture use.  As researchers noted, there is a scarcity of data on biogeochemical dynamics in the Amazon, which is needed for spatial-temporal research on pasture production and degradation.

Remote sensing has provided an efficient way to recover data for large areas of land.  The satellites that capture images actually capture information in several different bands of the electro-magnetic spectrum.  Some earth features reflect differently in each band.  For example, vegetation reflects highly in the near-infrared band of the spectrum.  Varying spectral properties of earth features allow scientists to use remotely sensed images to detect land cover types over large regions.  In this study, remotely sensed images were used to detect photosynthetic vegetation (PV) and non-photosynthetic vegetation (NPV).  Researchers compared these measurements with field data on nutrient levels and plant biomass to determine if there was a correlation between the two.

Using a previously developed algorithmic spectral-mixture analysis, researchers calculated estimates of PV, NPV and bare soil covers from LANDSAT imagery.  In the field, measurements for plant area index (PAI), live:senescent vegetation ratio, above ground biomass (AGB), plant canopy height, and specific leaf area (SLA) were collected.  They also field-tested for soil organic carbon (C), nitrogen (N), phosphorus (P) and calcium (Ca).  The remotely sensed estimates were compared to the field-based measurements.  The researchers found that both PV and NPV were highly correlated with field LAI and NPVAI measurements. LAI and NPVAI estimates were indicators of AGB, which in turn, was an indicator of pasture condition.  They also found that estimates of PV and NPV were highly correlated with soil organic C and N stocks, and were short-term indicators of N, P, and Ca.  The researchers concluded that remote sensing may be a viable approach to estimate pasture condition, carbon storage, and biogeochemical status at a regional level.

In this application, remote sensing was used to assess the impact of land use change on soil conditions.  Monitoring the health of developed land is an integral part of land management.  Evidence of pasture degradation can be an important step towards policy that supports a sustainable economy, especially in places with booming population growth.  In this study, remote sensing proved to be a cost-efficient alternative to field testing.

In another Amazon-based study, researchers measured the spread of human activity in the rainforest and its impact on fringe zones.  A study by Mena, Walsh, Frizzelle, Xiaozheng, and Malanson (2011) details the design of a spatial simulation model that was developed to assess the drivers of LULC change at the interface of frontier and forest in the Northern Ecuadorian Amazon (NEA).  Data used for the model included results from a socio-economic and demographic survey of colonist households conducted in 1990 and 1999, a satellite image of LULC change for 1973-2007, and GIS layers of infrastructure, geographic access and resource endowments of farms.  The researchers created a model that addressed how decision-making at the household level drives regional land use change.

The NEA has undergone an incredible amount of change due to colonization and resource extraction.  During the extraction process, oil companies built roads into previously isolated areas, thus exposing them to settlement.  Since the discovery of petroleum in the NEA in 1964, the population in this area has grown twice as fast as the rest of Ecuador.  This sudden increase of human activity has made the region an optimal place for studying relationships between human and environmental systems.

Settlers converted much of the land for agricultural purposes, expanding beyond subsistence crops to cash crops and cattle ranches.  With expansion came further in-migration and fragmentation of plots, increased off-farm employment, additional wage labor, growing market contact, and increased petroleum production.  All of these factors contribute to LULC change.  The goal of the researchers is to understand the dynamics of a complex LULC system given a set of rules and initial conditions.

The model contains four different modules: Initialization, Demography, Migration and Agriculture.  Within the model, there are several agents that undergo change over time.  The agents are the farmer, his family, a cell representing an area of land, and parcels within that cell.  The model uses a set of algorithms, based on social, economic and landscape attributes of survey data, satellite images and GIS layers, to determine the actions of farmers and their families.  The actions of the farmers determine the outcome of the LULC class for each parcel of land.  The possible LULC types are primary forest, successional vegetation, pasture, subsistence agriculture, commercial agriculture, and barren/urban.  The model simulates LULC change for over 2000 farms, over a given period of time, to develop a picture of LULC for the entire region.

The model was created using several assumptions; some of those are: (1) farmers make decisions based on immediate, current information, as well as prior knowledge (2) a farmer will always convert as much land as possible based upon labor and available resources (3) the off-farm labor pool always has sufficient labor to satisfy the needs of the farmer (4) all farms have high geographic accessibility to the NEA (5) a farmer will attempt to implement land change for profit maximization.

The model begins with the Initialization module, which creates the agents: the farmer, household members and their land.  Next, the Demography module uses algorithms based on regional rates of aging, fertility, mortality, marriage and to determine the household “roster.”  The roster includes all the members of the household, along with their age, gender, and marital status.  The module also implements historic data on prices of coffee, cacao, and cattle to determine the profitability of the farm.  If the farm has positive assets, meaning the family experienced a profitable year, the Agriculture module is used.  This module calculates any changes in land use on the farm.  Land use change could result from acquisition of more land, division of land for other family members, or change in the cropping system of the land.  If the family experiences a fruitless year, the possibility of out-migration is considered in the Migration module.  The entire process is repeated for every year, for each farm.  The model uses many sub-modules within each of the four main modules.

Many more factors than I described here were integrated into the model. Some of those include the amount of subsistence agriculture on the farm compared to commercial, the influence of a farmer’s neighbors on his decisions, and remittances that might be sent to a farm from family members that have migrated in-town.  Overall, the model seeks to consider all factors that would affect a farmer’s decisions, which ultimately have the greatest effect on regional land use change.

This model can contribute to research in sustainable agriculture, by providing a better understanding of the factors that drive land use change.  This model could provide the scientific evidence used to implement policy changes that would provide incentives for farmers to manage their land in a sustainable fashion.  This type of complex model is a unique example of interdisciplinary research.  Land use change is driven by the same factors that stimulate economic change, yet it affects both the human population and the natural environment, thus the need for an interdisciplinary approach.

GIS and remote sensing are tools for general monitoring and planning of agricultural areas.  It is up to the users of those tools to decide if they will use them for sustainable development.  However, due to the limitations of our resources and booming global population, sustainable development will soon be a requirement, rather than a choice.  The previous examples are evidence that GIS and remote sensing can easily be integrated into sustainable development.

Literature Cited

Asner, G.P., Townsend, A.R., Bustamantes, M.M.C., Nardoto, G.B., Olander, L.P. 2004. Pasture degredation in the central Amazon: linking changes in carbon and nutrient cycling with remote sensing.  Global Change Biology. 10:844-862.

Bandyopadhyay, S., Jaiswal, R.K., Hegde, V.S., Jayaraman, V. 2009. Assessment of land suitability potentials for agriculture using a remote sensing and GIS based approach. International Journal of Remote Sensing. 30: 879-895.

Ellis, E.A., Bentrup, G., Schoeneberger, M.M. 2004. Computer-based tools for decision support in agroforestry: Current state and future needs. Agroforestry Systems. 61:401-421.

Mena, C.F., Walsh, S.J., Frizzelle, B.G., Xiaozheng, Y., Malanson, G.P. 2011. Land use change on household farms in the Ecuadorian Amazon: Design and implementation of an agent-based model. Applied Geography. 31:210-222.

Tomer, M.D., Dosskey, M.G., Burkart, M.R., James, D.E., Helmers, M.J., Eisenhauer, D.E. 2009. Methods to prioritize placement of riparian buffers for improved water quality. Agroforestry Systems. 75:17-25.

Wu, W., Liu, H.B., Dai, H.L., Li, W., Sun, P.S. 2011. The management and planning of citrus orchards at a regional scale with GIS. Precision Agriculture. 12:44-54.

Discussion

No comments yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

  • No categories

Archives

%d bloggers like this: