Remote sensing of land use/cover changes in South and Southeast Asian Countries

Introduction

Land-Use/Cover Changes (LU/CC) are occurring rapidly in South/Southeast Asian (S/SEA) countries, generally associated with population growth, economic development and competing demands for land. In the region, the most common LU/CC changes include urban expansion, agricultural land loss, land abandonment, deforestation, logging, reforestation, agricultural expansion, etc. Specific to South Asia, forest cover has been increasing in countries like India, Nepal, and Bhutan due to sustainable afforestation measures; whereas, large-scale deforestation in Southeast Asian countries is continuing, for example, due to oil palm plantation expansion driven by the international market demand in Malaysia and Indonesia. Small-scale deforestation in most of the Southeast Asian countries is due to slash-and-burn agriculture by indigenous people, driven by poverty and population growth. Logging activities are common in Myanmar and Cambodia. In terms of urbanization, South and Southeast Asian countries contain 23 megacities, each with more than 10 million people. Megacities in these countries started as urban clusters which grow rapidly, merging into conurbations. Rapid urbanization is driving agricultural land loss, and agricultural intensification has been increasing due to less availability of land for growing food crops as, for example, in India, Vietnam, and Thailand. The drivers of LU/CC vary widely in the region and include such factors as land tenure, local economic development, government policies, inappropriate land management, land speculation, improved road networks, etc. Also, variability in the weather, climate, and socioeconomic factors drive LU/CC resulting in disruptions of biogeochemical cycles, radiation and surface energy balance.

Documenting LU/CC and the associated impacts gains significance in S/SEA as the results can be useful for informing policy and improved land management. Remote sensing, due to its multi-temporal, multi-spectral, synoptic and repetitive coverage capabilities is highly useful to monitor and document LU/CC and the associated impacts. The current special issue articles were part of the output from the LU/CC meeting organized in Chiang Mai, Thailand, 17–19th, July 2017 as part of the South/Southeast Asia Research Initiative (SARI), which is a NASA Land-Cover/Land-Use Change program-funded activity. The goal of the SARI is to promote and support innovative regional research, education, and capacity building, involving state-of-the-art remote sensing, natural sciences, engineering, and social sciences to enrich LCLUC science in South/Southeast Asia. SARI has been organizing focused thematic meetings and workshops in different countries of the region with the primary objective of strengthening regional science involving both developed and developing country scientists. A summary of the published papers is provided below:

• 1. Mapping Mangrove Forests Using Spectral Indices

Mangrove ecosystems are some of the important intertidal ecosystems in tropical and sub-tropical regions of the world. In addition to providing valuable renewable wood resources, mangroves are a valuable ecological and economic resource as they provide nursery and breeding sites for birds, fish, crustaceans, reptiles, and mammals, and provide protection against the coastal erosion. In S/SEA countries, mangrove ecosystems are fast declining due to overexploitation and coastal development. Thus, accurate characterization and monitoring of mangrove extent and area are essential to address economic, ecological, and societal benefit aspects. In this special issue, Ramdani, Rahman, and Giri (2018) use Landsat data in combination with the novel vegetation indices to discriminate mangrove species in South Sulawesi and West Java, Indonesia. The novel indices include spectral greenness, wetness, and brightness. These indices were transformed using principal component analysis to create a three-layered image on the greenness, wetness, and brightness to capture dynamic mangrove features. Using these indices, authors report improved accuracies of 92% and 89% with class precision of 86.2% and 84.2% for Bruguiera- and Rhizophora-dominated mangrove forests. The results are quite promising, and the new indices proposed in the study can be used elsewhere for mangrove delineation.

• 2. Urban Pattern Detection using Optimization and Neural Network Algorithms

Several countries in S/SEA are undergoing a transition from a predominantly rural to urbanized societies. The size, scale, and shape of cities vary and understanding how cities develop requires an integrated approach combining biophysical and socioeconomic data. Remote sensing can provide valuable information on urban density, compactness, and sprawl, which can be further related to socioeconomics to address sustainability. In this special issue, Bui et al. (2018) describe a novel hybrid algorithm entitled GMNN that combines a Grasshopper Optimization Algorithm (GOA) with Multiple-class Neural Network (MNN) for urban pattern characterization in Hanoi, Vietnam. Authors compare the GMNN algorithm with other approaches such as Support Vector Machines, Decision Trees, and Markov Neural Networks. They found that the GMNN algorithm outperforms other methods with an 87% overall accuracy. The study highlights the importance of metaheuristic algorithms for object-based urban area mapping useful for LU/CC studies.

• 3. Forest Cover and Temperature Change Monitoring Using Optical Data

Mapping and monitoring of forests are essential for forest management purposes. Forests are an important source of food, fuel and shelter. They provide a variety of ecosystem services including climate regulation, carbon sequestration, soil erosion control, maintaining air quality, supporting biodiversity, recreation and tourism. Forests influence local and global temperatures and the flow of heat through transpiration, which is a cooling process. Thus, declining forest cover can result in an increase in surface temperature. In a study by Khalid et al. (2018) use Landsat Thematic Mapper data for 1992, 2000 and 2011 to monitor forests using the short-wave infrared band and surface temperature fluctuations using the thermal infrared band observations over the Margilla Hills, Islamabad, Pakistan. Authors use a hybrid maximum-likelihood algorithm for land-cover classification and change detection and retrieve land surface temperature for different years. Authors report temperature differences sometimes reaching nearly 30 C due to land transitions from scrub forests to soil and water to agriculture (1992–2000) and also as a result of conversion of low vegetation to agriculture, scrub to pine and low vegetation to settlements (2000–2011). In the paper, authors describe land-use transitions and their impact on the surface temperature over a period of time.

• 4. Land-Cover Change Prediction Using Landsat data and a Cellular Automata Markov Chain Model

Modeling and predicting the future land-use changes is an important area of research as the results can aid in land-use planning, management, and environmental impact assessment. Remote sensing data can be effectively used to map land-use transitions, which can be integrated with simulation models to predict land-cover changes. In this special issue, Yulianto, Maulana, and Khomarudin (2018) provide such a modeling study in one of the watersheds in West Java, Indonesia. Authors used Landsat data for 1990, 1996, 2003 and 2009 to map land-cover changes and then integrate the results into a Cellular Automata and Markov Chain model for predicting future scenarios till the year 2050. Prediction results suggest increasing built-up areas, secondary forest and mixed garden, dry land farming, plantation, and wetland agriculture, with an average area of 173.7 ha/year, 3.4 ha/year, 155.5 ha/year, 37 ha/year, and 19.2 ha/year, respectively. They project a decline in primary forest and water bodies with an average area of 388.6 ha/year and 0.2 ha/year, respectively in the study area. The case study showcases the integration of remote sensing data and simulation modeling for addressing LU/CC in the study region. They also emphasize the need for integration of socioeconomic data for land-use change scenario studies, which was left beyond the scope of this study.

• 5. Monitoring Ground Water Irrigation Using Synthetic Aperture Radar Data

Use of optical data for LU/CC studies in tropical countries is limited due to persistent cloud cover. Synthetic Aperture Radar (SAR) data have advantages as they can penetrate clouds with multi-frequency and multi-polarization capabilities which can be exploited for a wide variety of LU/LCC studies. Most importantly, the use of SAR data for crop mapping and monitoring, including water resources is gaining significance as the data are sensitive to surface moisture content and temporal changes of agricultural practices. Sharma et al. (2019) use RADARSAT SAR data for mapping irrigated croplands in two different cropping seasons in southern India. Authors use both RADARSAT backscattering coefficients as well as a variety of polarimetric indices together with Support Vector Machines and report 92.7% overall classification and 93.9% kappa accuracies. Authors report that 13.4% of croplands in the study area rely on groundwater irrigation and have a double-crop rotation. The study demonstrates the potential of RADARSAT data for crop type mapping, including irrigation aspects.

Overall, the papers published as a part of this special issue cover a wide range of topics useful for LU/CC research. The articles generate interest in remote sensing, agriculture, ecological and environmental management aspects and highlight the importance of synergistic approaches. We hope that the papers published will trigger more novel ideas and shape the future research direction in remote sensing of LU/CC in SARI countries and elsewhere.