Mapping abandoned cropland in tropical/subtropical monsoon areas with multiple crop maturity patterns

Картирование заброшенных сельскохозяйственных угодий в тропических и субтропических муссонных районах с различными сроками созревания культур
Guilin Liu, Yuecheng Li, Chen Yang, Yongquan Lu, Desheng Jiang, Aojie Xu, Yuqi Zhong, He Yin
2024-02-08

LandTrendr algorithmMODIS, Landsat and Sentinel dataNDVI thresholdcropland abandonmenttropical/subtropical monsoon areas
Cropland abandonment has become a global land use change phenomenon that threatens food security. Identifying abandoned cropland in time and space can provide important data for cropland and agricultural management. In tropical/subtropical monsoon areas, sufficient water and thermal conditions allow crops to grow year-round, yet the specific crops grown vary in the dry and rainy seasons. However, the lack of optical remote sensing observations in the rainy season due to massive cloud cover and rainy weather makes it difficult to monitor abandoned cropland. Furthermore, when crops have multiple maturities and abandonment patterns, the mapping of abandoned croplands using remote sensing data at the annual scale is even more challenging. To resolve these problems, we proposed an approach for identifying abandoned croplands at the growing season scale. First, we used daily MODIS data to estimate the early and late growing seasons from 2013 to 2022. Then, we synthesized all available Landsat and Sentinel data to reduce the problem of inadequate remote sensing observations in the two growing seasons. Subsequently, we determined an NDVI threshold at the start of each growing season. In addition, we obtained the fitted NDVI curves for each growing season using the LandTrendr algorithm. Then, we determined the cultivated status by comparing the fitted NDVI curves with the NDVI threshold and then generated time series cultivated/noncultivated status maps. Finally, a temporal moving window was used to scan the abovementioned maps to identify abandoned cropland. The results showed that the proposed approach used to identify cropland abandonment achieved a favorable overall accuracy. From 2013 to 2014, a small amount of abandoned cropland concentrated in southern Leizhou was identified. In the early growing season during 2015–2017, a small area of abandoned cropland occurred along the western coast of Suixi. Since the late growing season of 2017, the abandoned cropland area has dramatically increased. Our method eliminates the interference that short-term fallow causes in identifying abandoned land and has high accuracy in areas with multiple crop maturities. Parcel-based recognition is more accurate in identifying fragmented croplands. Our results can provide reliable evidence for successful interseasonal cropland-use monitoring and food security conservation efforts.
1
A temporal moving-window analysis of seasonal cultivated-status maps enabled identification of cropland abandonment with favorable overall accuracy.
2
Daily MODIS observations were used to estimate early and late growing seasons from 2013–2022, while combined Landsat and Sentinel data addressed rainy-season observation gaps.
3
NDVI thresholds and LandTrendr-fitted seasonal NDVI curves were compared to classify cultivated versus noncultivated conditions and generate time-series status maps.
4
The method detected spatially and temporally varying abandonment, including areas in southern Leizhou during 2013–2014 and along western Suixi during the early growing seasons of 2015–2017.
5
The study introduces a growing-season-scale approach for mapping abandoned cropland in tropical and subtropical monsoon regions with multiple crop maturity patterns.

Abandoned cropland in tropical/subtropical monsoon areas, characterized by multiple crop maturity patterns and seasonal growing cycles

Spatiotemporal identification and mapping of cropland abandonment across early and late growing seasons despite seasonal observation gaps

Publication Details
Publication Date
2024-02-08
Journal
International Journal of Applied Earth Observation and Geoinformation
Publisher
Elsevier BV
ISSN
0303-2434
Cited by
11
Access Type
Author Information
Authors
Guilin Liu
Yuecheng Li
Chen Yang
Yongquan Lu
Desheng Jiang
Aojie Xu
Yuqi Zhong
He Yin
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%