Land growth analysis and anomaly assessment based on time-series remote sensing images

Case Background

In Lanzhou, Gansu Province, some newly reclaimed plots of land in Lanzhou New Area showed significant differences in yield in the fourth quarter of 2024, with some plots exhibiting significantly poorer growth, even though the variety, seeding rate, slope, and irrigation water source were all the same. Therefore, the farm hopes to use remote sensing to analyze the growth of different plots and assess the reasons for the differences in growth in order to make improvements in subsequent management.

This plot of land is located in the eastern part of Lanzhou New Area, covering an area of ​​approximately 2,900 mu (about 193 hectares). After being leveled, the land will be terraced, and corn will be planted there, starting in the 2024 season. The specific phenological periods are as follows:

Monitoring and evaluation methods

For comparative analysis, three plots were selected: Plot 1 was the plot that showed abnormalities, Plot 2 was another newly built plot in the southeast direction, and Plot 3 was a plot of land that had been cultivated for many years by a nearby village.

Vegetation index time series analysis was conducted using multi-source remote sensing data, with the data spanning from the beginning of 2023 before land leveling to the end of 2024 after planting.

Based on the phenological periods extracted from the vegetation index change curves, it was determined that the crop planted in the three plots was corn.

By comparing areas with significant differences in growth, assess whether the differences in growth have spatial distribution characteristics. At the same time, based on reference plots, assess the intensity of the differences in growth. Combine meteorological, soil and other conditions for comprehensive analysis to determine the causes of the differences in growth.

Evaluation results

Figure 1. Distribution of vegetation index before land leveling

Figure 2. Distribution of vegetation index after land leveling

Figure 1 is a composite map showing the maximum NDVI values ​​during the growing season before land leveling in 2023. Figure 1 shows that the area was mountainous before reclamation, and crops were planted on flat plots on both sides of the road, and the crops were growing normally (blue area in the figure).

Figure 2 shows the vegetation index distribution map after land leveling in 2024. It can be seen that the land cover type has changed significantly, with large areas being reclaimed as farmland. The target plot is located in the central part of the region and does not differ significantly from the surrounding areas in terms of topography and slope, but the vegetation growth is significantly poor and unevenly distributed. At the same time, the vegetation growth distribution has obvious spatial characteristics, with significant differences between plots.

Considering that the plots were all artificially divided and were planted for the first time, and that the management plans were all the same and the phenological periods were similar, the differences between the plots due to agricultural operations were ruled out.

By comparing the distribution of farmland before and after reclamation, it was found that the new farmland with better growth after reclamation has a certain similarity in spatial distribution with the old farmland before reclamation, that is, it extends from northwest to southeast, and the new farmland plots near the old farmland generally have better growth than other plots.

Figure 3. Time series of average vegetation index for different plots

(Purple and blue curves represent newly built plots of land, while green curves represent plots that have been cultivated for many years.)

Figure 4. Distribution map of different plots

Combining Figures 3 and 4, it was found that the newly built plots were growing worse than the reference plots, thus it was basically determined that the problem lay in the soil. After further communication with the farm's establishment personnel, the land leveling and covering methods were understood: the soil was basically covered in situ, with the best-growing areas using the original farmland soil, other plots with average growth also using topsoil for covering, and the worst-growing areas using the subsoil from the land leveling.

Through long-term human agricultural production activities, measures such as tilling and fertilization have increased the organic matter in the soil, improved its structure, and enhanced its fertility. Simultaneously, microbial activity also promotes the soil maturation process. However, raw soil, mainly formed from weathered rocks and deposited parent material, typically has poor structure and fertility, lacking organic matter and nutrients. Under natural conditions, the maturation process of raw soil is very slow.

Therefore, the cause of the difference in growth was determined to be soil conditions. Although the fertility of the raw soil is low, it can be gradually transformed into mature soil through improvement and fertilization measures, thereby improving land use efficiency. However, this improvement process requires a significant investment of human and material resources.

Based on the vegetation index distribution map, it is possible to accurately distinguish between areas covered by topsoil and subsoil, and to estimate the mixing ratio of subsoil and topsoil at that time based on the relative value of the vegetation index. This allows for the creation of soil improvement plans with different gradients for different areas, realizing variable operations for soil improvement and saving agricultural input costs.

Furthermore, remote sensing can be used to continuously monitor crop growth, assess the effectiveness of soil improvement, and adjust the variable scheme and implementation area year by year to achieve scientific management and sustainable utilization of farmland resources.