VIIRS/DNB Monthly and Yearly Nighttime Light Dataset in Beijing-Tianjin-Hebei Region

Chen, M. L.^1,2  Cai, H. Y.^1*

1. State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China;

2. University of Chinese Academy of Sciences, Beijing 100049, China

Abstract: The new generation nighttime light data, which sensed by the Visible Infrared Imaging Radiometer Suite Day/Night Band (VIIRS/DNB) are severely subject to stray light, resulting in the appearance of a large number of missing pixels in mid and high latitudes, especially in the summer. Therefore, to obtain the consecutive spatial-temporal nighttime light data and promote the application of nighttime light data, this study applied the cubic Hermite interpolation algorithm to interpolate the missing values from May to August in Beijing-Tianjin-Hebei Region. This dataset includes monthly and yearly average subsets. The time span is 2013-2018, the spatial resolution is 15², and the total data volume is 234 MB. Moreover, since ephemeral light was filtered out, this dataset is more applicable for stable socio-economic research, rather than ephemeral light research.

Keywords: VIIRS/DNB nighttime light; cubic Hermite interpolation; missing pixels; monthly average composite; Beijing-Tianjin-Hebei Region

1 Introduction

The new generation nighttime light (NTL) data, which sensed by the Visible Infrared Imaging Radiometer Suite Day/Night Band (VIIRS/DNB) are severely vulnerable to stray light, resulting in the appearance of massive distorting pixels in mid and high latitudes, especially in the summer (May to August)^[1]. To address this problem, NOAA provided two versions of the NTL, denoted by the VCM (VIIRS Cloud Mask) and VCMSL (VIIRS Cloud Mask with Stray Light), respectively. Pixels that contaminated by the stray light were eliminated in the VCM version, and therefore numerous missing pixels exist in this version. The VCMSL version covers more data toward the poles after corrected the distorting pixels by a stray light correction procedure that provided by Mills^[2]. However, two significant imperfections exist in the VCMSL version. One is the absent of 2012 and 2013 in this version, the other one is the quality still need to be improved^[2-3]. For example, radiation suddenly becoming low in some areas were found in the VCMSL version^[4]. Thereby, interpolating missing values is importance for obtaining consecutive spatial-temporal NTL data and promoting its application.

Given the fact that the missing pixels tend to exist over a large area, interpolation based on temporal interpolation method is more suitable than spatial interpolation method. The cubic Hermite interpolation is one of the most prevalent temporal interpolation algorithms. This algorithm has the advantages of high accurate and no overshoot^[5-6]. Thus this algorithm was used to interpolate the missing pixels of the VIIRS/DNB NTL data in the summer (May to August). Furthermore, the Beijing-Tianjin-Hebei Region is a representative economic and population agglomeration area in China, located in the middle latitude region and contaminated by the stray light in the summer. As a result, this study mainly focus on interpolating missing pixels in this region, producing a monthly average dataset without missing pixels (here after called the interpolated NTL data), and synthesizing annual average dataset on this basis.

2 Metadata of Dataset

The metadata of the VIIRS/DNB NTL data without missing pixels in Beijing-Tianjin-Hebei Region dataset^[7] is summarized in Table 1. It includes the dataset full name, short name, authors, year of the dataset, temporal resolution, spatial resolution, data format, data size, data files, data publisher, and data sharing policy, etc.

3 Methods

3.1 Data

The original data of this study were the monthly VCM composites from 2012 to 2019. When interpolating the missing pixels in May to August based on the cubic Hermite interpolation method, the eight months before May and the eight months after August of the VCM data served as the original data. Taking 2015 as an example, monthly VCM data from September 2014 to April 2015, and September 2015 to April 2016 are the original data for interpolation.

Table 1  Metadata summary of the VIIRS/DNB NTL data without missing pixels in Beijing-Tianjin-Hebei Region

(To be continued on the next page)

(Continued)

3.2 Algorithm Principle

Background noise and ephemeral light does not remove from the current VIIRS/DNB NTL data. Consequently, some pixels with negative values and abnormally high-valued exist in the original NTL data. The threshold method was used to eliminate the outliers. Pixels with negative values were reassigned to zero^[9]. The biggest median value of the whole study period in the study area, which was 285 nWcm^-2sr^-1, was served as the up threshold. Pixels with a value that larger than 285 nWcm^-2sr^-1 were reassigned to that value.

The core idea of the cubic Hermite interpolation is to construct a polynomial that no higher than cubic degree between each two adjacent nodes. Time nodes of a pixel are assumed to be . Correspondingly, its radiations are assumed to be  . After calculated at pixel level, the interpolation curve f (x) will: (1) ; (2) the polynomial between any two adjacent nodes is no higher than cubic degree; (3) satisfy the first derivative, but not necessarily the second derivative. That is to say the curve stay shape preserving^[10], and the results would not exceed the maximum value of the original data. Calculation were accomplish by MATLAB.

3.3 Data Processing

The flow of producing the monthly and yearly interpolated NTL data is shown in the Figure 1. Taking 2015 as an example, the monthly VCM NTL data from September 2014 to April 2016 (except May to August 2015) in Beijing-Tianjin-Hebei Region were collected as the original data. After the outliers remove and the cubic Hermite interpolation calculation, the monthly interpolated NTL data from May to August 2015 were produced. On this basis, combining with monthly VCM NTL data in other months of 2015, the annual average NTL data of 2015 was synthesized.

4 Results and Validation

4.1 Data Composition

The dataset consists of two subsets: One subset is the yearly average composite. Its time span is from 2013 to 2018, the other subset is the monthly average product, including monthly data of May to August of 2013 to 2018. The unit is nWcm^-2sr^-1.

Figure 1  Flow chart for producing the interpolated monthly and yearly nighttime light data

4.2 Data Products

An example of interpolated NTL data is shown in Figure 2: Both monthly (Figure 2(a)) and yearly (Figure 2(b)) data can significantly describe the outlines of Beijing, Tianjin, Shijiazhuang and other middle and small cities. They also reflect the radiation differences within the cities. In a certain city, in general, the largest radiations are commonly located in the airport, such as the Beijing Capital International Airport, the Tianjin Binhai InternationalAirport and the Tianjin Tanggu Airport (Figure 2(b)). The radiations of these places

Figure 2  The interpolated nighttime light data

can ar-rive more than 200 nWcm^-2sr^-1. The smallest radiations usually sprawl in the urban fringe. Their radiations tend to be less than 10 nWcm^-2sr^-1. The radiations in other region within the city are about 10–100 nWcm^-2sr^-1, in the developed commercial areas are generally between 50–100 nWcm^-2sr^-1.

4.3 Data Validation

The NTL data are mainly composed by stable artificial light. If the cities did not suffer any disaster or war, their radiation time series should fluctuate slightly^[11,12]. Therefore, radiation time series of three version of NTL data (the VCM version, the VCMSL version and the interpolated version) in the study area during September 2014 to April 2016 were compared. As shown in Figure 3, in the non-summer months, there is little variation between the three versions. In the summer months (May to August), however, obvious differences are easy to distinguish. In the VCM version, due to remove the contaminated pixels, its radiation drops sharply during the summer months, even drops to zero in June. On the contrary, the radiation time series of the VCMSL version and the interpolated version tend to be more stable resulting from stray light correction. However, further comparison reveals that the radiation of the VCMSL version in July is much small than other months. Spatially, its nighttime light contrast is weakened and the urban texture is not clearly depicted (Figure 4(b)). In contrast, the interpolated version was most stable, and its spatial distribution was reflective of the texture of the city (Figure 4(a)).

Figure 4  The comparison of three nighttime light data (taking July 2015 as an example)

5 Discussion and Conclusion

As one of the commonly used interpolation algorithms, the cubic Hermite interpolation has obvious advantages. It does not require the original data to conform to the statistical distribution. Its calculation only require a short time to accomplishment and does not require subjective intervention. Most importantly, its results are stable^[10]. Therefore, this algorithm is suitable for VIIRS/DNB NTL interpolation that with a large number of pixels and a short time series. In addition, the calculation of this algorithm is based on the average state of the original data, to some extent ensures that the interpolated results to be of stable, consequently avoiding the situation that the radiation become too small in certain months in the VCMSL version. Thus the interpolated version is more consistent with the fact that most nighttime light show slow changes. At the same time, the algorithm can effectively filter out fire and abnormally high-valued outliers, so that it is more effectively applied to the study of social and economic activities research.

Nevertheless, the cubic Hermite interpolation has its limitations. Firstly, the algorithm has a high requirement on the length of the original data. Secondly, the algorithm is easily subject to outliers. Third, since the interpolation calculation is mainly based on stable light, the results do not reflect ephemeral light. So the interpolated dataset is mainly applicable to the study of stable light that reflecting social and economic activities, rather than the ephemeral light research.

Author Contributions

Chen, M. L. designed the algorithms of the dataset and wrote the data paper. Cai, H. Y. contributed to the arrangement and revision of the paper.

References

[1]       Mills, S. D., Straka, W., Mills, S, P., et al. Illuminati the capabilities of the suomi national polar-orbiting partnership (NPP) visible infrared imaging radiometer suite (VIIRS) day/night band [J]. Remote Sensing, 2013, 5(12): 6717–6766.

[2]       Mills, S., Weiss, S., Liang, C. VIIRS day/night band (DNB) stray light characterization and correction [C]. Earth Observing Systems XVIII. International Society for Optics and Photonics, 2013, 8866(11): 350–354.

[3]       Bennett, M. M., Smith, L. C. Advances in using multitemporal night-time lights satellite imagery to detect, estimate, and monitor socioeconomic dynamics [J]. Remote sensing of environment, 2017, 192: 176–197.

[4]       Chen, M. L., Cai, H. Y. Interpolation methods comparison of VIIRS/DNB nighttime light monthly composites: A case study of Beijing [J]. Progress in Geography, 2019, 38(1), 126–138.

[5]       Geng, A. C. The application of matlab in teaching of cubic spline interpolation function [J]. Value Engineering, 2016, 35(18): 181–182.

[6]       Fritsch, F. N., Carlson, R. E. Monotone piecewise cubic interpolation [J]. SIAM Journal on Numerical Analysis, 1980, 17(2): 238–246.

[7]       Chen, M. L., Cai, H. Y. The monthly and yearly VIIRS/DNB nighttime light dataset with the cubic Hermit interpolation correction in Beijing-Tianjin-Hebei Region [DB/OL]. Global Change Research Data Publishing & Repository, 2015. DOI: 10.3974/geodb.2015.01.17.V1.

[8]       GCdataPR Editorial Office. GCdataPR data sharing policy [OL]. DOI: 10.3974/dp.policy.2014.05 (Updated 2017).

[9]       Shi, K., Yu, B., Huang, Y., et al. Evaluating the ability of NPP-VIIRS nighttime light data to estimate the gross domestic product and the electric power consumption of China at multiple scales: A comparison with DMSP-OLS data [J]. Remote Sensing, 2014, 6(2): 1705–1724.

[10]    Manni, C. On shape preserving C2 Hermite interpolation [J]. BIT Numerical Mathematics, 2001, 41(1): 127–148.

[11]    Li, X., Zhang, R., Huang, C., et al. Detecting 2014 Northern Iraq Insurgency using night-time light imagery [J]. International Journal of Remote Sensing, 2015, 36(13): 3446–3458.

[12]    Li, X., Li, D., Xu, H., et al. Intercalibration between DMSP/OLS and VIIRS night-time light images to evaluate city light dynamics of Syria’s major human settlement during Syrian Civil War [J]. International Journal of Remote Sensing, 2017. 38(21): 5934–5951.