# Vegetation index data products derived from near real-time MODIS and online services for CONUS based on NASA LANCE

### Workflow

The dataset described in this paper covers five types of indices, Normalized Difference Vegetation Index (NDVI), Vegetation Condition Index (VCI), Vegetation Condition Index (VCI), the mean reference vegetation (MVCI), the ratio to the median vegetation condition index (RMVCI) and the ratio to the vegetation condition index (RVCI) of the previous year. Figure 1 shows the overall workflow for producing all of the VI data products described in this article. Based on time resolution, our VI products can be grouped into daily product (NDVI) and weekly/bi-weekly products (NDVI/VCI/MVCI/RMVCI/RVCI). The main source of daily NDVI is LANCE-MODIS data. Other VI products are derived from daily NDVI data. The remainder of this section describes the calculation and processing details of the NDVI, VCI, and MVCI/RMVCI/VCI data products.

### NDVI

NDVI is the most widely used measure for quantifying vegetation conditions in remote sensing imagery. The daily NDVI data described in this study is derived from the RED band and the NIR (near infrared) band of MODIS data. It should be noted that the daily NRT NDVI data described in this study is produced based on the MODIS NRT product Terra Surface Reflectance Daily Level-2G Global 250 m (MOD09GQ NRT v061, dataset https://doi.org/10.5067 /MODIS/MOD09GQ .NRT.061) within a day of the sighting. Archived daily NDVI from 2000 to 2019 and derived weekly VI data are produced based on MODIS product Terra Surface Reflectance Daily Level-2G Global 250 m (MOD09GQ v061, dataset https://doi.org/10.5067/MODIS /MOD09GQ.061). Table 1 summarizes the surface reflectance band information for the MOD09GQ NRT product.

The first step of the NRT VI calculation is to mosaic the daily MODIS images and clip the area to the CONUS. Next, the NDVI is calculated using surface reflectance bands from the MODIS data, which are defined as:

$$NDV{I}_{i}left(x,yright)=frac{NI{R}_{i}left(x,yright)-Re{d}_{i} left(x,yright)}{NI{R}_{i}left(x,yright)+Re{d}_{i}left(x,yright)}$$

where NIRI and RedI represent the surface reflectance value for the near infrared band and the red band of each pixel. Based on the above equation, the NDVI value ranges from -1 to 1. The common NDVI value range for vegetation is 0.2 to 0.821. Weekly and bi-weekly NDVI data is calculated by capturing the maximum daily raw NDVI in the 7-day/14-day period for each week/two weeks using the Maximum Composite Value (MVC) method. This is different from the 8-day moving average method used in the MODIS VI (MOD13Q4N/MOD13A4N) products. The MVC method will keep the maximum value pixel of the daily NDVI charts in a week or two. This method can effectively reduce the effect of bad pixels (eg, cloud, fog, or no data) in a single daily NDVI.

To reduce the file size, we converted the NDVI data type from float to integer by scaling the index range by [−1, 1] at [0, 250]which can be defined as:

$$NDV{I}_{scaled}(x,y)=NDV{I}_{i}left(x,yright)times 125+125$$

### VCI

VCI is a vegetation index derived from historical NDVI22. It shows the percentage of the difference between the current NDVI value and the historical minimum NDVI value with the NDVI dynamic range, which is defined as:

$$VC{I}_{i}(x,y)=frac{NDV{I}_{i}left(x,yright)-NDV{I}_{min}left(x, yright)}{NDV{I}_{max}left(x,yright)-NDV{I}_{min}left(x,yright)}$$

where NDVII(x,y) represents the current NDVI value for the given pixel, NDVImaximum(x,y) and NDVImin(x,y) represent the maximum and minimum value of the historical NDVI time series for the given pixel.

Similar to NDVI data, the index range of VCI is scaled by [0, 1] at [0, 250]which can be defined as:

$$VC{I}_{scaled}(x,y)=VC{I}_{i}(x,y)times 250$$

### MVCI/RMVCI/RVCI

The MVCI and the RMVCI are indices measuring the difference between the value of the NDVI of the current year and the average and median value of the historical NDVI from 2000 to the current year. The equations of MVCI and RMVCI are respectively defined as:

$$MVC{I}_{i}(x,y)=frac{NDV{I}_{i}left(x,yright)-NDV{I}_{mean}left(x, yright)}{NDV{I}_{average}left(x,yright)}$$

$$RMVC{I}_{i}(x,y)=frac{NDV{I}_{i}left(x,yright)-NDV{I}_{median}left(x, yright)}{NDV{I}_{median}left(x,yright)}$$

where NDVII (x,y) represents the current NDVI value for the given pixel. NDVImean (x,y) and NDVImedian (x,y) represent the mean and median value of the historical NDVI in the same period for the given pixel.

RVCI, on the other hand, represents the ratio of the NDVI of the current year to previous years during the same periods. The equation is listed as follows:

$$RVC{I}_{i}(x,y)=frac{NDV{I}_{i}left(x,yright)-NDV{I}_{i-1}left( x,yright)}{NDV{I}_{i-1}left(x,yright)}$$

where NDVII (x,y) represents the NDVI value of the current year and NDVIi-1 (x,y) represents the NDVI value of the previous year.

To reduce file size and remove anomaly pixels, we limited the absolute value of the ratio to 125% and scaled the index range to [0, 250]. This conversion can be defined as:

V{I}_{scaled}(x,y)=left{begin{array}{cc}0 & if;V{I}_{i}left(x,yright) le -1.25 V{I}_{i}left(x,yright)times 100+125 & if-1.25