Regional high‐resolution cloud climatology based on MODIS cloud detection data

Abstract

Most satellite cloud climatologies come in the form of global, low-resolution datasets: so-called 'gridded' Level 3 products, resulting from the reprojection and spatio-temporal aggregation of swath (Level 2) data. Their coarse resolution means that global datasets are of limited usefulness in regional studies. In this paper we develop and evaluate a new, regional cloud climatology over Poland and its neighbouring countries (∼10% of the area covered by Europe), based on observations performed with the state-of-the-art cloud imager, the moderate resolution imaging spectroradiometer (MODIS). In contrast to the operational, global MODIS cloud climatology, which is delivered as a Level 3 product at a spatial resolution of 1 ∘ × 1 ∘ , this regional climatology maintains the MODIS nadir spatial resolution of 1 km/pixel. The resulting high-spatial-resolution climatology is compared with AVHRR and SEVIRI datasets, and surface-based (SYNOP) observations at the level of monthly and annual means. The results shows that the standard MODIS Level 2 cloud mask product MOD35/MYD35 can be successfully used to develop a regional, high-resolution cloud climatology. MODIS provides reliable estimates of cloud amount at the national scale (annual mean: 64.0% or 70.8%, depending on the MODIS data interpretation scheme), and correctly reproduces the annual cloud amount cycle (correlation between monthly means with SEVIRI/AVHRR >0.98). A comparison with monthly mean surface observations reveals a bias ranging from −1.1% up to 5.9%, and a root mean square error of 4.2% − 6.6%. MODIS data also correctly indicates the spatial distribution of clouds. However, local anomalies were detected that were identified as artifacts of the MODIS cloud detection algorithm. Those artifacts covered 9% of the study area, but had no impact on spatially-averaged metrics.

Remotely Sensed High-Resolution Global Cloud Dynamics for Predicting Ecosystem and Biodiversity Distributions

“…There are a few examples of finer-grain climatologies based on other sensors, such as HIRS (≈20 km) [ 19 ], AVHRR PATMOS-x [ 20 ] (≈11 km), and GridSAT [ 21 ] (≈8 km), but these are eight to 20 times coarser than possible with MODIS observations ( S1 Table ). There have been several regional–national climatologies assembled at finer (≤1 km) resolution from MODIS data, and these generally perform well in comparison with station observations and other meteorological satellites (e.g., [ 22 ]).…”

Inconsistency of surface‐based (SYNOP) and satellite‐based (MODIS) cloud amount estimations due to the interpretation of cloud detection results

“…This thematic map shows pixels classified into each of the four (in the case of MODIS) cloud detection categories. The cloud mask is the starting point for a climatological estimation of regional or global cloud amount (Levizzani et al, ; Ackerman et al, ; Kotarba, ). The first step is to decide what cloud fraction (0–100%) is used to represent each of the cloud mask categories.…”

A New Fully Gap-Free Time Series of Land Surface Temperature from MODIS LST Data

“…Therefore, if the land cover layer used is not up-to-date or water bodies change their size, differences in cloud detection over water bodies and their borders might be expected. This was observed by Kotarba [43] while analyzing MODIS collection 5 cloud product for the period 2004-2009 over Poland. However, they detected more clouds over water bodies than over water body borders.…”

“…A 68% of all missing values present in MODIS LST products were interpolated in time through LWR, while the remaining gaps were filled by means of TPS in the spatial interpolation step. The fact that there were more missing values during day than at night might be explained by the MODIS cloud detection algorithm that uses thermal and reflective bands for day overpasses, but only thermal bands for cloud detection at night, i.e., clouds are more easily detected during the day than at night [43,44]. The difference in the percentages of missing values among water bodies and their borders might also be related to the MODIS cloud detection algorithm that uses different spectral thresholds for different types of covers [45].…”