Jurnal Internasional Faktor-faktor Pengendali Atmosfer Formaldehida (HCHO) di Amazon seperti yang Terlihat dari Satelit – Zhang – – Ilmu Bumi dan Luar Angkasa

Download Disini
Jurnal Internasional Faktor-faktor Pengendali Atmosfer Formaldehida (HCHO) di Amazon seperti yang Terlihat dari Satelit – Zhang – – Ilmu Bumi dan Luar Angkasa

3.1 Spatial Relationship Between HCHO and Various Controlling Factors

Figure 1 shows the average distribution of multiyear (2005-2010) FC, EDVI, RF surface shortwave, air temperature 2 m and the number of HCHO columns from 2005 to 2010. Four subregions (boxes in each plot) were selected based on the characteristics of vegetation water content (i.e., EDVI) and FC distribution. It should be noted that we present two HCHO distribution numbers here (Figure 1 e and 1 f), one from all effective samples (Figure 1 e), another sample from the sample in days without fires according to MODIS observations (hereinafter, fire-free HCHO, Figure 1 f).

  image

Means several years (2005-2010) means (a) Different Emissivity Vegetation Index (EDVI), (b) number of fires ( the total number of fire pixels observed in each grid)), (c) Clouds and Radiation Energy Systems The earth surface radiation flux, (d) surface temperature re-analysis, (e) observe the HCHO distribution, and (f) observe the HCHO distribution without fire

In Figure 1 a, the EDVI pattern shows three areas with different types of vegetation: dense rainforests, rare vegetation savannas, and transition areas between them. The vast Amazon rainforest is clearly presented with EDVI is more than 0.015. east and north of Amazonas shows EDVI value of less than 0.005. Between the two, there is a transition area that shows EDVI 0.005-0.015. Meanwhile, Mount Andes in the west and Amazon River can be seen clearly by very low EDVI around 0.

D nature Picture 1 b, FC small in heavy rainforest (eg Subregions No.1 and 2) and in savanna areas are rarely southeast (eg, Subregion No.4). FC is large in transition areas with median EDVI, as subregion No.3 is marked (Figure 1 b). The reasons for this FC distribution are beyond the scope of this paper. We speculate that in EDVI rainforest is very high, trees contain a lot of water, which prevents them from burning. In these rare savanna areas, fire is also not flammable due to lack of fuel. However, in transition areas with moderate vegetation, VWC is a medium and with enough biomass fuel for combustion, fire is easy to occur.

Annual average shortwave radiation (Figure 1] c) in dense forests lower than in southeast savanna mainly due to the effect of cloud shade. Because there is always a high cloud fraction above a dense rainforest area, the average surface air temperature (Figure 1 d) shows a relatively uniform distribution with small regional differences. [1945900]

The spatial pattern of the average number of HCHO columns throughout the day (Figure 1 e) shows higher values ​​in overgrown areas (for example, rainforest areas) compared to areas that are rarely vegetated such as easetern savanna and the Andean mountain. The overall spatial correlation coefficient (hereafter, R) between HCHO and EDVI is 0.44 ( p

However, the area with the highest atmospheric HCHO dose does not match the area with the EDVI peak. the area with the largest FC (Image 1 b) In the transition from solid (high EDVI) to rare vegetation around 10 ° S, high HCHO, more than 1945,1915 15 × 10 15 ] molecule / cm 2 and a high FC value of 2,000–5,000 (Figure 1) both are found, concluding the effect of a significant magnification of fire on atmospheric HCHO in these places. the spatial relationship between HCHO and FC is 0.36 (1945-1920]

Although in all study areas, vegetation shows the strongest control effect on the HCHO atmosphere, the relative importance of fire can be higher than vegetation in areas with large FC. for example, in the area described by the dashed line rectangle in Figure 1 R between HCHO and EDVI is only 0.14 (1945-1919] p = 0.0019), but R between HCHO and FC are 0.37 ( p

In all study areas, HCHO shows a negative spatial correlation ( R = −0.37, p R at 0.27 ( p

Because the exchange of carbon and water are combined between vegetation and the atmosphere through photosynthesis and respiration, high EDVI implies both water in high vegetation and active physiological activity, which releases more VOC to enlarge atmospheric HCHO concentration. On the other hand, active fire also emits VOCs loudly and increases HCHO concentration. Emissions from biomass combustion (i.e., fire) are always tangled with biogenic emissions and make a combined control effect on the spatial pattern of atmospheric HCHO.

In satellite observations with coarse temporal resolution, it is difficult to disentangle the effects of these two. controlling factors at HCHO. In areas with frequent fires, the effects of fires can even cover the effects of biogenic emissions. However, from the point of view of modeling studies, it is important to know the biogenic emissions of VOCs. To obtain this knowledge, we recalculated the long-term HCHO using samples collected at that time without fire according to MODIS observations. This HCHO “fire-free” spatial pattern is shown in Figure 1 f.

In the area outlined by the dash rectangle with frequent fires in Figure 1 the correlation coefficient between the original HCHO and EDVI was only 0.14, but the correlation coefficient between fire-free HCHO and EDVI increased significantly. to 0.23 ( p

3.2 Temporal Correlation Between HCHO and Various Control Factors

The set of monthly average times of HCHO, FC, EDVI, air temperature and surface solar radiation in the 4 selected subregions are presented in Figure [1945902] 2. Related correlations are listed in Tables 1 with the value of p. 19459020. The effect of fire is not omitted here.

  image

Emissivity Different Vegetation Index (EDVI), HCHO, Fire Count, radiation, and monthly temperature values ​​in four regions. EDVI and HCHO averaged based on observation time, and fire calculation averaged per day in this region. (a) Region 1, forest located north of the Amazon River; (b) Region 2, forest located south of the Amazon River; (c) Region 3, latitude equal to region 2 , transition from forest to short vegetation; (d) Region 4, area of ​​short vegetation. [19659021] Table 1.
Temporal Correlation Coefficient Between HCHO and Monthly Control Factors

Area Fire Counting Radiant Flux EDVI
Region 1 Northern forest

0.36

p = 0.002

p = 0.00

p = 0.00

p

0.48 [19659006] [1945902]

0.19

p = 0.12

Region 2 South Forest

0, 86

[1945902]

p 19659160] p

0.12

p = 0.298 [19659031] 0.25

p = 0.034

Region 3 Mixed area

0.44

p ]

0.73

[1945902]

−0.363 [19659006] = 0.002

Region 4 Savanna [1965915] 0.20

p = 0.097

0.71

p [1945902]

p [194590960]

0.43

p [19659060] Note . Thick value: pass significance 95%; oblique value: not passed. EDVI = Emissivity of Different Vegetation Indexes.

In region 1 (Figure 2 a), the vegetation is very dense and the spatial average EDVI (green curve) remains relatively stable at around 0.012 throughout the year. Meanwhile, FC (black vertical bar) is very small here compared to other regions. With a state of stable vegetation and very little fire disturbance, the temporal variation of atmospheric HCHO is mainly controlled by solar radiation with a temporal correlation coefficient of 0.69, followed by air temperature with a temporal correlation coefficient of 0.48. The EDVI vegetation index only shows a temporal correlation of 0.19. Because there are almost no fires in this region, the correlation between HCHO and FC is not discussed.

Region 2 is also located in heavy rain forest but is affected by several biomass burning events around October. The number of HCHOs shows one high peak value in August and a small peak around March. The temporal correlation between HCHO and FC ( R = 0.86) is very high, and the high HCHO peak matches the FC peak. This illustrates that in this region, wild combustion emissions contribute the most significantly to VOC emissions and determine seasonal variations in HCHO concentrations. The role of the radiation controller is also evident in region 2. Small HCHO peaks around March (similar to March peaks in region 1) are associated with peak solar radiation. The HCHO value and radiation value are also highly correlated with a correlation coefficient of 0.58. However, the correlation between the monthly EDVI and HCHO averages is relatively weak (0.25). This is caused by fire contamination because it plays a dominant role in determining HCHO's temporal variation here. On a monthly average scale, there is no way to exclude the impact of a fire. In the following EDVI correlation analysis and fire-free HCHO using daily scale satellite observations, we found a high correlation between the two (see further discussion).

In region 3, the transition region (Fig. [1945927] 2 c), the EDVI value is around 0.01, lower than those in regions 1 and 2, indicating relatively less deep-water vegetation and physiological activity weaker. However, FC here is much higher than in region 2, showing very strong fire activity. HCHO has a very high temporal correlation with FC ( R = 0, 93), followed by temperature (0.74) and radiation (0.46). The correlation coefficient between HCHO and EDVI is negative (−0.36). This again shows that in areas with significant fires, fire will make dominant control of VOC emissions and consequently HCHO concentrations. In this situation, the role of vegetation is unclear and the results may be misleading.

In region 4 (Figure 2 d), the EDVI value is always lower than 0.008 which indicates low vegetation water content, and FC effects are smaller than those in regions 2 and 3. In this situation the HCHO temporal variation is again controlled by solar radiation (1945-1919) R = 0.71) and air temperature (0.86). It only shows mild correlation with EDVI and the correlation is not significant with FC.

As summarized in Table 1 in regions 2 and 3 with frequent fires, the temporal variation of HCHO has a strong correlation with FC (1945-1919) R 19459020 = 0.85-0.93) . Environmental conditions, solar radiation, and air temperature always show strong temporal correlations with HCHO.

The temporal correlation between HCHO and EDVI became unclear most likely due to fire contamination. However, from Figure 1. 1945 the effect of vegetation control on HCHO is the strongest on a regional scale (the spatial correlation coefficient is 0.46). Based on Figure 2 HCHO concentrations are very sensitive to FC, and the effects of fire can envelop the effect of vegetation on HCHO concentrations because both are entangled in monthly average data. To uncover the effects of vegetation, we use daily scale data to recalculate temporal correlations. And for vegetation conditions, daily observations are only available from microwave-based vegetation indices, such as EDVI, because they are less affected by clouds. The current optical vegetation index, such as MODIS NDVI, is only available at 8 or 16 day intervals and thus cannot be used to carry out the analysis.

Daily HCHO scatter plots on FC, EDVI, solar radiation, and air temperature in the four selected subregions are presented in Figure [1945-1958] 3 . Open points represent fire samples and the solid points each represent fire-free HCHO samples.

  image

Scatterplot the daily HCHO column versus the amount of fire, radiation flux, temperature, and Emissivity Different Vegetation Indexes in four regions. detected, and the solid point denoted, no fire detected (“fire free”). RMSE = root-mean-square error.

At the daily level, FC still shows high correlation with HCHO in regions 2 and 3 (Table [1945909] 2 [1945900]) radiation and air temperature still show good correlation with HCHO in all subregions. These results are all consistent with those at the monthly average level.

Table 2.
Temporal Correlation Coefficient Between Instantly Retrieved HCHO and Control Factors
Area Flux Radiation Temperature EDVI
Region 1 Forest North

0.33

(

0.50 / 0.52

(

[19659078] 0.45 / 0.36

[

0.08 / 0.02

(

Region 2 of South Forest

0.46

(

0.34 / 0.40

(

  0.15 / </i> 0.45 </b> </p>
<p> (</p>
</p>
</td>
<td class=

0.22 / 0, 32

(

Region 3 Mixed area

0.60

([19659006]

0.39 / 0.37

(

0.73 / 0.42

(

[1945038/ 0.24

(

Region 4 Savanna

0.13

(

  0.32 </i></b><i> / </i> 0.11 </p>
<p> (</p>
</p>
</td>
<td class=

0.69 / 0, 53

(

0.36 / 0.37

(

  • Notes . The left slant value is calculated using all samples; the right tilt value is calculated using the “fire free” sample. A thick value indicates a pass of 95% significance. EDVI = Emissivity of Different Vegetation Indexes.

However, the temporal correlation between HCHO and vegetation (i.e., EDVI) shows very different results at the daily level compared to the monthly average level. And this is mainly due to the elimination of the effects of fire. In regions 1 and 4 with relatively weak fire effects, the temporal correlation between HCHO and EDVI showed small differences between samples with and without fire. In region 2, R between HCHO and EDVI was increased from 0.22 to 0.32 after removing the fire sample. And in region 3, the effect of eliminating fire samples became stronger, and 1945-1919 R increased from negative (−0.38) to positive (0.24). The results of daily analysis strongly show that biogenic emissions can be covered by burning emissions, and while monthly data is too rough on a temporal scale to separate the effects of fire, the results may be misleading. It is important to use daily data to exclude fire samples to provide more reliable results from biogenic emission control factors.

Then the distribution of the temporal correlation coefficient between HCHO and fire (Figure 4 a), HCHO free fire and solar radiation (Picture 4 b)), air tempo (Picture 4 c), and EDVI (Image 4 b) studied. Only the results of the temporal correlation pass the 95% confidence test shown in Figure 4 .

  image

Correlation (pass 95% significance) between (a) HCHO and number of fire, (b) HCHO and radiation flux, (c ) HCHO and temperature, and (d) HCHO and the Emissivity Different Vegetation Index (EDVI). Correlations are calculated using daily data with fire in (a) and daily data excluding fire in (b) – (d).

In Pictures 4, in areas with a significant number of fire incidents, the number of HCHO and FC columns is highly correlated (1945-1919) R 19459020 = 0.6-0.8) This result is also consistent with Marbach et al. ( 2008 ) The relationship between HCHO and RF is almost all positive (Figure 4 b) except in some places with frequent fire incidents, the correlation is around 0.3 in the central rainforest , more than 0.5 in eastern rainforest, and 0.2 in the area of ​​short eastern vegetation. The correlation between temperature and HCHO is almost all positive in Figure 4 c.

In terms of vegetation conditions, in heavy rainforests, there is almost no positive correlation between fire-free HCHO and EDVI. On savanna, the correlation was significantly positive with a correlation coefficient of up to 0.5. Compared to the results in Figure 4 and 4 we speculate that, in rainforests, vegetation activity is strong and relatively uniform in time and space, and thus, VOC emissions do not depend on EDVI variations everyday, and radiation controls VOC emissions. But in the savanna area, vegetation is rare and vegetation conditions change dramatically over time and location, and therefore, the EDVI condition becomes as important as radiation.

3.3 Parameterization of Biogenic Emission-Induced HCHO with EDVI and Radiation

Contributions of biogenic emissions, namely emissions that are free from biomass combustion, into the HCHO atmosphere appeal to researchers in the field of environmental science. However, in real observation, these components are always entangled with other contributions, such as those originating from fire etc. We propose parameterization based on the experience of biogenic emissions induced by HCHO (B-HCHO) based on the above analysis.

We assume that the B-HCHO is dependent on RF and vegetation conditions based on above results and then Determine the associated coefficients with the daily collected HCHO-RF-EDVI data during 2005 to 2010.

 urn: x-wiley: 23335084: media: ess2307: ess2307-math-0002

The parameterization scheme, the coefficients of B and B 2 The sensitivity of B-HCHO to RF and EDVI. and has significant spatial variations, we show maps of B1 and B2 in Figure 5, 1945. In the rain forest, the radiation conditions dominate the emission, and therefore, B1 is significantly larger in the rain forest than that in transition area. And in the savanna area, vegetation condition is important for biology emission, and large values ​​of B2 are found.

 image

Bilinear formula coefficients (a, b); HCHO calculated (c) and between HCHO observed and calculated (d).

To assess the performance of this parameterization, we compare the modeled multi-averaged B-HCHO to the one derived from satellite observation after excluding all fire samples (Figures 5 c and 5 d). The calculated HCHO is generally comparing to the satellite observation. In the most areas except the coastal area along the Andean Mountain, the difference is small and the relative bias is under 20%. Based on the temporal variation of HCHO calculated and observed in the forest and savanna areas (Figure 6 ), the calculated temporal variation of HCHO (red dots) matches the blue dots well, and the correlation between the two is over 0.5 in both forest and savanna. This provides a strong evidence that in the Amazon region, the biogenic emissions are mainly controlled by radiation and vegetation conditions.

 image

Temporal variation of the calculated HCHO and observed HCHO in forest and savanna. Red dots denote calculated, blue ones denote observed.

In addition, we have tried adding the temperature to the model, but the coefficients of radiation and EDVI became noisy, and the HCHO density predicted bias in not improving.

Download Disini

Leave a reply "Jurnal Internasional Faktor-faktor Pengendali Atmosfer Formaldehida (HCHO) di Amazon seperti yang Terlihat dari Satelit – Zhang – – Ilmu Bumi dan Luar Angkasa"

Author: 
    author