the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Enhanced Quantitative Precipitation Estimation (QPE) through the opportunistic use of Ku TV-sat links via a Dual-Channel Procedure
Abstract. Earth – satellite microwave links such as TV-SAT can help for rainfall monitoring and could be a complement or an alternative to ground-based weather radars, rain gauges or satellites dedicated to Earth observation. Rain induced attenuation which is harmful for telecommunication is exploited here as an opportunistic way to estimate rain rate along the path link. This technique makes it possible to obtain rain measurements at a fine temporal resolution (a few tens of seconds) and with a spatial resolution of few kilometers, which is a good compromise for human activities such as civil security (watershed monitoring, flash flood), agriculture or transport. Among the advantages of this technique, one can note the low cost of the hardware used (which can be commercial) as well as that of its maintenance on site. However, measured attenuation does not directly provide rain intensity and some parameters have to be estimated. Among these, it is necessary to take into account the contribution of the natural radiation of the atmosphere. In this paper, we detail a theoretical framework allowing to estimate rainfall from the measurements of a low-cost sensor operating simultaneously over two parts of the Ku-frenquency band. Then this framework is assessed in a densely instrumented area in south of France, where it is shown that very good results are obtained when compared to rain gauges measurements, both in terms of overall rain accumulation and in terms of rainfall rate distribution. Then we apply this dual channel method in Ivory Coast, in the metropolitan area of Abidjan, where such an approach is very promising. It is shown that this technique when compared to rain gauge measurements give results far better than a single-channel naive approach neglecting natural radiation of atmosphere, but that there are still significant errors in the rain assessment, leading to a persistent underestimation of rain accumulation. Finally we discuss various effects that could lead to this remaining underestimation, opening the door to further studies.
- Preprint
(6196 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on amt-2024-88', Marielle Gosset, 15 Jul 2024
In this study the authors propose and assess a new method to improve rainfall measurement from earth-satellite microwave links (from TV satellites).
The principle of rainfall estimation from earth-satellite MW links in the KU domain was already explained and assessed beore by the authors.
Here they concentrate on a particular and technical aspect of the problem : how to improve rainfall estimation by a better separation between i) the rain induced attenuation of the received MW signal - which is the base of measurement, through a specifc attenuation vs rain rate relationship and ii) the posible increase of RS because of the natuarl MW emisionof the atmosphere and the droplets themselves.
The proposed method is based on a dual-channel approach where one of the channels is senstive only to the atmospheric noise/radiation.
Altogether the paper is well structured and written. The proposed methodology and the physical principles are well explained.
The experimental results show that the proposed dual channel method does substantually improve the estimation.
The authors also aknowledge that there are still unresolved issues and sources of uncertainties/erros which are not accounted for.
In my view the paper could be published once a few details in the processing and assumptions have been clarified.I also propose some suggestions for the wording.
TECHNICAL ISSUES OR QUESTIONS REQUESTING FURTHER EXPLANATIONS
PN is time independent in Eq 1 – doesn’t it vary in time - with temperature for instance ?
L88-89 : to be clarified : do you mean that the values of TA on the one hand and Tatm on the other hand do not vary with meteorological conditions ? which central frequencies are assumed to be close to each other ? not clear…..
L96-103 :
Eq 2 and 3 don’t really show the double impact of rain – the dependence of Ta on rain has not been detailed yet …..
Eq 11 and 12 :
I do not understand when and why the dependence of tr (and tatm) on frequency (as shown in Eq 4, 5 and 6) was dropped
The simplifications made to go from Eq 2 – which as a strong dependence on f through tatm) to Eq 11 which has lost the dependence of tr to frequency have to be explained and justified.
In my understanding Eq 11 should have a term in trA and one in trB
L175-179 – The simplification in Eq 11 implies that trchannelA = trchannelB … These paragraph implies that this in not true
I believe some steps are needed for the reader to follow all simplifications leading to simplified Eq 12 from Eq 2 to 6
L225-229 : did the authors have a chance to make some comparisons between this ARPEGE iso0° and the one from the MeteoFrance high resolution model AROME ? or with the MeteoFrance radar data (where the iso0° can be extracted using some polarimetric variable)- Same questions with Radiosoundings (or statistics from a satellite based radar such as GPM DPR ) to verify the reliability of those ARPEGE levels ?
L229 : it would be nice to have more information about these experiments…. Context/data/results etc…
L405 and Fig10 : Please be more precise : what do you lean but correlation between devices ? which data is compared ? time series of rain intensity ? which time step ? any filtering of outliers ?
It would be interesting to see the scatter plots and more statistics on the timeseries (KGE ?) in addition to the QQplots which provide only partial information.
It is not clear from the paper which relationship is used for attenuation-rainfall estimation – are the ITU parameters mentioned in L191-193 applied to all experiments (dual and single ) ? Was there any adjustment/calibration ? if yes how ?
STYLE/WORDING suggestions
L3 : link path and not path link
L7 : which can be commercial (what is meant ? off the shelve ) ?
THE measured attenuationL19 : what is small-scale or medium-scale rainfall intensity ? are you talking about resolution or intensities values ? Needs to be clarified – If you mean intensities why single out small and medium , heavy rainfall is the most damaging for ‘human/property damage’
L21 : Earth Observation satellite (rather than remote sensing ) ?
L35 ‘previous studies ‘ – which ones ?
L40 : combination of the signal from the satellite and a background noise that depends on the state… ?
L43 : and the baseline signal measured during dry period
In the presence of high rainfall rates, however,…
L47 : ‘The present study …. dual channel’. long clumsy sentence – Just get to the point , which is improving rainfall estimation from dual-channel measurement of TV satellite signal by accounting for background noise. No need to repeat low cost etc….
L50 : the assumptions it relies on ? (rather than the hypotheses it presupposes)
L52 inherent to rainfall estimation using…
L54 : section 2 introduces the physical principles (rather than ‘context’)
L56 : a physical device whose characteristics …..
L60 : physical principles
L61 : ground satellite : you should choose and keep one expression – Earth-Satellite as used before or ground-satellite …..LinkS or a link.
Careful with missing articles along the paper….
L89 : the central frequency of what and compared to what ? not clear – what are each other ?–
check format of the citation.
L104 : Various processes …. Links – do the process characterize the transparency ? do they influence the propagation of the link (or of the microwave signal along the link…) ? To be rephrased.
L264 : data processing rather than data treatment
L280 : not to go below the threshold (and not exceed …. Which means the opposite)
L314 : Eq 17 and 18 are redundant – you can give directly Eq 18 in dB ….
Citation: https://doi.org/10.5194/amt-2024-88-RC1 - AC1: 'Reply on RC1', Louise Gelbart, 12 Sep 2024
-
RC2: 'Comment on amt-2024-88', Anonymous Referee #2, 09 Aug 2024
Summary:
The authors present a novel analysis of data from satellite microwave links for rainfall retrieval. They show (for the first time, as far as I am informed) the impact and the potential usage of measurements of radiation from rain drops in the atmosphere on the received signal level of satellite microwave links. They give a good theoretical explanation of the effect they are investigating and show convincing improvements with their proposed new method based on a fairly large dataset. They discuss the remaining differences compared to reference data, which are still considerable for their analysis in Ivory Coast, in a detailed and open manner, clearly indicating future pathways for research. The results of their proposed method are encouraging and relevant in general, because most (if not all) other studies with satellite microwave links seem to not have taken the investigated effect explicitly into account. The paper is mostly well written, the figures are mostly clear and it fits very well for AMT. I suggest a minor revision, event though I have quite a few specific comments. But, none of my comments addresses and issue that I consider to be major.
## Specific commentsL23: „…low revisit time compared to rainfall dynamics.“ It is somehow clear what is meant here, but it should be formulated more precisely. Please rephrase.
L60: General comment on Section 2. Each subsection is understandable, but I am missing condensed info on how the improved dual band method is actually applied. It is not directly clear from how the equations are linked in the text, how Delta_G impacts the rain rate estimation via the power law. Maybe there should be an additional subsection that links things together, from t_atm via A to the power law, but explaining how it is done with the Std and the Dual method. Maybe this also fits as an extension of section 2.4.
Fig 1: What is the path length (affected by rain-induced attenuation) that is used for the calculation of the attenuation on the y-axis here?
Fig 1: Would it be possible to also show the increase in received signal strength for the increased brightness temperature for a given bandwidth, e.g. 1 GHz as used in your LNB?
L200: There is a lot of information provided in the paragraph that starts here, but it is not clear if one of the enhanced models is used, and if not, why?
L245: Since the lower and upper frequency band are directly adjacent, is there power leakage from one band to the other, i.e. for the described case where one TV-satellite only transmits in one of the two bands, how much does still leak into the other band of the receiver where it somehow contaminates the radiometer-like measurement?
L272: Is this method with the LSTM documented somewhere in more detail? What is the temporal granularity at which the classification is done?
Figure 3: Is there the potential of leakage of the Astra 19 signals into the receiver of the RS sensors? Or more specifically, what is the half-power beam width, or in general the gain pattern of the antennas? And how high is the noise floor of the radiometer channel of the RS sensors in relation to the potential leakage of Astra 19 signal into the RS receiver via the RS system’s antenna side lobes?
L296: What does „almost no signal“ mean here. How is it different from the setup in France and how does/could it affect the rain rate retrievals?
Figure 3 and Figure 4: What are the assumed melting layer heights for the plotted path lengths? That would be interesting to know. In Abidjan the elevation angle of the antennas is probably much higher because of being close to the equators, hence, I expect a much shorter path that is relevant for a typical melting layer height.
L310 and following: It is clear from the explanations here and from the shown plots that channel A and B have different P_atm which can be attributed to G_A and G_B. Did you also check that there isn’t an offset or some other inaccuracy due to the low-cost electronics of the LNB, which is not optimised to give accurate readings of received signal level?
Figure 5: Just a tiny detail, but you could use aligned y-axes here (they are slightly misaligned) and then remove the y-axis tick labels of the plot on the right.
Figure 5: What is the unit on the y-axis. If it is not dBm, what is the reference level for the dB given here?
L321: How do you want to assure that the rain-induced attenuation is strong enough to have t_R approx. 0? Please explain in the text how you identify these events.
L339: Where does the difference of P_A_Tot and P_B_Tot during normal operation (not pointing away form the satellite) come from? Is this due to different transmit power of the satellite in the two bands or can this also be an effect of different gains of the electronics for band A and B?
L350 and following: I unterstand the argumentation here on why Delta_G_p3 is used. But doesn’t this, the difference of Delta_G depending on what the current brightness temperature is, mean that Delta_G varies with rain rate? If yes, does this affect your results?
L374: Why does the existence of a dry season „explain the need for sufficient data to calculate Delta_G“? Do you mean that it is harder to get enough data with heavy rain due to the dry season? Please rephrase.
Figure 7: What happened during the period in September 2022 where signal levels for A and B both are increased for several days?
Figure 8: I would put the box plot with Delta_G_ref in the middle. But if you redo this plot, you might consider doing it with something else than boxplots since, here, the spread and distribution of the rainfall sums of the individual Ku sensors is not something we care about, at least not in this plot.
L388: This is a bit confusing. Does this mean that the values of Delta_G, as explained in L376 and 377, are used. Or did you do another analysis. Please clarify in the manuscript.
L391: Section 4.2 would maybe benefit from adding two or three subsections when discussing the results since there are different analyses carried out and discussed (gauges vs Ku, Ku SR - Std vs KU SR - Dual).
L395: without correction means that L does not use the +0.360 km (to account for melting layer) and the 0.2 dB for wet antenna? Section 2.4 does not specify what „with correction“ and „without correction“ precisely means, in particular for the melting layer height.
L398: Since you mention that it is important to account for both error sources, wouldn’t it be good to show both corrections (melting layer and wet antenna) separately in an updated Figure 9?
L402: It is not clear from the figure that the SR estimates are better than the ones from S sensors. In the plot we do not see which rain gauges corresponds to which Ku sensor. Most rain gauges are placed very close to a Ku terminal. Maybe the plot could be optimised to show e.g. each Ku sensor in a separate row of subplots each only with the rain gauges in the vicinity of its location or its path.
L406: What does HDR mean here? Probably HD Rain. But this abbreviation was not introduced.
L408: One reason why the correlation between rain gauges might drop faster with increasing distance compared to the Ku sensors is that the Ku sensors provide a path-averaged rain rate estimate which smoothens spatial extremes compared to the rain gauge measurements. This should be mentioned here in the text, because now the text sounds as if the gauges are inferior devices for rainfall measurement with the statement in the sentence before about the consistency of the HDR devices.
L415-L417: I do not unterstand the argumentation in these two sentences. Please rewrite.
Figure 11: Is this done with data from all gauges and all Ku sensors (separated by the applied method) or done with one pair of gauge and Ku sensor?
L425: What is a „directing coefficient“? Please clarify in the manuscript what is calculated here.
L428: I guess you mean „S-Std“ here and not „SR-Std“ based on what is described here. If not, I understood things wrongly. But maybe the text could be clearer.
L455: Why are the satellite signals received in Ivory Coast much weaker? Please explain in the text.
L484 (and following sentences): „…as the rain may be too light to be detected by the rain gauge“. Since the satellite link rainfall estimation also has a lower detection limit I would not agree with this argumentation. If you want to use this argument, please provide info on the lower threshold of the rain gauge data and of your rainfall estimates. A more likely cause for these false-positive rainy days could be that the rain event detection method, briefly described in section 3.2 but not explicitly validated, might produce false-positive rain events. This is a common challenge when processing attenuation data from terrestrial microwave links for which the raw data time series look very similar to the ones from satellite microwave links. Please elaborate on this and/or updated the text.
Figure 13: These plots should be larger.
L552: „…given their difference of nature“ is not very precise. You probably mean the different spatial integration characteristics and different operating principle in general. But it should be more precise in the text. Also, why exactly do we expect that the lowest quantiles are overestimated by the Ku-sensors?
L556: „…but also that the intensity-dependence of this underestimation seems to have been solved“. Please be more precise in the text. I do not understand what is meant here.
L559: You could cite Polz et al. (2023), which you already cited in the introduction, again here because they have analysed this effect in detail for terrestrial microwave links.
L581: This is the first time I read about „quasi vertically pointing“ in the manuscript. This should be either explained here, or maybe better, in the section describing the setup in Abidjan.
Figure A1: If I understand the caption correctly, I would name the data shown in the plot on the top „FP“ for false-positive (gauge has no rain, Ku detectors rain) and the data shown in the plot at the bottom „FN“ false-negative (gauge has rain, Ku detects no rain).
## Technical corrections (only partly documented, mainly done for section 4 and 5, due to limited time spent on this task):L3: „link path“ instead of „path link“
Equation 2: Should appear at end of sentence and not a top of the page.
L192: I have not seen the word „lineic“ been used a lot in this context. You might consider writing „path attenuation“. In the case of equation 13 here it is the „specific path attenuation“.
L413: The figure caption says that 30-minute resolution data is used. Here you write 1h. Please correct.
L429: Delete the „of in „of the atmospheric…“
L443: Better write „analysis“ instead of „study“ here because you only refer to the results of this section and not the results of the whole manuscript.
L448: Write „satellite with low signal strength“.
L454: There is a „was“ or „is“ missing in this sentence.
L512: Unclear what „by both of the rain gauges…“ means here. Please rephrase.
L527: I do not see „red crosses“ in Figure 14. I guess it should be „black crosses“ here.
L528: Same here. No „cyan dots“, maybe should be „blue dots“.
L545: Write „…more homogeneous rainfall distribution along the path affected by rain“ or something similar.
L547: From the text it seems Fig 15 and Fig 16 are not the ones that should be refered to here.
Fig 15: y-axis should maybe not be called „Station quantiles“ but something like „Ku-sensor quantiles“. On the x-axis write „gauge“ instead of „gage“.
Fig 16: „HDR“ on the y-axis is not used in the text except for two individual occasions. Maybe use something else here.
L573: write „rain rate“ instead of just „rain“.
L600: you maybe meant „because of the power-law“ instead of „but power-law…“
Citation: https://doi.org/10.5194/amt-2024-88-RC2 - AC2: 'Reply on RC2', Louise Gelbart, 12 Sep 2024
Status: closed
-
RC1: 'Comment on amt-2024-88', Marielle Gosset, 15 Jul 2024
In this study the authors propose and assess a new method to improve rainfall measurement from earth-satellite microwave links (from TV satellites).
The principle of rainfall estimation from earth-satellite MW links in the KU domain was already explained and assessed beore by the authors.
Here they concentrate on a particular and technical aspect of the problem : how to improve rainfall estimation by a better separation between i) the rain induced attenuation of the received MW signal - which is the base of measurement, through a specifc attenuation vs rain rate relationship and ii) the posible increase of RS because of the natuarl MW emisionof the atmosphere and the droplets themselves.
The proposed method is based on a dual-channel approach where one of the channels is senstive only to the atmospheric noise/radiation.
Altogether the paper is well structured and written. The proposed methodology and the physical principles are well explained.
The experimental results show that the proposed dual channel method does substantually improve the estimation.
The authors also aknowledge that there are still unresolved issues and sources of uncertainties/erros which are not accounted for.
In my view the paper could be published once a few details in the processing and assumptions have been clarified.I also propose some suggestions for the wording.
TECHNICAL ISSUES OR QUESTIONS REQUESTING FURTHER EXPLANATIONS
PN is time independent in Eq 1 – doesn’t it vary in time - with temperature for instance ?
L88-89 : to be clarified : do you mean that the values of TA on the one hand and Tatm on the other hand do not vary with meteorological conditions ? which central frequencies are assumed to be close to each other ? not clear…..
L96-103 :
Eq 2 and 3 don’t really show the double impact of rain – the dependence of Ta on rain has not been detailed yet …..
Eq 11 and 12 :
I do not understand when and why the dependence of tr (and tatm) on frequency (as shown in Eq 4, 5 and 6) was dropped
The simplifications made to go from Eq 2 – which as a strong dependence on f through tatm) to Eq 11 which has lost the dependence of tr to frequency have to be explained and justified.
In my understanding Eq 11 should have a term in trA and one in trB
L175-179 – The simplification in Eq 11 implies that trchannelA = trchannelB … These paragraph implies that this in not true
I believe some steps are needed for the reader to follow all simplifications leading to simplified Eq 12 from Eq 2 to 6
L225-229 : did the authors have a chance to make some comparisons between this ARPEGE iso0° and the one from the MeteoFrance high resolution model AROME ? or with the MeteoFrance radar data (where the iso0° can be extracted using some polarimetric variable)- Same questions with Radiosoundings (or statistics from a satellite based radar such as GPM DPR ) to verify the reliability of those ARPEGE levels ?
L229 : it would be nice to have more information about these experiments…. Context/data/results etc…
L405 and Fig10 : Please be more precise : what do you lean but correlation between devices ? which data is compared ? time series of rain intensity ? which time step ? any filtering of outliers ?
It would be interesting to see the scatter plots and more statistics on the timeseries (KGE ?) in addition to the QQplots which provide only partial information.
It is not clear from the paper which relationship is used for attenuation-rainfall estimation – are the ITU parameters mentioned in L191-193 applied to all experiments (dual and single ) ? Was there any adjustment/calibration ? if yes how ?
STYLE/WORDING suggestions
L3 : link path and not path link
L7 : which can be commercial (what is meant ? off the shelve ) ?
THE measured attenuationL19 : what is small-scale or medium-scale rainfall intensity ? are you talking about resolution or intensities values ? Needs to be clarified – If you mean intensities why single out small and medium , heavy rainfall is the most damaging for ‘human/property damage’
L21 : Earth Observation satellite (rather than remote sensing ) ?
L35 ‘previous studies ‘ – which ones ?
L40 : combination of the signal from the satellite and a background noise that depends on the state… ?
L43 : and the baseline signal measured during dry period
In the presence of high rainfall rates, however,…
L47 : ‘The present study …. dual channel’. long clumsy sentence – Just get to the point , which is improving rainfall estimation from dual-channel measurement of TV satellite signal by accounting for background noise. No need to repeat low cost etc….
L50 : the assumptions it relies on ? (rather than the hypotheses it presupposes)
L52 inherent to rainfall estimation using…
L54 : section 2 introduces the physical principles (rather than ‘context’)
L56 : a physical device whose characteristics …..
L60 : physical principles
L61 : ground satellite : you should choose and keep one expression – Earth-Satellite as used before or ground-satellite …..LinkS or a link.
Careful with missing articles along the paper….
L89 : the central frequency of what and compared to what ? not clear – what are each other ?–
check format of the citation.
L104 : Various processes …. Links – do the process characterize the transparency ? do they influence the propagation of the link (or of the microwave signal along the link…) ? To be rephrased.
L264 : data processing rather than data treatment
L280 : not to go below the threshold (and not exceed …. Which means the opposite)
L314 : Eq 17 and 18 are redundant – you can give directly Eq 18 in dB ….
Citation: https://doi.org/10.5194/amt-2024-88-RC1 - AC1: 'Reply on RC1', Louise Gelbart, 12 Sep 2024
-
RC2: 'Comment on amt-2024-88', Anonymous Referee #2, 09 Aug 2024
Summary:
The authors present a novel analysis of data from satellite microwave links for rainfall retrieval. They show (for the first time, as far as I am informed) the impact and the potential usage of measurements of radiation from rain drops in the atmosphere on the received signal level of satellite microwave links. They give a good theoretical explanation of the effect they are investigating and show convincing improvements with their proposed new method based on a fairly large dataset. They discuss the remaining differences compared to reference data, which are still considerable for their analysis in Ivory Coast, in a detailed and open manner, clearly indicating future pathways for research. The results of their proposed method are encouraging and relevant in general, because most (if not all) other studies with satellite microwave links seem to not have taken the investigated effect explicitly into account. The paper is mostly well written, the figures are mostly clear and it fits very well for AMT. I suggest a minor revision, event though I have quite a few specific comments. But, none of my comments addresses and issue that I consider to be major.
## Specific commentsL23: „…low revisit time compared to rainfall dynamics.“ It is somehow clear what is meant here, but it should be formulated more precisely. Please rephrase.
L60: General comment on Section 2. Each subsection is understandable, but I am missing condensed info on how the improved dual band method is actually applied. It is not directly clear from how the equations are linked in the text, how Delta_G impacts the rain rate estimation via the power law. Maybe there should be an additional subsection that links things together, from t_atm via A to the power law, but explaining how it is done with the Std and the Dual method. Maybe this also fits as an extension of section 2.4.
Fig 1: What is the path length (affected by rain-induced attenuation) that is used for the calculation of the attenuation on the y-axis here?
Fig 1: Would it be possible to also show the increase in received signal strength for the increased brightness temperature for a given bandwidth, e.g. 1 GHz as used in your LNB?
L200: There is a lot of information provided in the paragraph that starts here, but it is not clear if one of the enhanced models is used, and if not, why?
L245: Since the lower and upper frequency band are directly adjacent, is there power leakage from one band to the other, i.e. for the described case where one TV-satellite only transmits in one of the two bands, how much does still leak into the other band of the receiver where it somehow contaminates the radiometer-like measurement?
L272: Is this method with the LSTM documented somewhere in more detail? What is the temporal granularity at which the classification is done?
Figure 3: Is there the potential of leakage of the Astra 19 signals into the receiver of the RS sensors? Or more specifically, what is the half-power beam width, or in general the gain pattern of the antennas? And how high is the noise floor of the radiometer channel of the RS sensors in relation to the potential leakage of Astra 19 signal into the RS receiver via the RS system’s antenna side lobes?
L296: What does „almost no signal“ mean here. How is it different from the setup in France and how does/could it affect the rain rate retrievals?
Figure 3 and Figure 4: What are the assumed melting layer heights for the plotted path lengths? That would be interesting to know. In Abidjan the elevation angle of the antennas is probably much higher because of being close to the equators, hence, I expect a much shorter path that is relevant for a typical melting layer height.
L310 and following: It is clear from the explanations here and from the shown plots that channel A and B have different P_atm which can be attributed to G_A and G_B. Did you also check that there isn’t an offset or some other inaccuracy due to the low-cost electronics of the LNB, which is not optimised to give accurate readings of received signal level?
Figure 5: Just a tiny detail, but you could use aligned y-axes here (they are slightly misaligned) and then remove the y-axis tick labels of the plot on the right.
Figure 5: What is the unit on the y-axis. If it is not dBm, what is the reference level for the dB given here?
L321: How do you want to assure that the rain-induced attenuation is strong enough to have t_R approx. 0? Please explain in the text how you identify these events.
L339: Where does the difference of P_A_Tot and P_B_Tot during normal operation (not pointing away form the satellite) come from? Is this due to different transmit power of the satellite in the two bands or can this also be an effect of different gains of the electronics for band A and B?
L350 and following: I unterstand the argumentation here on why Delta_G_p3 is used. But doesn’t this, the difference of Delta_G depending on what the current brightness temperature is, mean that Delta_G varies with rain rate? If yes, does this affect your results?
L374: Why does the existence of a dry season „explain the need for sufficient data to calculate Delta_G“? Do you mean that it is harder to get enough data with heavy rain due to the dry season? Please rephrase.
Figure 7: What happened during the period in September 2022 where signal levels for A and B both are increased for several days?
Figure 8: I would put the box plot with Delta_G_ref in the middle. But if you redo this plot, you might consider doing it with something else than boxplots since, here, the spread and distribution of the rainfall sums of the individual Ku sensors is not something we care about, at least not in this plot.
L388: This is a bit confusing. Does this mean that the values of Delta_G, as explained in L376 and 377, are used. Or did you do another analysis. Please clarify in the manuscript.
L391: Section 4.2 would maybe benefit from adding two or three subsections when discussing the results since there are different analyses carried out and discussed (gauges vs Ku, Ku SR - Std vs KU SR - Dual).
L395: without correction means that L does not use the +0.360 km (to account for melting layer) and the 0.2 dB for wet antenna? Section 2.4 does not specify what „with correction“ and „without correction“ precisely means, in particular for the melting layer height.
L398: Since you mention that it is important to account for both error sources, wouldn’t it be good to show both corrections (melting layer and wet antenna) separately in an updated Figure 9?
L402: It is not clear from the figure that the SR estimates are better than the ones from S sensors. In the plot we do not see which rain gauges corresponds to which Ku sensor. Most rain gauges are placed very close to a Ku terminal. Maybe the plot could be optimised to show e.g. each Ku sensor in a separate row of subplots each only with the rain gauges in the vicinity of its location or its path.
L406: What does HDR mean here? Probably HD Rain. But this abbreviation was not introduced.
L408: One reason why the correlation between rain gauges might drop faster with increasing distance compared to the Ku sensors is that the Ku sensors provide a path-averaged rain rate estimate which smoothens spatial extremes compared to the rain gauge measurements. This should be mentioned here in the text, because now the text sounds as if the gauges are inferior devices for rainfall measurement with the statement in the sentence before about the consistency of the HDR devices.
L415-L417: I do not unterstand the argumentation in these two sentences. Please rewrite.
Figure 11: Is this done with data from all gauges and all Ku sensors (separated by the applied method) or done with one pair of gauge and Ku sensor?
L425: What is a „directing coefficient“? Please clarify in the manuscript what is calculated here.
L428: I guess you mean „S-Std“ here and not „SR-Std“ based on what is described here. If not, I understood things wrongly. But maybe the text could be clearer.
L455: Why are the satellite signals received in Ivory Coast much weaker? Please explain in the text.
L484 (and following sentences): „…as the rain may be too light to be detected by the rain gauge“. Since the satellite link rainfall estimation also has a lower detection limit I would not agree with this argumentation. If you want to use this argument, please provide info on the lower threshold of the rain gauge data and of your rainfall estimates. A more likely cause for these false-positive rainy days could be that the rain event detection method, briefly described in section 3.2 but not explicitly validated, might produce false-positive rain events. This is a common challenge when processing attenuation data from terrestrial microwave links for which the raw data time series look very similar to the ones from satellite microwave links. Please elaborate on this and/or updated the text.
Figure 13: These plots should be larger.
L552: „…given their difference of nature“ is not very precise. You probably mean the different spatial integration characteristics and different operating principle in general. But it should be more precise in the text. Also, why exactly do we expect that the lowest quantiles are overestimated by the Ku-sensors?
L556: „…but also that the intensity-dependence of this underestimation seems to have been solved“. Please be more precise in the text. I do not understand what is meant here.
L559: You could cite Polz et al. (2023), which you already cited in the introduction, again here because they have analysed this effect in detail for terrestrial microwave links.
L581: This is the first time I read about „quasi vertically pointing“ in the manuscript. This should be either explained here, or maybe better, in the section describing the setup in Abidjan.
Figure A1: If I understand the caption correctly, I would name the data shown in the plot on the top „FP“ for false-positive (gauge has no rain, Ku detectors rain) and the data shown in the plot at the bottom „FN“ false-negative (gauge has rain, Ku detects no rain).
## Technical corrections (only partly documented, mainly done for section 4 and 5, due to limited time spent on this task):L3: „link path“ instead of „path link“
Equation 2: Should appear at end of sentence and not a top of the page.
L192: I have not seen the word „lineic“ been used a lot in this context. You might consider writing „path attenuation“. In the case of equation 13 here it is the „specific path attenuation“.
L413: The figure caption says that 30-minute resolution data is used. Here you write 1h. Please correct.
L429: Delete the „of in „of the atmospheric…“
L443: Better write „analysis“ instead of „study“ here because you only refer to the results of this section and not the results of the whole manuscript.
L448: Write „satellite with low signal strength“.
L454: There is a „was“ or „is“ missing in this sentence.
L512: Unclear what „by both of the rain gauges…“ means here. Please rephrase.
L527: I do not see „red crosses“ in Figure 14. I guess it should be „black crosses“ here.
L528: Same here. No „cyan dots“, maybe should be „blue dots“.
L545: Write „…more homogeneous rainfall distribution along the path affected by rain“ or something similar.
L547: From the text it seems Fig 15 and Fig 16 are not the ones that should be refered to here.
Fig 15: y-axis should maybe not be called „Station quantiles“ but something like „Ku-sensor quantiles“. On the x-axis write „gauge“ instead of „gage“.
Fig 16: „HDR“ on the y-axis is not used in the text except for two individual occasions. Maybe use something else here.
L573: write „rain rate“ instead of just „rain“.
L600: you maybe meant „because of the power-law“ instead of „but power-law…“
Citation: https://doi.org/10.5194/amt-2024-88-RC2 - AC2: 'Reply on RC2', Louise Gelbart, 12 Sep 2024
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
232 | 43 | 23 | 298 | 7 | 9 |
- HTML: 232
- PDF: 43
- XML: 23
- Total: 298
- BibTeX: 7
- EndNote: 9
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1